Professional Documents
Culture Documents
INFLATION-LINKED RESEARCH
March 2010
GLOBAL INFLATION-LINKED
PRODUCTS
A USER’S GUIDE
PLEASE SEE ANALYST CERTIFICATIONS AND IMPORTANT DISCLOSURES ON THE LAST PAGE.
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Introduction
Alan James Global inflation-linked markets have been through an extremely difficult period in the two
+44 (0) 20 7773 2238 years since the last edition of this guide was published in February 2008, but in most
alan.james@barcap.com countries they have emerged from this turmoil stronger than ever. The economic shock
accompanying the financial market meltdown from September 2008 prompted a rapid fall
in inflation globally, with many countries experiencing deflation for the first time in 70 years,
but this was mild compared to the expected fall implied by the cheapening of inflation-
linked bonds. The extreme deleveraging phase that engulfed almost all financial markets
included the majority of off-benchmark investors in inflation-linked bonds being stopped
out of their positions. The impact was greater due to the acceleration in inflation in the
preceding year, which had led to more investors holding inflation-linked bonds versus
nominals as protection against inflation. After the extremes of Q4 08, the first half of 2009
was all about absorbing the float sold by these breakeven investors. During this period most
governments were unable to increase inflation-linked issuance despite their much higher
funding needs, but by the second half of 2009, as the markets normalised, supply began to
increase and 2010 is set to see significantly more supply in government inflation-linked
bonds than ever before.
Inflation swap markets became dislocated from inflation-linked bond valuations during late
2008, generally cheapening less than bonds but with a commensurate drop in liquidity.
Despite this illiquidity swap valuations were almost certainly less distorted during this phase
than deleveraging bond markets and were also probably more representative of inflation
expectations in the first half of 2009 as central banks began to engage in various forms of
quantitative easing. As asset swap activity in government inflation-linked bonds resumed,
this enabled relative valuations between inflation-linked bonds and swaps to move back
towards their historical normal ranges. Inflation volatility markets suffered notably more
than inflation swaps and liquidity still has not recovered in early 2010, despite markets once
again being relatively well defined. While inflation uncertainty is notably higher in the
aftermath of extreme fiscal injections and central bank quantitative easing programmes,
implied volatility remains an order of magnitude above both realised market and actual
inflation volatility.
As with previous versions of this guide, this edition aims to provide a solid base for those
looking to understand inflation-linked bonds and derivatives while also providing an
invaluable reference source for market participants. The first section of the guide focuses on
15 March 2010 1
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
the products within the inflation-linked universe, the second covers the features of each
individual market. This is followed by the section on inflation-linked indices and tradable
inflation-linked index products, before addressing important themes for the asset class,
including portfolio concepts, valuation themes and both the macro economic and
component drivers of inflation.
Government inflation-linked bonds tend to have a similar structure, where principal and
income are adjusted for changes in the relevant consumer price index between issue and
cash flow payment, subject to an indexation lag. In this guide, such bonds are referred to as
“inflation-linked”, “inflation-indexed”, “index-linked”, “real return bonds”, “inflation bonds”,
“inflation-protected securities”, or just “linkers”, with these terms used interchangeably.
Other bonds with alternative links to inflation are referred to as structured notes or bonds,
but the type of structure can vary markedly. The variety in inflation derivatives is even wider.
While calculations vary between countries and products, this publication tries to outline the
concepts behind them and their differences as well as providing the required rigorous detail.
With global monetary and fiscal policy in recent years having been aggressively set out to
avoid deflation, there are clear risks that this results in higher than intended inflation later in
this economic cycle. Indeed as highlighted in the concluding article, ‘Run inflation run’, the
economic benefits of a period of temporarily higher inflation may be considerable, which
may mean that policymakers are likely to tolerate such a period. It may not be imminent
however, with core inflation likely to stay subdued for some time as argued in ‘Why output
gaps still matter for inflation’. Given this backdrop of heightened uncertainty, inflation-
linked products are likely to continue to develop as a vital asset class for investors and
policymakers alike.
15 March 2010 2
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
TABLE OF CONTENTS
INFLATION PRODUCTS 5
Inflation-linked bonds ............................................................................................................................... 6
Inflation derivatives..................................................................................................................................20
INFLATION MARKETS 39
US.................................................................................................................................................................40
Euro area ....................................................................................................................................................52
UK.................................................................................................................................................................73
Japan............................................................................................................................................................88
Canada........................................................................................................................................................97
Sweden .................................................................................................................................................... 101
Australia .................................................................................................................................................. 110
Brazil ......................................................................................................................................................... 118
Mexico ..................................................................................................................................................... 125
Argentina................................................................................................................................................. 128
Chile.......................................................................................................................................................... 133
Colombia ................................................................................................................................................. 137
Uruguay ................................................................................................................................................... 139
Israel ......................................................................................................................................................... 140
Turkey ...................................................................................................................................................... 143
South Africa............................................................................................................................................ 146
Poland ...................................................................................................................................................... 151
Iceland...................................................................................................................................................... 153
South Korea ............................................................................................................................................ 155
15 March 2010 3
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
APPENDICES 299
Key information sources...................................................................................................................... 300
Summary sovereign table.................................................................................................................... 302
Real Yield Histories................................................................................................................................ 304
Breakeven Inflation Histories .............................................................................................................. 307
15 March 2010 4
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
INFLATION PRODUCTS
15 March 2010 5
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Inflation-linked bonds
Alan James Inflation-linked bonds are now an established segment of their respective local bond
+44 (0) 20 7773 2238 markets for most issuers, although with varying degrees of activity. Despite the
alan.james@barcap.com financial dislocations in the second half of 2008 significantly affecting most
government inflation-linked markets, all G7 countries still run inflation-linked debt
Khrishnamoorthy Sooben programmes alongside their nominal ones. Thirteen of the 20 largest countries in the
+44 (0) 20 777 37514 world weighted by GDP have inflation-linked markets. The Canadian model, introduced
khrishnamoorthy.sooben by Canada in 1990, has become the standard quotation format for linkers, with even the
@barcap.com UK switching to this framework for all new bonds. Latin American markets have not
adopted the Canadian format, but have similar, although non-standardised, features.
Corporate issuance of linkers where the inflation exposure is not swapped out has been
rare outside UK among the developed markets, though there are countries such as Chile
and Israel where there is more corporate debt that is inflation-linked than nominal.
The distinguishing feature between a nominal and an inflation-linked bond is the yield
metric used. Linkers offer a “real” yield which is net of inflation; this yield is therefore “real”
in the sense of true purchasing power, and just like a nominal bond is quoted in nominal
yield, a linker is quoted in real yield. The Canadian model has become the standard one for
inflation-linked bonds, with its innovative simplicity being that all calculations are done in
“real” terms, without any assumption to be made about the future path of the price index as
is necessary for instance in the traditional UK model. It is then straightforward to relate the
real yield to the linker’s price using the generic present value formulation which relates the
price of a nominal bond to its nominal yield:
P DU = P CU + AC
tm
C 100
=∑ +
t =t 0 (1 + RY ) t
(1 + RY ) tm
Where: PDU is the dirty price, unadjusted for past inflation accrual
PCU is the clean quoted price, unadjusted for past inflation accrual
15 March 2010 6
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
This price PDU does not include inflation which has already accrued on the bond’s principal
value; a final adjustment therefore needs to be applied to arrive at the true or monetary
settlement price. The “Base Index” is the price index reference value (Reference Index) on
the date from which inflation accrues, which is also the date from which interest accrues,
but this does not necessarily coincide with the initial settlement date for the bond. For
example, in France, bonds are issued with short first coupons with accrual calculated from
what would have been the previous coupon date had the bond existed, so in this case the
Base Index is calculated from the same day. The inflation accrual adjustment reflects the
cumulated change from the Base Index to the Reference Index value on the trade settlement
date and is computed as an Index Ratio:
In the Canadian model, the daily Reference Index is based on a lag from the current
settlement date. Typically, the reference for the first day of a month is equal to the price
index value three months back. For instance, in the euro area or US, the Reference Index for
1 February 2010 is the CPI value of November 2009. Reference indices for intervening days
are calculated by a linear interpolation on a standard actual/actual day count basis:
(t − 1)
Index = CPIm − 3 + × (CPIm − 2 − CPIm − 3)
Dm
Both the Reference Index and the Base Index are truncated to six decimal places and then
rounded to five decimal places before computing the Index Ratio, which is also rounded to
five decimal places. The settlement price of a Canadian model bond is then determined as
the product of the price PDU, which is non-adjusted for accrued inflation, and the Index Ratio
for the trade settlement date.
Canadian-style linkers are quoted in clean price terms (PCU), ie, net of accrued interest. The
real accrued coupon is calculated in the same way as for a nominal bond. The quoted clean
price and the real accrued coupon are added, and the sum (price PDU above) is multiplied by
the Index ratio to arrive at the cash settlement amount. For coupons paid, the (real) coupon
rate is multiplied by the Index Ratio, and likewise for the redemption amount. In many
countries with the Canadian model, but not Canada itself, the principal value at redemption
is floored at par even if the index ratio is below 1, ie, the price index at maturity is lower than
the Base Index. This deflation floor only applies to the principal value, coupons can be paid
off a principal amount that is less than par, including the final coupon.
The inclusion of a deflation floor proved a popular addition to the Canadian model when the
US Treasury first issued inflation-linked bonds (TIPS) with this feature, and most other
issuers have followed suit. French supply the following year thus opted for the floor too, as
have all subsequent euro issuers, while Sweden introduced floors for bonds issued from
15 March 2010 7
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
1999. However, Japan chose not to include floors on its inflation-linked bonds, as the value
of a floor would have been much higher than in other markets where deflation has been
very rare: the interpretation of the value of bonds would have been much more difficult. The
absence of floors in JGBis is a significant reason why the market became extremely
distressed in 2008/2009 as deflation fears increased. While the shorter end of the US TIPS
market had a similar supply/demand dislocation in Q4 08, when real yields were pushed
well above nominals, recently issued TIPS only very briefly traded through the level at which
their floor ensured a superior hold to maturity return to the nominal treasury curve. In 2009,
Japan announced that it was examining the possibility of issuing floored JGBIs. The UK
decided not to include a floor when it started issuing its new format bonds and nor did
South Korea. In most Canadian-style markets, coupons are paid semi-annually, consistent
with nominal bonds, with the exception of France, Sweden, Germany and Greece, where
annual payments are the norm for both linker and nominal issues.
Some countries have slightly different inflation lags, for example, in Japan, the three-month
lag is to the tenth of the month rather than the first, and in South Africa the lag is four
months. Japan also rounds to three decimal places rather than five when calculating the
Index Ratio, resulting in a less smooth accretion of inflation across a month. In Sweden, day
count conventions for inflation accrual are based on a linear rate that assumes 30 days in
each month, which means discontinuous accretion at month-end for months that are not
this length. The Swedish market also trades almost entirely on a real yield basis, with quoted
prices including inflation uplift, as Canadian conventions were only adopted after the
market began.
In more precise terms, carry on a linker is defined as the required change in the real yield
over the holding period considered such that the return from holding the bond is equal to
the repo cost. In common terminology, it is defined as the difference between the forward
real yield and the current or spot real yield. The return over the holding period has three
components: any coupon payments and/or coupon accrual, the inflation accrual and the
change in the quoted clean price. Given that the return depends on the inflation accrual, an
exact forward real yield or carry can therefore be calculated only up to the furthest
settlement date for which the Reference index value is known. For example, on 10 January
2010, the latest euro HICPx value known was for November 2009. This means that for a
euro HICPx inflation-linked bond, an exact carry was then calculable only up to the 1
February 2010 settlement date. On 15 January 2010, with the release of the December 2009
inflation data, the furthest settlement date for an exact carry computation then shifted to 1
15 March 2010 8
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
March 2010. Depending on the gap between the month of the last known CPI value and the
current date, an exact carry for a Canadian-style linker with a three-month lag can therefore
generally be calculated over up to a two-week to a month and a half forward horizon,
assuming an inflation index print date around mid-month.
The formal computation steps for a carry calculation are laid out in Exhibit 1, but the basis
point real yield carry in a given month can be approximated by:
[M/M inflation accretion + (real yield– repo rate)/12]/forward modified duration of bond
Shorter duration bonds thus have a notably higher sensitivity to inflation accretion than
longer issues. For example, consider a flat real yield curve of 2%, a known monthly inflation
accretion given the appropriate lag of 1% and repo funding at 5%. For a par bond with a
duration of 10, one-month carry is worth 7.5bp; [1% + (2%-5%)/12]/10, but for a 1y
duration bond the carry is worth 75bp; [1%+ (2%-5%)/12]/1.
The inflation lag in the Canadian model means that the valuation of the spot real yield may
be “optically” distorted by future inflation accrual which is already known but which has not
yet accrued on the bond. For example, at the start of a month during which the inflation
accrual is known to be very high, the real yield will tend to be biased down and therefore
appear rich, but a fair assessment of its valuation should take into account the positive carry
over the month. Looking at forward real yields offers a more realistic gauge of value ahead
of extreme carry periods. It is also possible to construct estimated forward real yields
beyond the known inflation data using forecasts, although it is only possible to lock in such
forward levels if the future inflation can be traded, for example, using inflation futures or
swaps. As discussed later in this guide, expected future carry can be an important feature of
pricing both real yields and breakeven spreads versus nominal bonds, in particular with
seasonal swings likely ahead of periods of expected very positive and negative carry.
PtDU is the dirty price, unadjusted for past inflation accrual, at settlement date t
PtCU is the clean quoted price, unadjusted for past inflation accrual, at settlement date t
IRt is the Index ratio at settlement date t, and can be a known or expected ratio
Repo is the repo rate used to finance the position and is assumed to be constant
= PTDU
0
* IRT0
15 March 2010 9
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Using the non-arbitrage principle for forward calculations, the total proceeds, ie, resale
value plus any reinvested coupon, from the bond at date T2 should be equal to the cash
settlement price in T0 uplifted with the repo cost. Given the coupon to be paid in T1, the
forward bond price in T1 has to be computed first; it is assumed that the coupon received
will be reinvested in the bond at the calculated forward price for T1. The unadjusted forward
price PT1DU at settlement date T1 should satisfy the same non-arbitrage principle and is
deduced from the equation:
⎧ ⎛T −T ⎞ ⎫
(C + P )* IR = (P
DU
T1 T1
DU
T0 )
* IRT0 * ⎨1 + ⎜ 1 0 ⎟ * repo⎬
⎩ ⎝ 360 ⎠ ⎭
The monetary coupon payment in T1 is equal to C * IRT1 . This is assumed to be reinvested
C * IRT1
in the bond. The additional inflation-unadjusted notional bought is therefore ,
PTDU
1
* IRT1
C
ie, .
PT1DU
C
Total inflation-unadjusted notional held in T2 is 1+ . The unadjusted forward price
PT1DU
PTDU
2
at settlement date T2 is then deduced from the equation:
⎛ ⎞
⎜1 + C ⎟ * PTDU * IRT = PTDU * IRT * ⎧⎨1 + ⎛⎜ T1 − T0 * repo ⎞⎟⎫⎬
( )
⎜ PTDU ⎟ 2 2 0 0
⎩ ⎝ 360 ⎠⎭
⎝ 1 ⎠
In theory, the forward real yield can then be deduced from PTDU
2
using the generic present
value formulation above.
In practice, the forward real yield for T2 can then be obtained through the “Yield” function in
Excel or the “YA” (Yield Analysis) page on Bloomberg, with both using PTCU
2
as an input.
15 March 2010 10
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
short-term interest rate. Similarly, Uruguay and Argentina turned to inflation linking in the
aftermath of the 2002 financial crisis (in turn, due to a large extent to currency risk
associated with large foreign currency-denominated public and private liabilities) when
sharply raising inflation expectations left inflation indexation (legally banned in Argentina by
the Convertibility Law in place during 1991-2001) as the only viable alternative to the dollar.
By contrast with this tactical (and in some cases ephemeral) reliance on inflation linkers, the
development of linker markets in Chile and Colombia owed itself to a long-standing policy
of building a stable unit of account for long-term securities that dated back to the 1960s.
While most Latin American linkers are quoted in real yield, in a format similar to Canadian-
style bonds, the price indexation methodologies can be very different. In most markets, the
intermediate Index Reference values are not determined via linear interpolation, but assuming
a geometric increase between the known published values. The experience of very high
inflation in many Latin-American countries means that the lag used in the price indexation of
linkers tends to be shorter than in the developed markets. In the extreme case, IPCA consumer
price indexed-linked bonds in Brazil have a lag that means their actual value is unknown in the
early part of each month. During this period, the bonds are priced off a consensus expectation
for the next inflation number, adjusting to the actual value when it is released. In Mexico the
UDI inflation index is published twice a month, creating a very short de facto lag in the bonds,
but otherwise, the daily interpolating of the Reference Index means that the mechanics are
very similar to the Canadian model. The details of conventions and calculations in each Latin
American market are described in the relevant country sections.
In most developed economies this factor is notably less important than it has been in the
past. With independent and transparent monetary policies, the gap between market and
government expectations of inflation is likely to be small. While there may be times when
divergences of expectations encourage issuance, the expectations mismatch is unlikely to
be the primary concern. For more recently developing countries with less established
monetary and fiscal institutions and capital markets there may still be occasions when
governments perceive that the markets’ expectations of price increases are too high,
particularly when institutional changes have been made to fight inflation more directly. The
substantial increase in issuance in Brazil in 2006 may have been partly motivated by such
considerations, as may have the resumption of issuance in Turkey in 2007.
15 March 2010 11
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
remain high. The market may be more willing to believe in the institutional changes made to
bring down inflation, if the government is seen to be “putting its money where its mouth is”.
The more inflation-linked debt a country issues, the less incentive it has to reflate the
economy and reduce the real value of the debt stock. The longer the expected lifespan of a
particular government or policy regime the more the strategy may be beneficial. This is
another argument that is not particularly relevant for developed economies with totally
independent monetary policies. It may be very significant for transitional economies that
have undergone periods of high inflation though; again, Turkey is a clear example of where
this may apply, while it may also have been a factor behind the significant increase in
issuance in South Africa in 2009.
Cyclical benefits
Issuance of inflation-linked bonds can have significant cyclical, as well as long-term liability
benefits for a government. When growth is strong, there is little pressure on public finances,
but inflation is likely to be higher. Equally, when growth is weak, prices are unlikely to rise
15 March 2010 12
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
quickly. Servicing linker costs should thus tend to be a fiscal stabiliser compared to servicing
nominal debt. The fiscal impact of a deflationary downturn on a country with a significant
stock of inflation-linked bonds ought to be less severe than a country with only nominal
debt. The UK DMO puts particular emphasis on the fact that inflation and the government’s
budgetary situation are likely to be highly correlated. Other than a “stagflation” scenario, the
main risk to this hypothesis is late in the economic cycle, when after a strong growth period
inflationary pressures may continue to grow even when output is already falling away, but
the counter-argument is that tax revenue also tends to lag output growth. Conversely,
issuing inflation-linked at the start of an economic upswing may be optimum timing, as it is
likely that during such a phase inflation risk premia will be high until policy acts to contain
inflationary pressures. It is also a time when funding needs are high, and it is advantageous
to extend the average life of the debt portfolio.
Risk diversification
Even governments with no natural preference for either real or nominal liabilities should
regard it as appropriate to have some inflation-linked liabilities within their debt, unless they
assign no probability to future inflation being lower than the market expects. A government
is better off having a balanced liability portfolio in the face of economic uncertainty. This
diversification benefit can mean that it is in a government’s interest to issue inflation-linked
bonds, even when implicit inflation is lower than the government’s inflation expectations.
As it is usually easier to sell longer-dated real return bonds than nominal issues, a benefit
also arises from reducing the exposure to short-term cash flow pressures.
In several Latin American countries, long-dated maturities have been issued in inflation-
linked before the nominal market was sufficiently developed for long nominal issuance, due
to fear of inflation eroding the value of nominal debt. Even when, such as in Mexico,
nominal curves have eventually extended to maturities as long as those of inflation-linked
15 March 2010 13
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
bonds, the duration of the long linkers remains notably longer due to the back-ended nature
of linker cash flows. South Africa has undergone a similar extension of its debt since the
introduction of inflation-linked bonds.
Social benefits
The existence of inflation-linked bonds may provide benefits to society beyond the funding
considerations. The ability to easily discern markets’ inflation expectations may be of benefit
to policy setters. In particular, there may be considerable benefits if breakeven spreads
between inflation-linked and nominal bonds help to avoid inflationary monetary and fiscal
policy errors. With central banks making no secret that they observe both spot and forward
inflation-linked breakevens, relatively stable spreads may also provide a self-reinforcing
credibility tool for inflation targeting. After the US FOMC indicated that the main market-
based series of inflationary expectations that it focuses on is the breakeven implied by 5y5y
forward TIPS, this series stayed in a tight 40bp range until the strong inflation volatility
experienced in 2008/2009. While it can be difficult for central bankers to ascertain how
much of breakevens is true inflationary expectations and how much comes from risk
premium, to the extent that the series itself becomes tied to policy credibility, this
differentiation becomes relatively less important.
Having a market-based reference of inflation expectations from linkers may also be useful
for economic agents in making decisions. The existence of inflation bonds could
theoretically reduce inflation uncertainty. This could encourage more savings, either directly
into inflation-linked bonds, or indirectly into assets for which there is a clearer real value if
there are inflation-linked assets for comparison. Putting a price on such benefits is difficult,
but as there seem to be few clear differences in behaviour between economic agents in
similar countries with and without inflation bond markets, it is unlikely to be very large.
15 March 2010 14
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
despite substantial domestic pension fund demand. In this case, increasingly active quasi-
and non-government supply subsequently developed along with inflation swaps trading,
with government issuance restarting in 2009 as funding needs rose.
One persistent criticism of governments issuing inflation-linked bonds is that any form of
inflation indexation is insidious and pernicious. If bonds are linked to inflation, there will be
increased pressure for other items to be linked to inflation too. The danger is that if the cost
of inflation is made less painful for individuals by widespread indexation, inflation may
increase until it reaches levels that are once again painful. This line of reasoning was
particularly prevalent in Germany, where, after multiple periods of hyper-inflation, it was
made illegal during the period of the Deutsche Mark for any debt to be indexed. While there
is some evidence to support the risks of creeping inflation from widespread indexation, and
countries such as Israel and Iceland have tried to wean themselves off indexation as a
consequence, this is a long way from saying that it is the fault of inflation-linked bonds.
There is no reason why bonds cannot be linked to inflation without general indexation
elsewhere. It should be relatively easy for a government to keep financial funding and other
price setting at arms’ length. If monetary policy is independent and can respond to the
inflationary effects of any indexation, the signalling benefit from inflation-linked bonds is
likely to partially offset the inflationary bias.
15 March 2010 15
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
temporary monthly inflation trends cause large deviations. In addition, the indexation lag
makes the real yield that is offered far from prefect at short maturities.
For those working on behalf of individual investors, such as pension fund managers, it also
makes sense for the inflation-linked bonds to be the risk-free asset for long-term investment.
In practice, they are only encouraged to do this actively if liabilities are tightly tied, such as in
UK-defined benefit pension schemes, as returns are often too low to justify the fees that most
investment managers take. For those investing for atypical individuals, whose consumption
basket has little linkage to consumer price indices, or whose investments occur in currencies
other than those in which consumption occurs, there is little incentive to consider
government inflation-linked bonds as the risk-free asset class. This does not mean that they
should not invest in them as part of a low-risk diversification strategy, but there is much less
argument for holding them as a core asset. Along similar lines, while there are clear
arguments for endowments to use inflation-linked government bonds as the basis for
investment decisions, for example, there is less of a case for corporates without clear
inflation-linked costs to invest spare cash anywhere but in the nominal market.
15 March 2010 16
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Accounting considerations
One issue that has been a limiting factor for non-government inflation-linked issuance,
particularly inflation paying, has been accounting treatment. In the US, FAS133 considers
the interest rate and the rate of inflation in the economic environment for the currency in
which a debt instrument is denominated to be clearly and closely related, allowing for
issuance of non-leveraged bonds. However, US Generally Accepted Accounting Principles
(GAAP) only allow for hedging of interest rate, foreign exchange and credit, making it
extremely difficult for a US corporate to take exposure in inflation swaps without marking to
market even if revenues are explicitly inflation-linked. Historically, the UK had the most
straightforward treatment, encouraging the market to develop. Bonds could be accounted
for on an amortised cost basis with uplift accounted for as a finance charge, with the same
reasoning as in the US, but inflation swaps could also normally be held on an accruals basis.
In July 2008, the International Accounting Standards Board (IASB) amended the IAS 39 rule
with clarification regarding the hedging of inflation. The hedge accounting of inflation is
allowed only if changes in inflation represent a contractually-specified element of cash
flows of the instrument considered. Other than Argentina, every county with an inflation-
linked market now either accepts international accounting standards, as in Europe, or has
an accounting agency seeking convergence with them.
Fundamental considerations
Fundamentally, the reasoning behind most corporate structural paying/issuing of inflation-
linked instruments is not that different from a government’s, though there can also be
shorter-term value and cash flow considerations. Corporates with significant revenues or
assets linked to inflation have clear potential to benefit from issuing or paying inflation, but
there is an argument for large firms with balanced inflation exposure to consider the
inflation market as a diversified source of funding. If there is a premium for the inflation-
linked asset class, then firms are likely to benefit from funding in inflation unless they have a
cost base or liabilities more closely tied to inflation than their revenue or assets. Such firms
include those with large wage bills that track price inflation relatively closely, but no linkage
between revenue and rising average prices.
The most common reason for corporates to issue or pay inflation is if their revenues are
inflation-linked. Sometimes this cash flow linkage is obvious, for instance, regulated utilities in
some countries have the rate at which they can increase prices linked to inflation. The main
reason for the lack of a corporate inflation-linked bond market in the US is that such inflation-
linked controls are relatively rare. Explicit, contractually-based inflation-linked revenues are the
most natural source of supply. In most countries other than the US, hedge accounting of such
explicit cash flows is possible, encouraging direct payment of the inflation stream. Other
inflation linkages are subtler, for instance supermarkets, whose sales will be similar to the
inflation basket and so their prices will rise in a similar vein. It is only in countries such as the
UK and South Africa, where there is a significant inflation risk premium, that there has been
issuance from issuers with such indirect inflation linkages though.
Corporate balance sheets are full of real assets, so offsetting these with real liabilities ought
to be appealing. This should be the case, particularly for firms whose physical assets are a
high proportion of their total value. For example, a land-owning company might be a
natural issuer of inflation even if it does not have direct revenue linked to inflation. Also, just
as having inflation exposure within a government debt portfolio acts as a fiscal stabiliser,
having it within the portfolio of a corporate can tend to act as a profit stabiliser. This is
particularly the case for industries with strong cyclical cash flows, eg, white goods retailers,
even when there is no clear long-term revenue link to inflation. With accreting inflation-
15 March 2010 17
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
linked bond cash flows skewed later than those of nominal bonds, coupled sometimes with
slightly more depth of demand for long-dated inflation linked than nominal exposure, this
can make inflation-linked issuance appealing to firms wanting to pay for their investment
with longer-term funding.
The benefits of extending investor reach by issuing inflation-linked bonds can be more
important for corporate issuers than governments. Companies can more easily take
advantage of issuing into specific pockets of demand than governments, for instance
issuing more structured bonds and using inflation derivatives to align this with their liability
needs. On the other hand, there are times when it is cheaper for firms to align their needs
specifically using inflation swaps rather than issuing bonds, particularly for smaller
amounts. The lower liquidity of inflation-linked bonds than nominals can limit the depth of
demand for non-government issuance, as has been the case in the UK. Added to often
narrower assets swaps for inflation-linked government bonds than their nominal
counterparts, it is often cheaper for a corporate with explicit inflation-linked revenues to
issue a nominal bond and then swap this exposure into inflation. On the other hand, for
publicly-listed companies unable to hedge account for inflation swaps, the mark-to-market
valuations of long-term exposures may create unacceptable volatility in the results. This has
meant that in the UK in particular it is more common for unlisted companies, such as those
controlled by private equity firms, to pay inflation or real rates than public companies with
similar inflation-related cash flows.
The following chapter discusses more complex structured notes in more detail, but these,
as with simple year-on-year coupon linked bonds, very rarely involve any net inflation
supply into the market, with the inflation risk swapped out.
15 March 2010 18
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Bond Inflation-linked
price cash flows
Inflation -linked
bond buyer
15 March 2010 19
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Inflation derivatives
Khrishnamoorthy Sooben Inflation derivatives are now established products. Asset swapping has encouraged a
+44 (0) 20 777 37514 parallel development with the cash market, although the US remains a bond driven
khrishnamoorthy.sooben@ market with relatively limited swap activity. Liability-driven demand is an important
barcap.com feature of both the UK and euro markets, with the nature of liabilities in the former
having led to the development of a non-linear market. In Europe and to a lesser extent
Alan James the US, complex structured products issuance has emerged, and this has led to a market
+44 (0) 20 7773 2238 for inflation caps and floors.
alan.james@barcap.com
[
notional * (1 + fixedrate)
tenor
]
−1
Counterparty Counterparty
A B
⎛ InflationI ndex t + tenor ⎞
notional * ⎜⎜ − 1⎟⎟
⎝ InflationI ndex t ⎠
Source: Barclays Capital
One issue to be aware of is the potential confusion from the terminology used when
inflation swaps are traded. Paying or receiving in inflation swap parlance normally relates
to the inflation leg. The receiver/payer in an inflation swap will receive/pay accrued
inflation and will pay/receive the fixed rate. This is opposite to the convention in the
nominal swap market where the receiver/payer is understood to be receiving/paying the
fixed rate. Alternatively, a long/short position in an inflation swap implies receiving/paying
inflation (ie, the floating element) versus paying/receiving the fixed leg, which is again the
opposite of nominal swap parlance. Stating exactly which leg is being received and paid
clears any confusion.
The start date from which inflation accrues on the swap is 6 January 2010, as the spot date
is the trade date plus two business days. The inflation index reference value for 6 January
2010 is calculated as an interpolation between the October and November 2009 CPI values,
in the exact same way as for US TIPS. For standard UK RPI and euro HICPx swaps, the
lagging principles are notably different from the cash market as they trade on a non-
interpolated basis, with a two-month lag in the UK and three months in euro. This means
that a euro HICPx swap traded on any given day of a particular month will have the same
starting index reference value, which will be the published index value three months prior.
For example, all standard euro HICPx swaps traded during January 2010 pay inflation
accruing from October 2009 – referred to as the base month. In February 2010, the base
month for all swaps shifts one month and changes to December 2009.
The traded price indices in standard inflation swaps are naturally the same as for their
corresponding bond markets. On the other hand, settlement dates may be different from
bonds, as they are generally in line with nominal swap markets instead. In the UK, a t+0
settlement date is typical, while in the US and euro area, t+2 is standard. The settlement can
be especially important when dealing with non-interpolated swaps. In particular, care
should be taken when establishing new positions in euro swaps at the end of each month,
as not all market quotes at this time will be on a consistent basis, with some quoting on the
basis of a lag to the settlement date and others versus the execution date. Here, stating
exactly which base month is being traded avoids confusion.
15 March 2010 21
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
PRCPI t : projected CPI for index reference date t years after the start reference date
PRCPI tm : projected CPI for index reference date t years and m months after the start
reference date
= CPI 0 * (1 + ZCt )
t
We have: PRCPI t
ln (PRCPI t +1 ) − ln (PRCPI t )
m
We then approximate PRCPI t , before applying the seasonality overlay, as:
m
PRCPI tm = CPI 0 * exp{ [ln (PRCPI t +1 ) − ln (PRCPI t )]* m / 12 }* exp ∑ MS n
n =1
140
130
125
120
115
Jan 11 Jul 11 Jan 12 Jul 12 Jan 13 Jul 13 Jan 14 Jul 14 Jan 15 Jul 15 Jan 16
Source: Barclays Capital
15 March 2010 22
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Due to the seasonality overlay, the segments between two annual points on a projected CPI
curve will not be straight lines, but will have an oscillatory pattern. This is illustrated in
Figure 3 1, where projected annual FRCPIx values are computed from quoted ZC FRCPIx
swaps on 7 January 2010 and intermediate values calculated via interpolation plus a
seasonality overlay. From the projected CPI curve, a complete ZC swaps curve can be built
by calculating the annualised rate of growth of the index from the starting reference date to
each future reference date. The magnitude of the oscillation due to seasonality on the
projected ZC curve will naturally decrease in longer maturities as the ZC rate is expressed as
an annualised rate, as in Figure 4.
3.0
2.5
1.5
1.0
Maturity
0.5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Source: Barclays Capital
1
Only a five-year horizon is shown on the chart as the oscillatory pattern becomes less visible as the range of values
on the Y-axis is increased.
15 March 2010 23
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Projected index value for September 2014 at trade date = 215.3 * (1.0315^5)
= 251.41 (rounded 2dp)
Trend rate of inflation between October 2013 and October 2014 = ln(251.99) – ln (243.58)
= 3.392% (rounded)
15 March 2010 24
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
At the trade unwind date, the projected index value for September 2014 =
= Projected index value for October 2013 * exp (Trend inflation * 11/12) * exp (November
+ December +…..+ September m/m seasonal)
= 243.58 * exp (3.392% * 11/12) * exp (-0.392% + 0.083% +….-0.068%)
= 251.34
This is equivalent to working backwards from the projected index value for October 2014,
i.e.:
Projected index value for September 2014 = {Projected index value for October 2014/
exp(Trend inflation/12)} / exp (October m/m seasonal)
= {251.99/exp (3.392%/12)} / exp(-
0.027%)
= 251.34
To calculate the P&L on the position, we compare the notional capitalised on the basis of
the two projections for the September 2014 index value.
Notional capitalised using projection on trade date = (Projected index value for
September 2014 on trade date / September 2009 index value) * Notional
= (251.41/215.3) * £100mln
= £116.774mln
Notional capitalised using projection on unwind date = (Projected index value for
September 2014 at unwind date / September 2009 index value) * Notional
= (251.34/215.3) * £100mln
= £116.741mln
The capitalised notionals calculated correspond to the cash flows at maturity on the fixed
leg of a swap paying inflation from September 2009 to September 2014.
The difference in the cash flows is -£32922. This has to be discounted to the trade unwind
date by the zero coupon discount factor, which is around 0.86. This means that if one goes
long inflation on the 5y swap on 17 November 2009 and unwinds it with an equivalent
swap, the P&L (excluding transaction costs) would be -£32922 times 0.86 (ie, -£28314).
To look at the valuation of existing swaps from another perspective, carry each month will be
determined by the difference between the actual monthly inflation accretion and the traded
breakeven level plus monthly seasonal factor. Hence, even if inflation in a given month is
negative, if this decline is less than implied by the seasonal model then carry on the position
that is receiving inflation can be positive. Obviously, if the market price of the remaining
position moves the other way, then positive carry may be scant consolation and as with
inflation-linked bonds, carry is a much more significant factor at shorter than long maturities.
By definition, the fixed rate of a ZC inflation swap is set at inception such that the value of
the swap is nil. Hence, at inception, the swap has no sensitivity to the nominal discount
factor, ie, zero nominal duration. On the other hand, the position will be sensitive to changes
in the ZC inflation swap rate; this sensitivity is known as the inflation duration. Even though
it is a zero coupon instrument, the effective inflation duration of an inflation swap will be
15 March 2010 25
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
less than its tenor due to the cash flow being paid at the end of the period. A seasoned swap
trade is likely to have acquired value, for instance because breakevens have changed or
because realised inflation levels have been different from the levels implied by the swap at
inception. To determine the present value of a seasoned trade, its future expected final
value has to be multiplied by a nominal discount factor to arrive at its present monetary
value. Thus, the dv01 of an existing inflation swap with respect to movements in the
inflation breakeven will be determined by the ratio of its remaining tenor to one plus the
discount rate to the power of this tenor. As the inflation swap acquires value, it will start to
gain a residual nominal duration, as a lower nominal yield takes the present value closer to
the expected final cash value, although in the event of a negative expected final value, this
residual duration is also negative. For a spot position, this nominal duration is usually a
minor second order consideration, but the importance of effective convexity that stems
from inflation and rates tending to move in the same direction can assume notably more
importance for forward inflation swap positions.
Although not quoted nearly as much on dealer screens as the zero-coupon structure,
indicative y/y rates are easily derivable from the zeros in the same way as is true for coupon
interest rates from zero-coupon interest rates. For longer maturities, convexity distortions
can become relatively large, but most structured notes are quite short. The most common
15 March 2010 26
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
use of the y/y swap remains for hedging structured products, both new exposure and
unwinds of old bonds that have been sold back to issuing banks. The floating leg of a y/y
inflation swap is equivalent to a collection of consecutive forward inflation rates, and
therefore the y/y structure can be useful for hedging inflation caps and floors, either on a
stand-alone basis, or as features within structured products.
⎧ ⎡⎛ CPI m−3 ⎞ ⎤ ⎫⎪
Notional* dayfrac ⎪⎨ ⎢⎜ ⎟⎟ − 1⎥ + X %⎬
⎜ CPI
⎪⎩ ⎣⎝ m −15 ⎠ ⎦ ⎪⎭
Source: Barclays Capital
together the growing bond and swap markets and has encouraged a complementary
development since.
With very few natural payers of euro area inflation other than governments, there is a
natural tendency for asset swaps in inflation-linked bonds to be somewhat cheaper than in
nominal government bonds. However, with a range of government issuers, the potential
asset swap demand base is substantial, and has generally helped limit the degree to which
breakevens in the bond and swap markets trade out of line. This was not the case however
towards the end of 2008 and start of 2009, when European bond breakevens dropped
significantly relative to inflation swaps; but the fact that many speculative players and
natural buyers of asset swaps were no longer active meant that acute relative distortions
persisted over several months.
In the UK, there has been considerable asset swap activity in gilt linkers. However, up to
2007 this was dwarfed by asset swapping of non-government issuance, which drove the
development of the swap market. When such corporate supply was heavy, gilt linkers
sometimes traded significantly more expensive in asset swap than nominal government
issues. In 2008 and 2009, UK corporate inflation-linked issuance other than from Network
Rail was extremely limited, prompting gilt linkers to generally trade cheaper in asset swap
despite unprecedented gilt linker asset swap buying.
In the US there are very few sources of inflation other than TIPS, while Treasuries are not an
asset class that is bought by asset swap investors as they are too expensive. This left US CPI
swap breakevens much wider than TIPS breakevens, but prior to mid-2008 demand for TIPS
asset swaps developed whenever issues traded at a discount to the general collateral (GC)
repo rate spread to Libor offering a positive carry position. JGBi asset swap demand
developed along similar lines, with demand cheaper than Tibor-GC. The widening of
spreads between interbank and general collateral rates in the second half of 2007 helped
shake up positions in both markets even as global nominal asset swap liquidity suffered.
However, the deleveraging in 2008-09 distorted relative valuations to a larger extent than in
the UK or euro market.
For a market in which the inflation swap curve is relatively well defined, linker asset swap
pricing is best thought of as akin to that in the nominal market with the additional step of
transforming all real cash flows into nominal form, as ultimately the valuation is versus
Libor or equivalent nominal floating rate and not a real floating leg. Hence, the inflation-
linked bond cash flows should be projected into nominal space using the inflation swap
discount factors and then discounted by the nominal swap curve. The difference between
this valuation and that of the inflation swap curve will then provide the asset swap value.
15 March 2010 28
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
There are, however, several ways that this calculation can be done, three of which are
particularly relevant to inflation-linked bonds. The first of these is the par-par asset swap
calculation that is the conventional quoting method for euro inflation-linked asset swaps.
Most other markets use a proceeds asset swap calculation. Both of these methods create
inconsistencies comparing different bonds though and in particular comparing inflation-
linked bonds with nominals. For comparative purposes, we use a zero spread (z-spread)
calculation instead. While conceptually this method is straightforward, it is almost entirely
an analytical tool, with trading in this format extremely difficult as pricing is on an
iterative base that would make the paperwork intractable in anything but a contract for
difference format.
Set out below is an example of a par-par asset swap calculation for the OAT€i20. The
inflation index used to transform the real cash flows is that implied by HICPx zero inflation
swap breakevens, and the discount factors are of the Euribor curve. Being par/par, it is
necessary to net out the deviation from par of the bond’s dirty cash price, from the PV of
the par priced bond flows, and the PV of the Libor leg, to find the PV of the asset swap
assuming no spread to Libor. Via the swap PV01, the necessary spread to give the overall
structure a PV of zero can be found, which is the mid or fair-value asset swap spread, in this
case 42bp assuming negligible value to the embedded inflation floor. The par par asset
swap is not a particularly logical calculation to use for bonds, which will tend to deviate
from nominal par over their lifetime as they gain inflation accretion, but it is relatively
straightforward and became convention in the euro area when most asset swaps were on
short maturity bonds with limited history. It should be highlighted that the frequency of the
Libor payments will not necessarily be the same as the cash flows on the bond in an asset
swap. For example, the underlying nominal swaps market in Europe most commonly trades
on a semi-annual payment convention, whereas OAT€is carry annual coupons.
15 March 2010 29
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Proceeds asset swap spreads are generated in a similar manner to par-par calculations with
some minor differences. In a proceeds spread, there is no exchange of cash flows upfront as
the swap notional is set to be equal to the bond’s dirty price. Therefore, the biggest
difference is that the bond premium (or discount) in step C of Figure 6 must be discounted
from the maturity date, reflecting the fact that the swap notional will be different from 100
at the final maturity of the asset swap. Proceeds asset swaps avoid any issues with historical
accretion and hence are more appropriate for markets where bonds have been accreting
inflation for years, as most TIPS had when the market took off and particularly in the UK. A
par par asset swap tends to exaggerate the value of the asset swap for a bond trading above
par, as bonds with inflation accretion will usually be. For comparison, we consider the
proceeds asset swap on the same OAT€i20 below, although euro bonds are rarely quoted in
this format.
Neither par-par nor proceeds asset swaps are an ideal valuation measure to compare
inflation swaps versus bonds. The back-ended cash flows of inflation-linked bonds mean
that the value of a floating basis point now is less than that of an inflation-linked bond
discounted on its own real yield curve. Hence, a 1bp move in real yield will be worth more
than a 1bp in asset swap in either of these methodologies. Deviations of asset swap levels
away from Libor flat will tend to become more distorted the longer the maturity of the
linker. In an extreme case, a 50y bond may have an expected average principal in present
value terms around twice its current level and hence a 1bp move in the real yield of the
15 March 2010 30
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
bond with no move in the inflation swap curve will move the asset swap by around 2bp. It is
thus relatively difficult to compare asset swap levels across inflation-linked bonds and
particularly difficult to compare them with asset swap levels of nominal bonds. For this
reason, we prefer to use a z-spread asset swap methodology.
Z-spread asset swaps are a widely used analytical tool within nominal bonds to smooth out
micro distortions and compare relative value, but are more important for linkers as the
distortions are notably larger. The calculation of the z-spread asset swap is done so that, by
construction, a 1bp move in the real yield of the bond will move the asset swap by 1bp as
well. A z-spread asset swap spread between a nominal and inflation-linked bond thus
provides a consistent measure to the richness of bond versus swap breakevens at that
maturity. The calculation involves taking the cash flows of a linker in a similar way to a
proceeds asset swap and calculating their present value. The next step is to iteratively adjust
the uplift factor affecting the bond cash flows in a parallel fashion until the present value
matches that of the same cash flows priced from the swap leg. The amount that the swap
curve has to be bumped to match the value of the bond cash flows is the z-spread asset
swap. It will be narrower than asset swaps calculated by the other methods, substantially so
in the case of long maturity bonds.
15 March 2010 31
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Figure 8: Example of zero-spread (z-spread) asset swap calculation (using the OAT€i20)
Bond price 123.16
PV 126.57 123.16
Difference 3.41
Z-spread 0.29%
Source: Barclays Capital
While the z-spread asset swap is an ideal analytical tool, it is a very difficult format to trade.
There have been multiple attempts made to make a more representative tradable asset
swap measure for inflation-linked bonds, for instance by allowing the value of Libor to
accrete at an agreed rate upfront (either the breakeven inflation rate or a conventional
amount) that will compensate for the back-ended linker cash flows. Alternatively, linker
asset swaps could be reset with each coupon to allow for the accretion of the bond, fixing
the accretion rate at the actual rather than expected level, but this would involve a new
transaction after every coupon point. Conceivably, such a reset process could be less
frequent, eg, every five years, and it would still remove much of the distortion. This
approach may be feasible as long-term swap contracts usually have break clauses
embedded within them anyway so they could instead be reworded as a reweighting clause.
Any such attempt to prevent distortions remains fraught with difficulties, though, and as
practitioners become comfortable with using z-spreads for analytics but asset swaps for
trading purposes, the calls for an alternative measure may reduce.
15 March 2010 32
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
balance sheet. The total return of the bond or bonds is received versus paying out a floating
rate cash flow. The spread applied to the floating side of the swap is required to
compensate the dealer for the balance sheet cost of holding the asset or assets underlying
the swap agreement. In markets such as US TIPS, where general collateral repo yields are
significantly lower than Libor, receiving a total return swap is likely to be at Libor minus,
whereas in less liquid or lower-rated markets, it is likely to cost more than Libor to receive
the total return. The credit rating of the institution offering the swap can also have a bearing
on the pricing, with those with a lower rating potentially requiring less spread to hold the
assets on balance sheet. While there is a counterparty credit risk involved in a total return
swap, this can be mitigated by ISDA/CSA agreements or by frequent resets of the swap.
The market for total return swaps in developed inflation-linked market indices has increased
sharply, partly due to the movement toward separating ‘Alpha’ and ‘Beta’ in portfolios. For
example, active managers can use a total return swap on the Barclays US Inflation-Linked
Index to gain their Beta exposure, simultaneously freeing up capital to pursue Alpha-
generating trades. The total return swap effectively converts their benchmark from the
index to Libor plus the spread charged for using the dealer’s balance sheet. In the US or
Europe, it is likely to be a relatively cost-effective leveraging strategy for investors who are
unable to repo bonds directly themselves. The additional advantage for indexed investors is
they do not have to rebalance their portfolio exposure as their beta index changes once it is
held in total return swap format rather than in physical bonds. Total return swaps can also
be used as a way for investors to short the bonds without directly having exposure to the
repo market. It is important to note that if the total return of the underlying instrument is
negative, then the counterparty receiving the economic performance will have negative
cash flows on both legs of the total return swap. One notable feature when distortions
between bonds and swaps were very high in late 2008 was the use of total return swaps to
express linker asset swap positions.
notional × totalretur nt ,t −1
Counterparty Counterparty
A B
notional × (Libor ± spread )
15 March 2010 33
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
2004 saw the emergence of leveraged notes, which typically paid a multiple of y/y inflation
or a fixed rate plus a multiple. Issuance in Europe during 2004, 2005 and 2006 was €10bn,
€3bn and €7bn, respectively, with the total not significantly above that of 2003, but the
leverage factor in most notes issued during these three years implied a much greater impact
on the swaps market. Coupons were floored, although floors set below zero were relatively
common in 2004. The leverage factor amplified not only the impact in swaps space but also
the implicit notional on any embedded floors. One format that had a particularly important
effect on the underlying markets was notes that offered a leveraged exposure to the spread
of euro and French inflation, often forward starting. These notes helped to correct a
relatively extreme richening of French swaps versus euro inflation that had been driven by
strong Livret-A related hedging demand for French CPIx. Similar notes have also been sold
with a UK RPI versus euro HICPx spread. In addition to having an inflation floor, sometimes
higher than zero percent, such leveraged notes often have an inflation cap embedded within
them as well. A combination of inflation and nominal CMS coupon-linked structured notes
proved to be relatively popular with investors, particularly in 2006. For instance, a note
might pay the better of inflation or 10y CMS, or alternatively the worse of the two plus X%.
Similarly, floating rate notes have been offered at Euribor + Y%, for example, but capped at a
multiple of euro HICPx.
In 2006-07, the main focus in euro inflation-linked structured note issuance was on inflation-
range accruals, which typically aimed for the y/y inflation to remain in a tight range around
the ECB’s target of inflation close to but below 2%. In such structures, a large fixed nominal or
floating rate would be paid if inflation remained within the range, compared to a low or no
payout otherwise. Buyers of such range accrual notes were effectively sellers of inflation
volatility. These products therefore offered interesting value as implied cap and floors volatility
was relatively high at that time, in contrast to low realised inflation volatility. Range accrual
notes helped push down cap/floor vol in the euro area significantly, but investors in these
bonds suffered when actual inflation started to move well above the upper end of the ranges
offered towards the end of 2007. With realised inflation pushing towards 3% in 2007,
cap/floor implied volatility drifted higher as some assumptions in pricing models were
changed but also because of some unwinding of the loss-making range accrual notes.
15 March 2010 34
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Italy was the country in which the demand for structured inflation notes took off most
notably. The majority of notes sold there in 2003 and 2004 went to individuals via bank
distribution networks. While most issues sold in Italy were linked to euro inflation, there were
as many notes linked to Italian inflation in these two years as total issuance linked to US CPI.
In 2006 and 2007, there were inflation coupon notes sold linked to at least seven different
euro area domestic inflation measures, most commonly Spanish CPI, plus euro-denominated
issuance in Danish and Polish CPI. A variety of other emerging market inflation indices were
also offered in y/y inflation note formats, both in local and hard currencies. Risk tolerance
subsequently fell, however, and even before the worst of the market turbulence in 2008,
there was a tendency towards relatively straightforward products.
Figure 10: Structured note inflation issuance in US CPI and euro area indices (bn)
25
US Euro (US$ equivalent)
20
15
10
0
2002 2003 2004 2005 2006 2007 2008 2009
Source: Barclays Capital
In 2008, the fear of high inflation boosted demand for inflation-linked structured notes from
retail investors. Products paying inflation with leverage once again became widespread.
Many of these notes were structured with a higher leverage than before and with floors
above zero. Issuance also thrived in the first half of the year as banks in particular took
advantage of expensive credit/funding for financials. Even though implied vols and swaps
surged to unprecedented levels, the high funding levels meant that sufficiently attractive
pay-offs could easily be structured to cater for the retail investor base. After summer in
2008, issuance dried out, as focus turned to deflation while an explosion in volatility made
the cost of floors embedded within most structures extremely expensive. Issuance started
to recover again in Q2 09 after global central banks had engaged in quantitative easing
strategies, which significantly increased inflation fears despite negative realised inflation,
but leveraged notes have been scarce.
Hybrid inflation-linked notes have been a feature of the structured product market almost
since its inception. One of the first large structured inflation notes in the euro area, sold by
the Italian Post Office in 2001, was linked to the performance of the Italian MIB Stock Index
but with the return floored at an inflation-protected principal. Similar best-of-hybrid
products have also been seen, for instance notes linked to the S&P500 and the Euro Stoxx
indices. Other hybrid notes offered the best of inflation accretion and commodity basket
performance. All relied on a sufficiently high real yield to leverage up on the risky leg, and in
the euro area in particular, demand for this kind of product fell sharply as real yields fell in
2005. Inflation-linked credit products, for instance inflation-linked CDOs, then became more
popular, helping to lead to a rebound in total structured note issuance in 2006. While more
highly structured inflation-linked notes are unlikely to see any significant resurgence for
some time, there is scope for hybrid notes to return as real yields rise.
15 March 2010 35
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Caps and floors are most commonly quoted at 5y and 10y maturities, with a concentration of
interest at integer values, 0%, 1% and 2% floors and 2%, 3% and 4% caps. However, the
development of the range accrual market in particular has helped better define the cap/floor vol
smile, as the strikes on these notes occurred at a varied range usually between 1 and 1.75% on
the floors and 2.5% to 3.25% on the caps. While caps and floors related to structured notes have
traditionally been the most important flows, by 2007 the market started to reach a point where
flows unrelated to underlying product were becoming almost as important. Participants in other
vol markets are increasingly getting involved in inflation caps and floors, while activity at the long
end has also picked up notably on the back of pension fund demand.
Inflation caps and floors are traded in most of the major inflation bases and not just because
of embedded floors in structured notes. The UK cap and floor market is relatively well
defined as a result of LPI curves, given the preponderance of pension liabilities that are LPI
linked, even though there was very little activity in LPI swaps in 2008 and 2009 as realised
inflation volatility exploded. The skew stemming from structured note floors in the euro
market has been traditionally extreme. High floors could previously be structured in notes
as they were deep out of the money when inflation swap rates were stable around 2%.
When structured note demand in the US drove cap/floor vol there to several times that in
euro HICPx, there was flow to hedge one with the other and notes also offered on this basis.
As swap rates fell towards the end of 2008, having spiked higher earlier in the year, the
sizeable street short in floors as a result of past structured issuance meant that losses had to
be marked-to-market, creating a rush to hedge, whether in floors or swaps. The delta-
hedging led to a self-reinforcing bearish momentum in euro HICPx swaps and cap/floor
implied vol in the euro area shot higher. The same pressures were seen even more
extremely in the US despite the size of exposures being smaller.
In a market where there is no underlying supply of inflation volatility, pricing of low strike
floors comes most obviously from hedging with low strike nominal options. This became
very difficult in the US and euro area in late 2008, with this market also relatively distressed.
It was never a feasible proposition in Japan, which combined with initial restrictions on
investment on bonds without a guaranteed principal led to floors pricing in Japan at
extremely expensive levels. This was only partially offset by some notes including caps too,
with the vol skew relatively extreme. It did prompt some structured notes being sold
elsewhere in the world taking advantage of this pricing though. Coupled with the relaxation
of investment restrictions on JGBis, which effectively killed off the market for competing
products whose main additional feature was principal protection, prices of 0% floors in
Japan became less expensive in 2007, even though breakevens were notably lower than in
previous years. Nonetheless, activity in Japanese core CPI floors became focused on final
maturity, most commonly 10 years, rather than a yearly basis, limiting the degree to which
it was the options market that drove dislocations in late 2008.
15 March 2010 36
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
The inflation volatility market in France grew as a result of liability-driven demand linked to
the Livret A savings rate. The y/y inflation linkage involved in this rate meant a significant
degree of convexity pricing versus zero coupon swaps even though it had no theoretical
floor attached to it. From 2004-08, the rate was computed as an average between 3-month
Euribor and the y/y FRCPIx rate. From 2008, an element of non-linearity was added to the
formula, with it being the minimum of French CPIx y/y +0.25% and 0.5x the sum of CPIx
and the average of 3mth Euribor and eonia. If products replicating the more exotic formula
were structured, this would have significantly increased the implied vol at lower strikes in
both inflation swaptions and caps. However, with the rate being set arbitrarily for most of
2008 and 2009, there has been no demand to precisely hedge the new formula. At the start
of 2009, the official rate-setting was further incremented with the condition that the
magnitude of a rate change cannot exceed 1.5%. It was also decided that the Banque de
France could recommend a change in the rate even between the twice-yearly revision dates.
All the above contributed to weaken the once strongly perceived formulaic link between
French inflation and the Livret A rate.
LPI is interesting from a modelling perspective because the annual accretion means that it is
path-dependent, meaning that simple cap/floor vol is only an approximation and that
simple two currency inflation or real rate volatility models will not provide closed solutions.
It does help define a broad smile in the UK market though, with supply of LPI from property
rentals often at non-standard levels (albeit rarely outside the 0-5% range). Models that were
developed to price LPI swaps can generally be extended to capture most other non-linear
pricing that has been seen in the inflation and real rate space. A result of this is that the
development of the UK LPI swaps market in the last decade encouraged complexity to be
added elsewhere. Modelling of inflation volatility product is discussed in more detail later in
this publication in the section ‘Inflation Volatility Pricing Considerations’.
LPI is not a unique concept to UK RPI. In particular, some pension liabilities linked to Irish
CPI are also in an LPI format, capped at 4% and floored at 0, which used to be the legal
minimum. Another significant inflation liability that involves an LPI formula, albeit one
15 March 2010 37
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
where the cap and floor are so far out of the money that it is often ignored, is the so-called
TFR rate in Italy. The accretion of TFR funds, traditionally a percentage of wages withheld
and then paid when a worker leaves a company but now increasingly transferred into
pension holdings, is 1.5% + 75% of Italian FOIx CPI inflation but with a cap of 7.75%, ie, at
8.33% inflation and at least implicitly with a floor at -2% inflation.
TIPStions
Options on TIPS, known as TIPStions, and options on TIPS breakevens have been trading in
the OTC market since early 2005, though activity in 2009 was almost non-existent. The
evolution of the TIPStions market was a natural extension of the TIPS market in that it gives
portfolio managers and other real money investors an easy way of managing portfolio risk
and returns without having to make outright purchases or sales of their TIPS holdings. It
also allows leveraged investors to implement strategies they have used in the cash market
in a more efficient manner. A call TIPStion gives the buyer the right to purchase TIPS at a
preset price on the expiration date of the option. The market convention for quoting the
options is typically in terms of at-the-money spot. Quoting in this manner reduces the
complications involved with calculating forward prices in TIPS. TIPStions pay off just like
standard options with the expected pay-off for a call being the maximum of the price at
expiry minus the strike and zero, whereas a put TIPStion pays the maximum of the strike
minus the price at expiry and zero. While TIPStions are simple products at first glance, there
are some challenges that makes pricing non trivial. We look at the pricing aspect in more
detail in the ‘Inflation volatility and delation floors’ section.
Swaptions
Swaptions in both UK RPI and euro HICPx area can be quoted, although there is no
interbank market. Both real rate and breakeven swaptions can be traded. A well defined
swaps curve, which is the case in both UK RPI and euro HICPx facilitates forward pricing in
swaptions, while this is not trivial for bond options.
Swaptions in the euro area have involved relatively short tenors. Prior to 2009, the presence
of inflation futures at least helped to offer a fully visible, albeit illiquid, curve off which to
price such exposures for less than one year. The development of pension fund interest in
swaptions helped drive demand for longer-term expiries as well as maturities. The more
complete definition of the inflation swaps curve makes this pricing more straightforward.
The presence of conditional inflation indexation of Dutch pension funds has created the
prospect for the supply of inflation volatility, as well as the Netherlands becoming an
increasingly important source of demand. Nominal swaption volatility remains the most
important driver of real yield vol, while activity in the inflation cap and floor market helps to
define breakeven swaption vol pricing.
In the UK, interest in the swaptions market is usually for very long-dated expiries and
maturities, with most options being for five years or more, while the most common
maturity is 30 years. What limited structured inflation note market there is also
encourages activity in long tenors, with increasing activity in long-term notes that are
callable or puttable. Swaption activity remains uncommon compared to LPI activity, but
the definition of the LPI swaps activity helps to provide a reference volatility curve and
smile as a starting point for pricing.
15 March 2010 38
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
INFLATION MARKETS
15 March 2010 39
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
US
Michael Pond The US Treasury began issuing inflation-linked bonds, most commonly known as TIPS,
+1 (212) 412 5051 in 1997. By 2000, TIPS had overtaken the UK inflation-linked market to become the
michael.pond@barcap.com largest market of its type, reaching a value of $600bn by February 2010. The Treasury
has varied its issuance pattern over time and currently issues at 5y, 10y and 30y
maturities. Despite the depth of the TIPS market and the commitment from the
Treasury, corporate and inflation swap activity is limited, although total return swaps on
TIPS are regularly used for leveraged exposure.
The basket of goods and services and the item weights are determined from the Consumer
Expenditure Survey (CEX). Since the CPI is a fixed-weight index, the implicit weights remain the
same from month to month. A related concept is the relative importance of an item. Relative
importance in essence means that if the price of a particular item rises more than the average
price increase of items in the basket, then the relative importance of that item increases. To
illustrate, the price of crude oil, as measured by the WTI, had risen from around $20 per barrel
in January 2002 to near $90 per barrel in December 2007. A result of that increase is that the
relative importance of energy rose from 6.2% to 9.7% during the same time period. The charts
highlight the change in the relative importance of the eight major categories from 1997 to
weights set in December 2008. Annual weights are typically released in mid-February for the
previous year.
Figure 11: 1997 CPI-U weights Figure 12: December 2008 CPI-U weights
Transportat Transportat
ion 17.6% ion 15.3%
Housing
Housing
39.6% Apparel
Apparel 43.4%
3.7%
4.9%
Source: Haver Analytics, Barclays Capital Source: Haver Analytics, Barclays Capital
As can be seen from Figure 12, one of the most significant categories in terms of weights is
housing, over half of which is an imputed measure called ‘Owners’ equivalent rent of
primary residence’ (OER), which attempts to capture price changes if those consumers who
15 March 2010 40
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
own their home were to rent instead. BLS measures the change in implicit rents over time
by matching owner units to renter units with similar characteristics. The characteristics
included location, structure type and other general traits such as age, number of rooms and
type of air conditioning. As owners pay for utilities separately, BLS calculates the pure rent
of the matched renters by removing the value of any landlord-provided utilities and
furniture. As utility prices tend to fluctuate more than actual rents, imputed pure rents tend
to be negatively correlated with utility prices, primarily natural gas prices. Before 1983, the
BLS used an asset price approach in computing the shelter component of CPI; because this
method was driven by interest rates and house prices it was much more volatile than the
current method and core CPI volatility has declined since.
Energy prices can also have a significant impact on m/m CPI prints and in fact are
historically responsible for over 50% of the volatility there. Energy currently makes up
around 8.7% of the total CPI basket. This weight has grown since TIPS were first issued, so
energy is even more important to inflation-linked investors now than it had been in prior
years. While the BLS only publishes the relative importance for each December, the
weightings do change from month-to-month. As seasonal factors affecting Energy CPI are
at their worst in December, the BLS relative importance data understate the average weight
of energy. Gasoline (Motor Fuel) is the most important component of Energy CPI, both
because its weight is higher and because it tends to be more volatile than the other
components, electricity, home fuel oil and utility gas service (natural gas).
5/16/1996 Treasury Secretary Rubin announces the intention to issue Treasury inflation-
indexed securities (www.treas.gov/press/releases/rr1073.htm)
9/25/1996 President Clinton and Treasury Secretary Rubin announce the terms and
conditions of the first Treasury inflation-indexed security
1/29/1997 First 10y TIPS auction
4/9/1997 First 5y TIPS auction
4/8/1998 First 30y TIPS auction
6/30/1998 Final rules on fungible inflation-indexed STRIPS published
9/1/1998 Treasury begins selling series-I savings bonds
9/29/1998 Treasury announces regular quarterly schedule for TIPS and discontinues 5y TIPS
1999 Fed conducts first TIPS pass
11/30/2000 TIPS are stripped for the first time
10/31/2001 Treasury eliminates 30y TIPS because of lower borrowing needs
7/15/2002 First 5y TIPS matures
4/18/2002 Treasury conducts TIPS buyback
4/30/2003 Treasury expands 10y TIPS auctions to 4 per year with two new issues per year
2/8/2004 CPI futures begin trading at CME
5/5/2004 Treasury announces the introduction of 20y TIPS and reintroduction of 5y TIPS
7/27/2004 First 20y TIPS auction
10/26/2004 First reintroduced 5y TIPS auction
1/15/2007 First 10y TIPS matures
1/22/2008 TIPS Index market value hits $500bn
2/22/2010 First reintroduced 30y TIPS auction
Source: US Treasury, Barclays Capital
15 March 2010 41
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
There was limited initial support for TIPS as investors used them primarily as a tactical
trading vehicle. The small number of participants resulted in both low trading volume and a
low beta relationship to nominal yields. Breakevens were generally the main catalyst for
investment decisions. In November 2000, the iSTRIPS market was launched when the TII08s
became the first TIPS to be stripped. iSTRIPS allow investors to trade the TIPS coupon and
principal components separately, where the principal component carries the floor, although
to date there has been only scarce interest in iSTRIPS. While iSTRIPS have not been a huge
success, it was important that the Treasury encouraged stripping activity as a sign of it’s
commitment to the TIPS program when many market observers were questioning the
durability of the asset class. Despite this commitment, the Treasury reduced TIPS issuance
commensurate with reductions in the nominal calendar until only an annual 10y note, with
just one re-opening auction, existed in 2001 (Figure 14).
With five years of history and the 5y TIPS issued in 1997 having matured in 2002, the TIPS
market finally started to gain broader acceptance. Consultants began recommending TIPS
in earnest and due diligence and approval processes were introduced. There was also
increased interest in real return mutual funds and other funds tied to the TIPS Index as
investors began to make diversification allocations into TIPS as a new “asset class”.
Increasing demand led to significant growth in 2004, as the Treasury issued nearly as many
TIPS that year as it did in the previous three. They also announced a major expansion of the
program to include two 5y auctions and two 20y auctions per year, in addition to the
existing quarterly 10y note auction cycle. Alongside growth in the cash market was a
developing inflation derivatives market. CPI futures began trading at the CME in early 2004
and volume in the CPI swaps market increased significantly along with issuance in inflation-
linked corporate notes. The US inflation market continued to develop in 2005, particularly
on the derivative and structured note side.
From 2005 to mid-2008, average daily trading volume leveled off because of an increase in
demand from structural investors such as pension funds and insurance companies where
investments tend to be passive in nature. Related to this demand, there was significant growth
in Inflation-Linked Total Return Swaps activity from investors looking to receive the return of
the TIPS Index or a Global Inflation-Linked Index in a passive nature. There was a significant
shift in the investor base beginning in the second half of 2008. Hedge funds, after making up
around half of TIPS market flows in 2007, largely exited the asset class during the financial
crises. After the market cheapened significantly following this deleveraging process, real
money investors began to increase structural allocations significantly. Foreign central banks
began to buy because of diversification benefits and as a de-facto currency hedge. Domestic
15 March 2010 42
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
real money increased structural allocations in part because of medium-term inflation risks
associated with stimulative fiscal and monetary policy and also because of a realization by
many that the right allocation to TIPS within a well diversified portfolio is not zero.
90
80
70
60
50
40
30
20
10
0
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
(expected)
15 March 2010 43
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Figure 15: TIPS issuance as a % of all Treasury note and bond issuance
12%
10%
8%
6%
4%
2%
0%
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Estimate
TIPS as % of Total Gross Coupon Issuance
Source: US Treasury, Barclays Capital
Increased issuance has driven an improvement in liquidity as well. In the initial years of
market development, trading volumes tended to spike only around auction weeks as they
were seen as liquidity events in an otherwise low-volume product. While auction periods are
still seen as liquidity events, this pattern has changed and trading volumes are more
consistent. While trading volume increased only slightly over the 2000-02 period, when it
averaged $1.87bn per day, average daily trading volume increased significantly in late 2004
and was $3.73bn in 2003, $5.95bn in 2004 and took a sizable jump up to $8.77bn in 2005.
Trading volume has since levelled off around $8-10bn/day. We believe this trend is being
driven by increased activity by long-term investors whose structural investments tend to be
passive and hence of a lower turnover than more tactical investors. The significant increase
since 2005 in investors looking to receive the returns on the TIPS Index on a Total Return
Swap, which by nature is a passive vehicle, supports this view. Since the middle of 2008,
average daily trading volume has declined to a trend near $4.5-5bn. This is because of the
structural shift from hedge funds to real money investors, who typically take more of a buy-
and-hold approach to investing and so tend to trade less.
12%
10%
8%
6%
4%
2%
0%
Feb 97 Feb 99 Feb 01 Feb 03 Feb 05 Feb 07 Feb 09 Feb 11
Source: US Treasury, Barclays Capital
15 March 2010 44
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Figure 17: TIPS market value and average daily volume ($bn)
12
550
10
450
8
350
6
4 250
2 150
- 50
Jan-00 Mar-01 May-02 Jul-03 Sep-04 Nov-05 Jan-07 Mar-08 May-09
Returns
Through 2009, TIPS have had an annualized since inception return of 6.65%. This compares
to the 6.57% annualized return on a basket of comparable maturity nominal Treasuries. The
greatest annual return for TIPS was 17% in 2002 and the worst was -1.7% in 2008, the only
year that the TIPS Index had a negative return. Relative to the nominal comparator Index,
the biggest underperformance was -18.4% in 2008, but that was followed by the largest
outperformance of 16.2% in 2009.
Figure 18: TIPS – historical performance and risk Figure 19: Return/risk versus nominals and equities
3
10%
2
1
5%
0
0% -1
-2
-5% -3
1999 2001 2003 2005 2007 2009 1998 2000 2002 2004 2006 2008
TIPS structure
TIPS, along with most major inflation-linked bond markets except the majority of issues in
the UK, follow the Canadian model, where the security pays a fixed coupon on the inflation-
adjusted principal. The principal is adjusted on a daily basis using an index ratio that
quantifies the rate of growth in inflation or deflation between the issue date and settlement
15 March 2010 45
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
date. The index is lagged three months from the settlement date; for example, for 1 January
2010, the CPI-U for October 2009 applies. We compute the index ratio as follows:
Index Ratio = Reference Index/Base CPI Index; where the Base CPI Index is the Reference
Index at issue date and,
where:
For settlement amounts, real accrued interest is calculated as for ordinary Treasuries. Clean
price, which is the trading price and does not include either the inflation or coupon accrual,
and accrued are each multiplied by the Index Ratio to arrive at a cash settlement amount.
For coupons paid, the (real) semi-annual coupon rate is multiplied by the Index Ratio, and
likewise for the par redemption amount (with the cash value subject to the par floor).
Floor
In addition to the above structure, TIPS have an embedded floor such that at maturity, the
investor gets the greater of par or the inflation-adjusted principal. Since the inflation-
adjusted principal is the par amount times the index ratio (which is the ratio of the reference
CPI to the base CPI), this is another way of saying that, at maturity, the index ratio is floored
at 1 as it applied to the principal. The pay-off on the principal amount at maturity can be
written as:
It is important to remember that the “strike” on the floor is at par, or an index ratio of 1, not
where the index ratio is at the time of purchase. For this reason, the floor value of newer
bonds tends to be more valuable because the index ratio is typically lower than seasoned
TIPS. As an example, the TIIApr13 Index ratio, for a trade that settles on January 25, 2010, is
1.02332. This means that there has been 2.33% cumulative inflation accrued since issuance
in April 2008. The floor would kick in if the index ratio fell below 1, so the inflation accrued
since issuance would first need to be fully reversed out in a period of deflation. With about
3.2 years left, there needs to be 0.71% annualized deflation to maturity for the floor to be at
the money at maturity.
15 March 2010 46
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Before September 2008 the market put very little value on this embedded option. However,
during the financial crises when breakevens out to the 9y turned negative and investors
were increasingly risk averse, the value increased significantly: for example, at one point the
real yield spread between TIIApr13s and TIIJul13s was 200bp with most of this difference
explained by the floor. The non-linear inflation market began quoting cumulative caps and
floors around early 2009 and the value of the embedded option could then be priced
separately from TIPS.
Tax
On August 25, 1999, the Internal Revenue Service published ‘Final regulations’ covering the
tax treatment of inflation-indexed instruments. Investors should consider the entire
document, but a key paragraph is detailed below:
“The final regulations provide rules for the treatment of certain debt instruments that are
indexed for inflation and deflation, including Treasury Inflation-Indexed Securities. The final
regulations generally require holders and issuers of inflation-indexed debt instruments to
account for interest and original issue discount (OID) using constant yield principles. In
addition, the final regulations generally require holders and issuers of inflation-indexed debt
instruments to account for inflation and deflation by making current adjustments to their
OID accruals.”
Thus, the inflation escalation of principal in the US is taxable as income annually, even
though the Treasury will be making the inflation payment at maturity. This creates a
phantom inflation tax, which for non tax-exempt investors such as insurance companies
and individual investors may make ownership in TIPS unattractive. To ameliorate this
problem the Treasury in 1998 issued a Series I Savings Bond program targeted at individual
investors. These bonds are tax exempt for 30 years.
Owing to this ‘phantom income’ tax issue associated with TIPS, many retail or other taxable
investors view nominal Treasuries and corporate inflation-linked notes as more tax efficient.
While it is true that TIPS are disadvantaged from a cash-flow perspective, we have shown
that they are not necessarily penalized on the expected after-tax total return versus nominal
Treasuries. Other inflation-linked structures pay out the inflation on a monthly coupon
rather than accreting on the principal, so the investor is taxed on income actually received.
However, as typically state taxes are paid on these bonds as well, the advantage is not as
clear cut as many perceive.
Rules and regulations governing the tax treatment of TIPS can be found at the following
link: ftp://ftp.publicdebt.treas.gov/gsrintax.pdf.
15 March 2010 47
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
proved natural buyers of TIPS. The insurance sector is notably less important than in Europe,
mainly because inflation-linked life policies are much rarer, although inflation-linked
structured issuance by insurance companies has become more common.
One of the main reasons the Treasury reintroduced 5y TIPS issuance in 2004 was to
encourage central bank buying. Foreign official institutions have become an increasingly
important feature of the market, but until 2009 remained relatively small in TIPS compared
to their nominal Treasury holdings. At the start of 2009 we began to see structural
investments in TIPS from foreign central banks and sovereign wealth funds because of
diversification benefits and as a de-facto currency hedge.
Pension reform may encourage more buying of TIPS by private defined-benefit pension
schemes, but since most private funds do not have explicit cost-of-living adjustments
(COLAs) allocations to TIPS would be mainly for diversification purposes. The absolute scale
of private-defined benefit inflation-linked liabilities is considerably smaller than the more than
$2trn state and local government sector. Importantly, state and local government pension
scheme liabilities have relatively more explicit price indexation rather than wage indexation,
and also much more frequently have indexation commitments beyond the period when a
member of the pension scheme is an active contributor. State and local government schemes
are already the largest pension fund buyers of TIPS and their liabilities mean the potential for
increased buying is substantial. On the other hand, federal pension reform is unlikely to affect
the state sector significantly, so the relative importance of the private sector defined-benefit
sector may increase. As in other countries, to date few TIPS have been bought for defined-
contribution pension schemes, which remain skewed heavily towards equities.
iSTRIPS
The US Treasury over the years has developed the inflation-indexed security market in a
similar fashion to the nominal Treasury market. Hence, the development of a full yield curve
has led to increased issuance, increased investor demand and the development of a deep
and liquid market. Allowing TIPS notes and bonds to be stripped into zero coupon
instruments is another step in this development process. STRIPS is an acronym for Separate
Trading of Registered Interest and Principal Securities. A TIPS security can be divided into its
two components – coupon and principal. Each coupon cash flow, along with the principal
payment is made into a real zero-coupon instrument, or iSTRIP. All TIPS issues are now
eligible for stripping and Barclays Capital has been an innovator in this area, first stripping
TIPS in November 2000. To date there has been only scarce interest in iSTRIPS. Most trades
have been in lieu of structured products or derivatives. While iSTRIPS have not been a huge
success, it was important that the Treasury encouraged stripping activity as a sign of its
commitment to the TIPS program when many market observers were questioning the
durability of the asset class.
The US Federal Register sets forth basic conventions for the stripping and future settlement
prices of zero-coupon inflation instruments. The complete formulas may be found at the
following link for CFR 356.36 Appendix B. The link for this register is as follows:
www.access.gpo.gov/nara/cfr/waisidx_02/31cfr356_02.html.
Principal component
There is only one principal component (corpus) per TIPS issue. The par amount is the
original face value of the bond to be stripped in $1,000 increments. The principal
component retains one of the key attractions to TIPS. The embedded floor in TIPS only
15 March 2010 48
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
applies to the principal component so holders of the principal at maturity receive the
inflation-adjusted principal value or the par amount, whichever is greater.
If, on January 15, 2011, the reference CPI is equal to 219.5, an owner of the principal
component will receive:
If, however, the reference CPI at maturity of the bond were somehow less than the base CPI,
resulting in an index ratio of less than 1.0, the inflation-adjusted principal will be less than
par and the investor will, accordingly, receive the $1,000,000 face value.
The principal component trades at a discount to par and for trade settlement purposes will
settle in the intervening period using the same methodology as above, substituting the
current reference CPI into the equation. So for example, on January 23, 2010 the reference
CPI was 216.28558. Therefore, if the January 2011 principal iSTRIP was priced at 99 for that
settle date then market value would be calculated as:
Interest component
The US Treasury faced a hurdle in the initial formation of the strips program, as each TIPS
issue having its own base CPI would have a different inflation accrual index. To make issues
fungible with each other, the Treasury had to create a two-step process: first, removing the
inflation indexation to allow for stripping and then re-adjusting the zero coupons for their
inflation accrual. The embedded deflation floor in the TIPS security stays with the principal
component, making the coupon component a true real rate security; the development of
the inflation derivative market allows buyers of coupon iSTRIPS to purchase inflation floors.
Hence, a buyer of coupon iSTRIPS can effectively create “P” if necessary.
The interest component (coupon) from a particular TIPS issue is transferred at an adjusted
value initially, which is established using the CPI reference value for its original issue (dated)
date. The adjusted value represents the reset of the inflation accrual to 100, with an inflation
adjustment made to an investor at maturity. In this way, coupons with the same maturity
from different TIPS are now fungible and the coupon strip would be inflation adjusted at the
same rate. All such components with the same maturity date have the same CUSIP number,
regardless of the underlying security from which the interest payments were stripped.
The US Treasury, in the Federal Register, sets the stripped interest component and its
adjusted payment valuation. The Treasury established that the adjusted valuation (AV)
calculation would be as follows:
15 March 2010 49
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
C = quoted coupon
P = $1,000,000 par amount
CPI = 174.04516 base CPI on issue (dated) date
AV = adjusted value
AV = ((C/2) *P) *(100/CPI))
or ((0.035/2) * 1000000) * (100/174.04516) = $10,054.86
Source: Barclays Capital
In this example, with a $1,000,000 notional stripped, 10.1 of $1,000 bonds are created.
Bundled with other issues, there is sufficient liquidity created to generate round lots of
bonds. Prior to maturity, a buyer/seller of a coupon would settle a trade as follows:
Using the example, assume that the January 2011 note is purchased with a settlement date
of January 23, 2010 and a reference CPI of 216.28558, if we assume the price is 99 the
coupon would settle at:
At maturity, the amount payable on a coupon strip is made via the following formula:
Following on our example for the principal strip, assume that in January 2011 the reference
CPI (at maturity) is 219.5 and thus, final payment would be $1,000, 000 * (219.5/100) =
$2,195,000.00.
US inflation derivatives
US zero-coupon CPI swaps have adopted an interpolated base index format, in the same
way as the French CPIx market. This more closely aligns the swaps market methodology
with the bond market, which also features an interpolated daily reference index. This serves
to smooth out the discontinuities in swap breakevens at month-end that occurred when the
market first found its feet using the HICPx-style non-interpolated format. The index used is
the CPI-U not seasonally adjusted index with a three-month lag, the same as that for TIPS.
The Bloomberg ticker for the index is CPURNSA <Index>. Barclays Capital’s indicative CPI-U
zero-coupon swap Bloomberg page is BCAP3.
While zero-coupon-style swaps are the most active structure traded on the inter-broker
market, y/y structures are most commonly demanded by the US retail sector. A primary
driver of US swap activity thus far has been hedging related to inflation-linked MTN deals,
15 March 2010 50
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
although issuance in this sector has been extremely limited since 2008. Typically, these
corporate deals pay y/y inflation on a monthly basis plus a fixed spread (with a floor usually
set at zero on the sum of inflation plus the fixed coupon). Paying the inflation uplift out
rather than accreting the principal as with TIPS is done primarily to avoid the phantom
income tax problem associated with TIPS. Options on CPI Swaps, TIPS and breakevens,
along with other non-linear inflation products, such as caps and floors, have traded in the
US, but only in limited fashion thus far.
Linking the cash TIPS and CPI swaps market is asset swaps. Trading in asset swaps in the
US has tended to occur on a tactical basis when valuations appear at the edge of a range.
The general lack of payers in the inflation swap market thus far in the US means that CPI
swaps breakevens and bond breakevens are more loosely connected than say in the euro
area. The gap between cash breakevens and CPI can be explained by funding costs of a
cash breakeven position including expected repo differentials and balance sheet costs.
CPI futures were introduced in the US in February 2004. However, the initial contract
specifications did not proved useful and it is no longer traded. While not likely soon, we
do see scope for reintroducing an inflation contract based on either monthly NSA CPI
settings or monthly contracts on the y/y format, similar to the HICPx future in Europe,
because this would be more useful for both risk management and relative value trading if
it were introduced.
The total rate of return swaps market has taken off notably, beginning in 2006. TRS
provides an alternative to cash as a way to gain long or short exposure to cash
instruments, particularly by investors looking to match index returns. Many investors use
inflation-linked TRS to gain beta exposure to the asset class while generating alpha
returns in some other product.
15 March 2010 51
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Euro area
Khrishnamoorthy Sooben France kick-started the European linker market with an issue linked to French CPI ex-
+44 (0) 20 777 37514 tobacco in 1998 and launched its first bonds linked to the euro HICPx in 2001. Greece,
khrishnamoorthy.sooben Italy and Germany have since followed, with Italy now the largest issuer in euro HICPx.
@barcap.com The development of an inflation-linked debt programme in Germany has been slow, but
is likely to gain momentum in the years to come. The depth and complexity of the
inflation derivatives market has increased substantially, with end users active in
inflation swaps, while structured note activity has encouraged the development of a
volatility market. Activity in individual euro country inflation indices remains limited,
with exposures mostly hedged via the broader euro HICPx market
In 1996, the Eurostat statistical agency was charged with creating “common statistical
standards for consumer price indices”. The headline all-items HICP Index, also known as the
MUICP or Monetary Union Index of Consumer Prices, is the main inflation reference for
monetary policy for the European Central Bank (ECB). The ECB has a mandate to maintain
price stability, which has been defined by the ECB as a level of MUICP inflation close to but
below 2%. MUICP inflation swaps traded before the launch of the first French euro HICPx-
linked bond, but after France had issued its first bond, HICPx became the most widely used
inflation swap base. As most Italian domestic inflation liabilities also exclude tobacco, the
Italian government used the same index for its first €i bond, confirming the benchmark
status of euro HICPx for both bonds and swaps.
15 March 2010 52
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
HICP indices, in common with most other CPI indices nowadays, are geometric chain-
weighted Laspeyres indices. While there is annual chain indexation at the start of each year
to reflect changes in consumption weights between and within countries, some euro
countries, particularly Germany, have detailed re-weightings only every five years. The
German resetting in February 2003 caused significant revisions to the inflation profile, but
all bonds and swap contracts are based on unrevised index values. Following the German
rebasing in February 2008, Eurostat decided to allow changes to impact the HICP series only
from January 2008. This introduced a notable structural break in the package holiday sub-
component of the series in 2008.
Final euro area inflation data is usually released around the 17th of the following month, but a
flash estimate of MUICP inflation is released around the end of the month in which data are
collected and individual countries release data in advance of the euro total, leaving only limited
uncertainty in the final release. Final January inflation data are released very late in February as
a result of extra calculations needed for annual re-weighting, which causes a complication for
the bond market. When an inflation reference value is unknown for settlement, the official
formula to calculate the index ratio is to extrapolate the last known y/y inflation rate to the
latest index value. This is a poor approximation for January m/m inflation as the seasonal
factor for this month is the most extreme negative of the year. As a result, the market rarely
trades on the official convention at the end of February, preferring to short settle or not trade
at all if the index ratio is unknown. The rebasing of the HICPx Index at the end of February
2006 (base year 2005 = 100) did not materially affect valuations on bonds or swaps, for which
the original reference HICP has been rescaled accordingly. Since February 2006, the index is
published to two decimal places rather than one.
There have been few major revisions to the composition of the euro HICPx in recent years.
There is relatively little standardisation of quality adjustment measurement at present, but
Eurostat has been focusing on gaining consistency. The use of hedonic pricing to adjust for
changes in quality is likely to become more widespread as a result, which over the long term
may produce a marginal downwards bias in the index. More important is the consideration
of housing, with owner-occupied housing currently excluded from HICP indices. Housing
rents make up 6% of the HICPx, whereas national accounts data suggest that around 16%
of all consumption spending is on housing (ie, actual plus imputed rentals). The ECB has
highlighted the need to include owner-occupied housing, for instance in its July 2005
monthly bulletin. This suggested that any inclusion would most likely be on a net
acquisitions basis (ie, attempting to strip out the price of land from house prices) as this is
an investment rather than consumption element. Eurostat, along with national statistics
agencies, has conducted various pilot studies on this with mixed results. Thus, while a
decision on the inclusion of owner-occupied housing was originally intended to be made by
15 March 2010 53
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
2006, no decision has been taken by the start of 2010. If agreement cannot be reached, it is
possible that a separate house price measure is produced instead. If owner-occupied
housing were included in the HICPx, over the long run it would be likely to bias the price
level upwards but it could also significantly increase medium-term inflation volatility.
While covering a notably smaller and more homogeneous area makes the French CPIx Index
more volatile than euro HICPx, this is mostly offset by having a lower weighting for fresh
food and energy. Although the level of French prices is only marginally above that of the
euro area average, France would become relatively more expensive as the euro area
expands eastwards. The index is usually released around the middle of the month, just
ahead of the euro area data. Until 2005, preliminary data were released early in the month
with the final series published after the euro data, but Eurostat encouraged INSEE, the
national statistics office in France, to publish earlier. Since March 2005, there has been a
single CPI released mid-month. The unrevised index is used for bonds and swaps. If the
series is rebased all reference calculations are adjusted accordingly. An unpublished early
estimate of French HICP is provided to Eurostat for the euro HICP flash estimate and, as
estimates for most other large euro countries are published around this time, inferences on
the French data may be drawn if the euro flash estimate surprises in either direction.
The calculation methods for the French CPI and HICP are relatively similar. Both use
geometric aggregation at the lowest strata sub-indices, and have the same methodology for
quality adjustments. There has been no clear, long-term bias between the two series. The
difference is that HICP takes into account expenses net of rebates, while CPI uses a gross
basis. This is particularly important for the healthcare component where rebates from the
state are substantial in France. This leads to a much higher weight for healthcare in the CPI
(10%), but also leads to inconsistencies between the two series when healthcare reform
affects the degree of public subsidy, for instance reducing the items on which they are
available, which causes a jump in the HICP healthcare series without affecting the CPI. While
in the long term it is likely that healthcare costs will increase notably higher than inflation,
this may well be offset if the trend for reduced state subsidies continues.
15 March 2010 54
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
4%
3%
2%
1%
0%
-1%
-2%
Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Jan 09
Source: INSEE, Eurostat
Price convergence assumptions among countries in the euro area are an important element
in projecting the evolution of a domestic index relative to the euro area. If euro area
economies are closely related, open and share the same currency, then we can assume that
domestic price levels will tend to converge. Ultimately, this implies that in the long run, price
levels will be very similar across countries of the euro area. However, by definition, inflation
rates will diverge during the convergence process among countries that initially have
significantly different price levels. For example, Spain currently has a lower price level
compared to the euro area average. If there are strong convergence forces, then prices in
Spain will tend to adjust upwards to catch up with the euro area average, thereby
mechanically pushing Spanish inflation relatively higher. The inverse would apply for a
country with a higher-than-average price level.
15 March 2010 55
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
that close trade relationships between a country and the rest of the euro area will tend to
encourage price convergence in the products that are effectively traded. Hence, if the
difference in the general price level is due primarily to the difference in price levels in the
services sector, convergence may not be as strong given that services are often not easily
traded among countries.
Eurostat data up to 2008 show that the price level convergence between the major euro area
countries has been relatively strong. The enlargement of the euro area will, however, have an
impact on euro HICP. First, the higher the weight of a new country in the euro area HICP
basket, determined by the share of its “household final monetary consumption expenditure”,
the higher its potential effect on euro HICP. Second, the further its price level is from the
current average, the higher will its inflation tend to diverge from other member countries at
the beginning. In the past, additions to the euro HICP have not introduced any strong bias in
the index. However, with the euro area expanding towards eastern European countries,
where price levels are lower, this will introduce an upward bias on euro HICP inflation.
130 BE DE ES FR
GR IE IT NL
120
110
100
90
80
70
60
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Source: Eurostat, Barclays Capital
France
France first announced its intention to issue inflation-linked bonds on 3 December 1997.
The legislation to enable the launch of the new asset class was passed on 3 July 1998. On 15
September 1998, the OATi 3% Jul 2009 was syndicated, with the bond frequently re-opened
by subsequent auctions. The market was widely consulted on the main characteristics of
the new bond, including the choice of inflation index to which it would be linked. It was
decided that the bonds would adopt the Canadian methodology that was fast becoming the
preferred global structure, but including a principal floor as the US had done. The timing of
the first issue just ahead of the start of the euro area was not a coincidence. This monetary
union was expected to intensify competition for financing in nominal bonds, and the hope
was that France would gain a first-mover advantage by being the first euro area country to
issue inflation-linked bonds.
The inflation index was agreed as INSEE’s official measure of French national CPI,
excluding tobacco. There was considerable debate ahead of the initial launch of French
15 March 2010 56
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
inflation-linked bonds as to whether to link the first issue to French inflation or to that of
the then forthcoming euro area. The arguments for the domestic index included the
likelihood that national inflation would be a better liability match for the government.
However, international appeal would clearly be broader for a euro index. In 1998, the final
decision almost certainly came down to practicalities. At the time, the disadvantages of
Eurostat’s European Harmonised Index of Consumer Prices for the EMU area were
material as it was a relatively new, untested index with no track record. Full index
coverage was not yet complete in some countries, which left an index in flux and an
associated fear of revision risk.
A second linker, the OATi 3.4% Jul 2029, was launched a year later in September 1999,
again linked to the French national CPI ex-tobacco. The same issuance route was followed,
with an initial syndication and occasional re-openings. Growth in the outstanding market
value of these two bonds was slow but steady. There was some disappointment that the
instruments did not seem to be capturing the imagination of investors in euro area
countries outside of France. In October 2001, France addressed this issue head on by
launching the OAT€i 3% Jul 2012 linked to euro HICPx. Again, this bond was launched via
syndication, but with its size boosted by some direct exchanges out of the OATi09. There
were some fears ahead of this issue that the launch of a second inflation-linked product
may harm the liquidity of existing OATi bonds, but in fact the move gave a new lease of life
to the sector as a whole. Not only did turnover in the new issue quickly grow, but interest in
the existing issues was heightened too.
France responded to an increase in interest and demand in the sector with a significant
increase in the pace of supply. It has issued new bonds each year while auctioning existing
issues nearly every month. The OAT€i Jul 2032 was syndicated in 2002, including some
exchanges out of the OATi29. The OATi Jul 2013 was the first issue to be launched via
auction in 2003. The Agence France Trésor (AFT) decided to revert to a syndication method
to launch its new OAT€i 2020 at the start of 2004, but then launched the OATi11 and
OAT€i15 via auction later in the year. The OATi17 was also launched via auction in
September 2005. In April 2006, the first BTAN linked to euro HICPx, the BTAN€i10, was
launched via auction. After consultation with primary dealers, the AFT recommended a T+3
settlement date for the BTAN€i (nominal BTANs settle T+1) to be consistent with other
inflation-linked bonds. For the launch of the OAT€i40 in March 2007, France switched back
to syndication, as for the OATi23 in February 2008. In January 2010, the OATi19 was issued
via auction.
The AFT has steadily increased linker issuance since the product was launched. Prior to
2009, it was committed to a minimum of 10% of its total bond issuance each year to be
in inflation-linked bonds, but with the possibility to issue significantly more if justified by
demand. €17bn was issued in 2005, 14% of gross bond supply, compared with €24bn in
2004 when demand was particularly strong. In 2006, €18bn was issued compared with
€17bn in 2007. 2008 saw €15bn but issuance dropped to €12bn in 2009, as the AFT
responded to the fall in demand for linkers. The strategy to reduce issuance from the end
of 2008 and over 2009 was combined with a pragmatic approach, which consisted of
tapping specific issues that were in demand. For instance, in October 2008, when fears
about deflation were intense and risks of poor action were significant, only OATis were
issued, given that the only demand was for French inflation due to Livret A hedging. In
2010, as in 2009, the aim of the AFT is for about 10% of total issuance to be in inflation-
linked bonds. Auctions normally occur on the third Thursday of every month, excluding
August and December. It is usual for one OATi and one OAT€i to be auctioned each
month except when a new bond is launched.
15 March 2010 57
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Figure 26: French linkers: historical performance and risk Figure 27: Return/risk French IL versus nominals and equities
Greece
Greece launched the GGB€i 2.9% Jul 2025 in March 2003 via syndication. The bond was
priced against an interpolated OAT€i real yield curve on 18 March, for €1.25bn. It was re-
opened via syndication twice in 2004 and once in 2005 and 2006. Greece launched the
GGB€i 2.3% July 2030 via syndication in March 2007 for €3.5bn, with pricing against the
OAT€i 3.15% 2032 on a real yield basis. It was tapped for €4bn in January 2008, again via
syndication. A few weeks before launching the GGB€i30, Greece made a private placement
of a 50y bond linked to euro HICPx. The liquidity of the Greek issues has generally been
lower than other European linkers, probably due to their small weight in the market. The
fact that they fell out of the main Barclays Capital inflation-linked indices at the end of
December 2009 because of ratings downgrades may affect their structural valuations
versus other issues and contribute to keeping poor liquidity conditions.
GGB€i bonds have exactly the same calculation conventions as French OAT€i bonds, with
inflation accrual linked to HICPx, and even the same coupon payment date, 25 July. Before it
entered the euro area, Greece issued bonds linked to domestic CPI in small size in 1997, also
in a Canadian format, but the last of these redeemed in 2007.
Figure 28: Greek linkers: historical performance and risk Figure 29: Return/risk Greek IL versus nominals and equities
5% 1
0% 0
-5%
-1
-10%
-15% -2
-20% -3
2004 2005 2006 2007 2008 2009 2004 2005 2006 2007 2008 2009
15 March 2010 58
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Italy
Italy announced its intention to issue its first inflation-linked BTP on 5 September 2003 and
syndicated a €7bn 5y bond within five days. The speed of the ground-breaking transaction
took many in the market by surprise, but the issue was quickly accepted, enabling a
syndicated re-opening in October to bring the bond to over €10bn. The BTP€i 1.65% Sep
2008 followed an almost identical model to French OAT€i bonds, except that it paid semi-
annual coupons like conventional BTP bonds. The bond was initially priced using an
interpolated spread to the nominal BTP curve, but a maturity matched conventional bond
was auctioned the week after the launch, enabling straightforward trading of the breakeven
inflation spread.
The choice of maturity for the first BTP€i was determined by heavy domestic retail demand
for inflation-linked notes. The Public Debt Division of the Italian Department of the Treasury
had noted that a considerable amount of swapped 5y MTN notes with inflation-linked
coupons had been sold in Italy and without a 5y point on the OAT€i curve it was relatively
difficult for issuers to hedge their inflation exposure. Italy hoped to capture both swap-
hedging demand and to appeal directly to individuals who had been buying the structured
notes. More than 220 investors bought the initial syndication, with the majority placed in
Italy. Much of the remainder went to the UK and US, a combination of derivative houses and
long established, international inflation-linked investors, with relatively little going to other
euro area countries. The first re-opening syndication in November 2003 redressed this
imbalance, with almost 40% being allocated to French investors. Further syndicated supply
of the issue in 2004 brought its face value up to €13.4bn. The majority of this bond quickly
became held versus inflation swaps, much of which remains locked away to maturity.
Having started issuance with a relatively opportunistic 5y maturity, the next new bond was
a 10y, the BTP€i 2.15% September 2014. The bond was initially syndicated in February
2004 for €5bn and subsequently built up via syndications and then auctions to a notional
size of €14.5bn, in line with the target size indicated at launch. The BTP€i 2.35% 2035 was
syndicated in October 2004 for €4bn and has subsequently been re-opened by syndication
and then auction. The BTP€i10 was syndicated in January 2005 as the second 5y issue and
has been built up by auctions subsequently. In June 2006, the 10y BTP€i 2.1% September
2017 was sold for €4bn via syndication, and tapped via auctions afterwards. The BTP€i
2012 was the first Italian linker to be launched via auction in March 2007, but the Tesoro
reverted to the syndication method a few months later in June to launch the BTP€i 2023, its
first 15y linker. Both bonds have been re-opened since via auction. In addition, Italy has
issued private placements of two ultra long-dated euro HICPx linkers maturing in
September 2057 and September 2062. These, together with Greece’s 50y linker private
placement, sparked two-way interest in ultra-long euro HICPx swaps. In May 2009, the 10y
BTP€i 2019 was launched via syndication but thereafter re-opened via auction. After almost
one and a half years, Italy came back to the new issue market in October 2009 with the
syndicated launch of the BTP€i 2041.
In 2005, Italy overtook France as the country with the largest stock of bonds linked to euro
HICPx. Like France, Italy also reduced its inflation-linked issuance significantly from the third
quarter of 2008 and over 2009, as a response to reduced demand, with the scheduled
auction of October 2008 even cancelled. Italy has indicated its intention to continue with
the monthly issuance of BTP€is. Auctions take place near the end of the month, the day
before nominal supply and settling on the last day of the month. Except for the 5y BTP€i12,
the launch of new bonds has been done via syndication. The Tesoro had previously
indicated it will launch 5y maturity bonds via auction, but its 2010 Guidelines for Public
Debt Management indicate that new securities will be placed via syndication. Effective from
15 March 2010 59
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
the start of 2010, BTP€i auction announcements carry a range for the total amount to be
issued across the securites involved. Previously, a maximum amount was announced for
each individual BTP€i auctioned.
As with a conventional BTP, a BTP€i pays its coupon every six months, but its yield is
quoted on an annual basis. Calculations work in exactly the same way, with inflation accrual
calculated on a daily interpolated basis between the inflation data from three and two
months previously. Italy chose the same index as France mainly for market convenience, as
it is the index most widely used in inflation swaps and MTN bonds as well as OAT€is,
although domestic Italian indexation has historically usually excluded tobacco, too.
Figure 30: Italian linkers: historical performance and risk Figure 31: Return/risk Italian IL versus nominals and equities
12% 3
10%
2
8%
1
6%
0
4%
-1 Equity
2%
-2 IL
0%
Nominal Bonds
-2% -3
2005 2006 2007 2008 2009 2005 2006 2007 2008 2009
Germany
The intention to issue euro HICPx-linked bonds was announced by German finance ministry
officials in November 2004, but it was not until March 2006 that the initial bond was launched.
The inaugural inflation-linked bond meant that all G7 countries were issuers of inflation-linked
bonds. The 10y Bund€i 1.5% April 2016 was issued via syndication with an initial size of €5.5bn,
including €0.5bn retained to boost liquidity in secondary market trading, and was priced against
the nominal Bund January 2016. The German Finanzagetur indicated its intention to tap the bond
up to three times to a volume of €10-15bn. The bond was re-opened in September 2006 via
syndication for €3.5bn, including €500mn retained. Germany switched to an auction procedure
for the second tap of the Bund€i16 in April 2007 for €2bn. The second euro HICPx German
linker, the OBL€i 2.25% April 2013, was issued in October 2007 for €4bn. The launch was
conducted via auction. The third German linker came in June 2009. The Bund€i 2020 was the
first euro linker launched after the episode of extreme, deflation-led fall in breakevens during the
second half of 2008. This reaffirmed the commitment of Germany to the inflation market, given
that the development of the German real €i curve has been slow compared to what was broadly
expected when the programme was launched in 2006. The Finanzagetur announced in its 2010
issuance outlook that it intends to issue €3-4bn inflation-linked bonds quarterly.
The focus during the launch of the first German linker was on seasonality pricing. Before the
launch of the Bund €i16, bonds linked to the euro HICPx had maturities of 25 July (French
and Greek linkers) and 15 September (Italian linkers). The fact that the Bund €i16 would
redeem on 15 April 2016 meant that it would accrue less positive inflation seasonality
compared to an equivalent OAT€i or BTP€i. Although the bond was being priced against the
15 March 2010 60
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
nominal Bund January 2016, the market was also focused on estimating fair value versus
the OAT€i15, and estimating the difference in the seasonality component of the two bonds
was key in determining the fair value of the Bund€i16. From that point of view, the German
linkers are useful. By trading on a different seasonality point, they have helped to refine the
seasonality pattern that is perceived by the market.
Figure 32: Euro linkers – historical performance and risk Figure 33: Return/risk Euro IL versus nominals and equities
-4% -2
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
For French government issues, each linker has one principal component, identified by its
own ISIN. Conventions for the principal component are exactly the same as for the
underlying issue except that there is a floor on the principal. Hence an investor receives at
least the face value of the position at maturity. If inflation has occurred since the underlying
bond was issued, the investor receives the face value multiplied by the index ratio, ie, the
reference inflation index at maturity divided by the initial reference inflation index value for
the underlying bond. While the value of the par floor option on its own is usually very small,
the guarantee of nominal principal repayment may be worth notably more for investors
who are unable to buy bonds without floors.
The coupon-stripping process for BTP€is, initiated as from 2008, holds an additional
complexity compared to French bonds. Investors can actually request that the principal be
split into a nominal component and a floored inflation uplift component. This introduces a
major innovation to the principle of iSTRIPS, given that this three-component stripping
model does not exist in any other inflation-linked bond market.
15 March 2010 61
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Inflation strips provide increased flexibility to hedge inflation-linked liabilities. Demand may
come from insurance and pension sectors in countries with inflation-linked liabilities but with
difficultly accessing inflation swaps, for example. For accounting purposes, inflation-linked
strips can be more favourable than inflation swaps as they are considered unleveraged
inflation-linked bonds on a held-to-maturity basis, with no embedded derivatives under an IAS
accounting regime. Co-incidentally, euro countries where pension funds do not use many
swaps in general and inflation swaps in particular also tend to have some of the longest
inflation liabilities. Spain, Finland and Ireland, for example, have long domestic inflation
exposure within pension funds. Ultra-long iSTRIPS could therefore be of interest for pension
funds in such countries, but activity in stripped euro linkers has never taken off.
The pressure for life insurers and pension funds to address liabilities with the introduction of
IAS19 has been similar to that in the UK, but far fewer liabilities are explicitly inflation-linked
in the euro area and of these, most are linked to indices very different from euro HICPx.
Pension reform in Europe may crystallise more demand to hedge long-dated, inflation-
linked liabilities. An example is the ‘TFR’ (trattamento di fine rapporto) in Italy, which refers
to money held back by firms from Italian employees’ salaries and paid back to them as
severance pay when they retire. The amount retained is 6.91% of gross earnings and under
the TFR the funds collected accrete in value annually at 1.5% plus 75% of the change in the
level of the Italian FOI (workers) CPI inflation index ex tobacco. Since 1 July 2007, all
companies with 50 or more employees are required to transfer further funds collected into
pension funds or to the control of the national government pensions system, INPS, although
cash collected prior to 2006 remains with companies. Firms have been required to provide
collective pensions options to employees, or new TFR money will be transferred by default
to INPS where it will accrue at the normal TFR rate. So far, the take-up of options that have
an inflation linkage has been limited, with a tendency for funds to be transferred to INPS,
but the size of such exposures is likely to grow over time. While previously, TFR funds were
released to employees when they changed employers, now they will be released only at
retirement, implying less turnover in the liabilities and hence more incentive to hedge.
The development of the FRCPIx inflation-linked bond and swap markets has been largely
driven by the decision to partially link the remuneration rate on Livret A savings accounts to
the FRCPIx inflation rate. The decision, taken in 2004, was meant to depoliticise the rate-
setting decision. The original formula determined the rate as half the y/y FRCPIx rate plus
half the 3mth Euribor rate plus 25bp, rounded to the nearest 25bp, and was used in the
twice-yearly revision of the rate. The Livret A (called Livret Beu when distributed by the
Credit Mutuel network) rate is used to determine the remuneration rate on various other
savings accounts and in 2004, around €270bn of instant access account deposits became,
de facto, linked to French inflation. Funds collected on Livret A and Livret Bleu savings
15 March 2010 62
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
accounts have traditionally been centralised at the DFE (Direction des Fonds d’Epargne), an
agency administered by the state-owned financial institution, CDC, which finances social
housing schemes at lending rates that depend on the Livret A rate.
From an ALM perspective, there is therefore an automatic hedge for funds centralised at the
DFE and used to finance social housing. However, hedging of the exposure to French CPIx
inflation is needed from commercial banks which distribute savings accounts linked to the
Livret A rate, either through OATi bonds or FRCPIx swaps. The reform of the Livret A
undertaken in 2008 and effective from the start of 2009 had important implications with
regards to the latter. From 1 January 2009, the distribution rights for the Livret A were
extended to the whole banking network in France (as opposed to only La Banque Postale,
the Crédit Mutuel and the Caisse d’Epargne in the past). This led to a surge in total Livret A
and Bleu outstandings by more than €17bn in January 2009. Probably more relevant is the
fact that the centralisation rules with the DFE changed and under the new regulation,
commercial banks would be allowed to keep Livret A funds on their balance sheets,
although under some conditions. The expected and realised increase in Livret A
outstandings therefore implied a substantial increase in hedging demand and the impact on
the French inflation market was obvious.
Other than the Greek €i for domestic investors, the uplift of euro inflation-linked bonds is
generally taxable. This is one of the main reasons for the existence of the structured
inflation market, on which taxation is paid only on coupons as they are paid. OATis and
OAT€i bonds are taxed in very much the same way as other French government bonds, ie,
the inflation accrual is taxable for domestics while there is no withholding tax payable for
international investors. Retail investors can pay all withholding tax at maturity or sale.
Institutional investors pay tax both on interest received and annually on inflation as it
accrues. BTP€i bonds follow the same tax rules as conventional BTPs. This means that
domestic entities are taxed on inflation uplift as well as real returns. International investors
are exempted from paying withholding tax as long as they are within countries that Italy
does not define as tax havens, and they send in the necessary initial documents that are on
the Treasury website. At the time of writing, countries excluded from the “white list” of tax-
exempt countries include Switzerland, as well as some offshore tax havens.
15 March 2010 63
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
has also been considerable issuance of structured notes whose cash flows are linked to
inflation. The accreting style issuance has been directed at similar institutional investors
to those buying government linkers and has generally not involved accompanying
derivative transactions. Most structured note issuance has been to individual investors,
particularly in Italy, and the inflation exposure of these notes has largely been hedged
using inflation swaps.
Government-style bonds
Agencies, quasi-agencies and regions have mostly issued inflation-linked bonds in a
government-style format. The largest non-government issuer has been the Caisse
d’Amortissement de la Dette Sociale (CADES). This sovereign agency was created in 1996
as a vehicle to consolidate and service the debts of the French social security funds. Its
revenue comes from a ring-fenced, fixed tax on income called CRDS, making it a natural
issuer/payer of French inflation. As of mid January 2010, it had €10bn face value of bonds
linked to French CPIx, €12bn by market value, with four CADESi benchmark issues with
maturities ranging from July 2011 to 2019. These issues have been built up via multiple
syndications and in practice have traded very much like OATis, albeit with lower liquidity,
helping to define the curve when there were relatively few government issues. CADES
stated that it aims to issue €0.5bn in CADESi issues in 2010.
Other notable issuance has come from Réseau Ferré de France (RFF), the owner of French
railway infrastructure, whose 2023 HICPx-linked bond reached €2bn face value, having
initially been syndicated for €800mn in February 2003. Caisse Nationale des Autoroutes
(CNA), which grants loans to toll road companies, issued a €600mn 2016 French CPIx-
linked bond in 2001. The Italian agency, Infrastrutture (ISPA), which has inflation-linked
revenues from some projects such as high-speed railways which it funds, issued a €750mn
2019 bond in February 2004. This issue was the first benchmark inflation-accreting bond to
be linked to Italian inflation, specifically workers inflation (FOI) excluding tobacco, although
there has been much more structured supply linked to this base. The first major true
corporate inflation-accreting bond came from Veolia Environnement in June 2005, a euro
HICPx-linked €600mn 2015 bond. As an owner of a range of utilities in France and across
Europe, Veolia ought to be well suited to using the inflation-linked market as part of its
funding strategy. Terna S.p.A. became the first Italian listed company to issue a bond linked
to the Italian FOI Index ex tobacco in October 2007. The €500mn issue matures in
September 2023 and was priced against the BTP€i September 2023, which carries the same
maturity date. Given its revenues derived from regulated activities, Terna is a natural payer
of inflation. Toll road operators may well become significant issuers/payers of inflation
across Europe as the market develops, particularly those with explicit inflation linkages for
their tolls. In August 2008, France Telecom issued a 10y bond linked to euro HICPx for an
initial €350mn amount.
15 March 2010 64
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
volumes fell back to around €3.5bn in 2007. The widely dispersed nature of this issuance
means that while there are several issues of over €500mn, liquidity is very limited. In 2003,
most issuance was 5y, but as yields fell and the type of demand changed, issuance has
moved longer, with 10-15y supply as common as 5y in recent years.
Most issuance in 2003 was of bonds paying an annual coupon at the rate of inflation plus a
fixed percentage but with a fixed principal. Coupons were usually floored at the fixed rate,
although in 2004 higher floors became more common. Coupons were often backward-
looking, eg, paying inflation from the previous year and high fixed coupons early in the life of
the bond were commonly offered as enticements. As distribution of this type of bond
became more widespread, there was increased interest from corporates as well as
individuals, but fees were such that most institutional investors were deterred. With the slight
revival of issuance in 2006, products with coupon payments linked to inflation (mostly euro
HICPx) with leverage became more popular. These notes often carried a fixed attractive
coupon at the beginning and a cap and/or floor on the subsequent floating payments. 2006
also saw several structures with coupons linked to the differential between euro and French
ex-tobacco inflation. The issuance of such products thrived on what was perceived as an
anomaly on the forward breakeven differentials in the swap market. Indeed, forward French
inflation breakevens were higher than on the euro HICPx curve because of Livret A-related
demand. These notes proved popular as French inflation was expected to be lower than
European inflation over the medium term. 2007 saw the appearance of inflation range
accruals. Pay-offs here are dependent upon the length of time that inflation remains within a
specified range. The development of these products marked a stepping stone as they helped
increase liquidity in the market for inflation caps and floors, but with realised inflation
breaking above the top of the ranges specified by the end of 2007, holders of the notes
suffered poor performance and demand for them waned. In 2008, notes paying a coupon
linked to a multiple of the y/y inflation rate were widespread, with activity driven by demand
from retail investors given the high inflation environment. The structuring of attractive pay-
offs was also relatively easy as result of expensive credit/funding for financials.
Most structured inflation notes have been issued by financials, although opportunistic
swapped issuers including the EIB and KfW have also been involved. A rise in real yields
would increase issuance to individuals although the hefty levels of 2003-04 or 2008 will
likely be seen only if there is once again an inflation scare. There has been a clear trend
towards more institutionally-focused structured issuance, which is likely to be less sensitive
to rate levels. The fees involved in this kind note are often lower due to smaller distribution
fees, providing access to investors who are unable to take direct advantage of the
development of the inflation swap market. The area in which there is still ample room for
further development is that of domestic inflation bases. Issuance has been relatively
significant from time to time on some bases, but the market is still far from having reached
a self-sustained dynamism. For example, in 2006, there was significant issuance in
structures linked to Spanish inflation, but supply dried up in 2007, before picking up
strongly during the first half of 2008, when most of the structured notes issued in Spain
were linked to the domestic index. If structural demand develops for individual domestic
inflation, eg, on the back of evolutions in pension industries, then this could force the
development of two-way markets.
15 March 2010 65
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Mercantile Exchange (CME) contract was first listed on 19 September 2005, while the Eurex
HICPx future started trading on 21 January 2008.
The CME HICPx future contract is tradable on the CME electronic Globex platform between
8:00am and 4:00pm London time. The price of each contract at expiry is 100 minus the y/y
change in HICPx. Each monthly contract refers to the inflation release from the previous
month, which will almost always be released during the contract month. Thus, the value of
the January 2008 contract expires based on the y/y inflation rate of December 2007 HICPx.
The contract trades until 4pm London time on the day before the inflation data is released.
Prices for the contract are available directly from the CME and can also be seen on
Bloomberg, where the contract code is AAxx <index>. Thus, to see the price of the whole
futures strip, on Bloomberg, the code is AAA <index> CT, while the Jan ‘07 future is AAF7
<index>. The Reuters code is <0#EHI:>.
Reference HICP 100.00 – annual inflation rate in the 12-month period preceding the contract
Futures Index month based on the Eurozone Harmonised Index of Consumer Prices
excluding tobacco (“HICP”) published by Eurostat.
Contract months Twelve consecutive calendar months.
Trading venue Available for trading on CME “Globex” weekdays from 8:00am to 4:00pm
and hours (London time).
Minimum price 0.01 Index points or €100. (Leaving the notional value of the contract
fluctuation approximately €1,000,000.)
Last trading day 4.00pm (London time) on the business day preceding the scheduled day the
HICP announcement is made in the contract month.
Final settlement By cash settlement on the day the HICP announcement is made. The final
price settlement price shall be calculated as 100 less the annual % change in HICP
over past 12 months, rounded to four decimal places, or 100 –
[100*(HICPt/HICPt-12)-1].
Note that a price of over 100.0 suggests deflation during the 12-month period.
Source: CME
The format of the Eurex HICPx future contract is almost identical to the CME. Prices are
quoted as 100 minus the y/y change in euro HICPx during the previous month. The first
difference is that Eurex quotes a strip of 20 monthly contracts compared with only 12 for
the CME alternative. But the main innovation lies in the twice-daily auction mechanism, in
an aim to to encourage liquidity. Every day between 9.45am and 10am and between
4.45pm and 5pm central European time, the designated market makers were expected to
quote at least 12 of the 20 listed contract months. The price at which orders are matched is
determined at the end of the twice-daily auction through a netting process. The netting is
designed to determine the price that results in the highest executable volume. Prices can be
monitored on Bloomberg, where the code for the whole strip is HICA <index> CT. The
Reuters code is <0#HICP:>
On both the CME and Eurex contracts, there has been almost no activity since mid 2008,
but previously, the front CME HICP future was a good predictor of inflation compared with
most economists, thereby highlighting its use to ascertain fair economic value. The Eurex
future had the potential to be more successful: the fact that a strip of 20 monthly contracts
is quoted should have encouraged its use by dealers as it enables the hedging of fixing risks
even beyond the one-year point. Fixing risks are indeed very high on short-dated inflation
swaps, thereby hampering activity and resulting in wide bid/offer spreads.
15 March 2010 66
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Euro HICPx
Euro HICPx swap broker volumes began to pick up substantially from late-2002, driven by a
rise in the need to hedge retail products and structured MTNs. In 2003, it accelerated further,
aided by the issuance of BTP€i08, the most suitable hedge for most of these exposures.
Typical monthly broker volumes moved from €500mn in mid-2002, to €5bn by mid-2004, a
level that was maintained in 2005 even though structured issuance fell notably. Liquidity in
the market has been increasing since, with typical monthly broker volumes rising to €15bn in
2007. Total volume traded surged in 2008 in the interbank market but mainly because of the
Lehman collapse. Prior to the financial turmoil starting in 2007, direct interest in trading
inflation swaps, from proprietary desks and hedge funds for instance, meant volumes did not
suffer even when issuance hedging flow slowed down, while liability hedging rose steadily.
However, thereafter, most speculative investors significantly reduced their activity in the
inflation swaps market, which coincided with a period of low liability-hedging demand as
Dutch pension fund solvency ratios dropped. Government linker asset swaps trading also
represents a significant proportion of volumes in the euro area inflation swap markets. The
relatively high level of activity in asset swaps has generally helped align swap and bond
breakevens more closely than has been seen in other markets. The private placements of
ultra long-dated linkers by Italy and Greece in 2007 has raised the prospect of the euro
inflation curve stretching out to the 50y maturity as in the UK, but the depth of demand at
such extreme liabilities remains relatively limited. During the financial turmoil from 2007, the
fact that many speculative players or natural investors in linker asset swaps were no longer
active led to a wide distortion between swap and bond valuations in 2009. Ultimately, the
cheapness of linker asset swaps attracted demand, mainly from bank treasuries, but the
investor base has probably shrunk relative to before the crisis
The benchmark format for euro HICPx quotes is the zero-coupon structure, with a standard
lag of three months, meaning that the base inflation index for the swap is the value of the
HICPx three months before settlement. Towards the end of the month, swaps on the next
base month also begin to trade. There will be a discontinuity in the quoted “breakeven” at
the time of the roll from one month to the next, reflecting typical seasonality between
months. Barclays Capital displays live prices for the currently trading base month on
Bloomberg page BISW1. The most commonly-traded maturities are the 5y and 10y, but
annual maturities above 2y trade regularly out to 30y and sometimes longer. Activity at the
front end is more limited, but the inflation futures offer transparency in the 1y swap point,
as well as defining a tradable short-end curve.
Activity in inflation caps and floors on euro HICPx has been boosted in 2007 following the
issuance of inflation-range accrual notes. Given that these notes rely on inflation remaining
within a specified range, this means that dealers selling them were actually long inflation
volatility and they have tried to offset their risks by selling inflation caps and floors in the
market. The increased trading activity on inflation options has thus provided a clearer
picture of the implied inflation volatility. Hefty issuance in leveraged notes over 2008 also
encouraged hedging activity in the cap/floor market, which led to a significant rise in
implied volatility. The presence of embedded floors in previously issued notes and increased
volatility in realised inflation led to a sharp repricing of the cap/floor vol surface in 2008-09
(we cover this in more detail in the ‘Inflation volatility and deflation floors’ section later in
this publication). The link between the vol and the swaps market became tight towards the
end of 2008 when increasing deflation fears pushed short-dated swaps lower, leading to a
downward spiral of valuations at the short end of the curve as dealers sought to delta-
15 March 2010 67
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
hedge their short floor positions. Activity in breakeven and real yield swaptions is muted,
with no broker market, but there was an increase in trades during 2006 and 2007 when the
underlying swap markets became more liquid.
French CPIx
The French CPIx market is slightly less transparent than that in euro HICPx, and there is
much less structured inflation issuance to be hedged. The default format for French CPIx
zero-coupon swaps is also different, as it is based on the same interpolated daily reference
value as used for OATis, rather than the monthly format used in other euro area markets.
This avoids the discontinuity of the HICPx method and aligns swap methodology more
closely with bond methodology. On the other hand, it discourages short-term tactical
trading within the same monthly inflation base that is seen in HICPx and makes it more
difficult to compare market movements within a given month, as it can impose a drift on
the level of breakevens.
The French CPIx market is the oldest major euro area inflation market, first trading in 1998,
just ahead of the launch of the first OATi and before the HICP market that began almost as
soon as the euro currency was created in 1999. Short-dated issuance by French agency
CADES and the earlier development of the OATi bond market allowed the French CPI swaps
to initially have better liquidity than HICPx. Some domestic real money investors started
using swaps to match cash flows as the market developed. However, it was not until trading
of the basis against euro HICPx began, with broker screens becoming readily visible, that
the market gained any depth. However, compared with the size of the underlying bond
market, outright paying flows have often been more common than in euro HICPx, while
liability hedging is significantly broader than just Livret A-related flows.
The decision to link the Livret A French public sector savings rate to inflation from August
2004 greatly heightened activity levels in French CPIx. With banks being restricted on the
amount of bonds they could hold versus their liabilities, the flow became more and more
skewed towards using derivatives as a solution. A variety of swap types have been used to
hedge liabilities, including zero coupon across the curve, tailored maturity inflation swaps
matching the inflation element of reset on the Livret A rate and Livret A swaps covering the
nominal as well as the inflation element of the reset. In aggregate, there has been a
tendency for hedging activity to fall when real yields fall and also to move longer on the
curve, with relatively little apparent sensitivity to breakeven levels. Bank hedging demand
has at times led to more significant deviations between OATi asset swaps and those of
nominal OATs than in the €i market. There is a significant percentage of OATis that are held
in asset swap form until maturity as a result of the Livret A hedging pressure, but there is
also active two-way asset swap flow.
In 2006, the richness of French CPI swaps triggered substantial issuance in structured notes
paying euro versus French inflation, usually with a leverage factor. These notes were
actually hedged in the corresponding swap markets, correcting the squeeze created by
Livret A-related hedging. FRCPIx again richened significanty versus euro HICPx swaps
toward the end of 2008/beginning of 2009, given the extension of Livret A distribution
rights to the whole banking network in France. However, the fact that the formula was
changed in 2008 and then over-ruled most of the time in 2008 and 2009 implies that
French inflation is no longer a good hedge for Livret A-related liabilities, at least from a pure
formulaic perspective. This should help realign structural valuations between the FRCPIx
and euro HICPx swap markets to more fair levels.
15 March 2010 68
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
German CPI
Any indexation to German CPI was illegal prior to 2003, but the German government’s
intention to issue inflation-linked bonds, even though this was euro HICPx, spurred a
market to develop. While there are very few explicit inflation liabilities in Germany, the
higher real yield that a German inflation basis offers compared with one linked to euro
inflation may be of interest to retail and corporate investors. On the other side of the
market, the tendency towards utility privatisation may well produce paying flows, while, as
the market develops, banks are likely to be happy to offer German inflation versus euro
HICPx across the curve. Until now, supply in German inflation has come primarily from
property securitisation and rental leases. Banks are willing to pay German inflation through
swaps because of its weight in the euro HICPx, such that the basis risk involved is less of an
issue than for other domestic inflation indices.
Pension liabilities in Germany are traditionally backed by company assets, which means that
implicit or even explicit inflation risks tend to be ignored. However, a trend towards funded
schemes in Germany implies that embedded domestic inflation risks in liabilities are likely to
be addressed on a larger scale. However, the extent to which this will effectively create
substantial demand for German CPI is debateable. First, liabilities are not solely to the
national headline CPI but can be based on various other indices. In the latter case, where the
basis risk of the relevant index versus German CPI is high, hedging needs may as well be
met using the euro HICPx swaps market. Furthermore, even in the case where the relevant
index is the German CPI, pension funds may prefer to hedge euro HICPx swaps, given that
the basis risk versus European inflation is perceived to be low.
15 March 2010 69
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
from corporates to hedge TFR exposure, which is 1.5% plus 75% of FOI ex tobacco
inflation. The TFR reform, effective since July 2007, was expected to prompt an increase in
demand for Italian inflation. Companies with 50 employees or more are now required to
provide collective pension options, including options offering a nominal guarantee but with
a return target consistent with the TFR. Insurance companies managing these pension
liabilities should naturally be driven to receive Italian inflation. The pension liabilities linked
to inflation should also grow more steadily than in the past given that money held back
under the TFR will now be available to employees only at retirement. However, no marked
pick-up in demand for the FOI ex-tobacco inflation has been observed, with a tendency for
funds to be transferred to the national government pensions system.
There has been a very strong convergence in the Italian price level towards the euro
average. FOIx swaps have generally traded at a positive spread versus euro, but the spread
collapsed in 2008, probably due to some restructuring of some positions linked to Italian
inflation, and traded below in 2009 as inflation fell. Economic slack in Italy and loss of
competitiveness implies that the differential between the Italian and euro inflation is likely to
shrink, but the effect on FOIx swaps could be offset if TFR-related demand picks up.
Spanish CPI
The market for Spanish CPI developed in 2004 with some limited swapped structured note
issuance. Structured note issuance, however, rose strongly in 2006. These notes typically
had 10y and 15y maturities, and the hedging of these triggered a strong demand and
richening in Spanish CPI swaps as the street was left short. The rise in demand coincided
with a period when actual inflation was high on the back of the real estate boom, thereby
accentuating the richening versus euro HICPx swaps. Demand for Spanish CPI is mainly
from the insurance and retails sectors, while natural supply comes from residential property
leases and property securitisation. Spread levels to euro HICP swaps decreased in 2007,
from the very rich levels observed in 2006, helped in part by speculative selling from some
who viewed the spreads as unsustainable and a narrowing of realised inflation spreads.
The first half of 2008 saw heavy activity in Spanish CPI swaps on the back of structured
note issuance. The interesting feature of structured note activity in Spain, as opposed to
other euro area countries, during 2008 was that the underlying was commonly the
domestic index. This led to a sharp relative richening in Spanish CPI swaps, while the long
end lagged versus euro probably due to infrastructure-related paying flows. The general
price level in Spain is lower than the euro area and there had been convergence until 2008.
The volatility of Spanish CPI is notably greater than euro HICP given a heavy weight for the
food and fuel components. Inflation in Spain has been notably lower than the euro average
over the 2008-09 crisis, and structured notes linked to Spanish inflation dried out in 2009.
Spreads have been volatile versus euro HICPx swaps, but tightened significantly towards the
end of 2009, given the relative outlook in inflation.
Dutch CPI
The largest potential for domestic inflation protection demand, and especially in long end, is
undoubtedly in the Netherlands. The size of Dutch defined benefit pension schemes means
that demand for domestic inflation ought to be considerable. Indexation is usually to either
sector-specific or general wage inflation rather than Dutch CPI and, in the absence of large
natural inflation payers, most pension funds have accepted that they cannot match their
liabilities. Some funds are willing to pay a significant premium to source specific inflation
though, inducing payers mainly from the infrastructure and property firms. This is a market of
one-way flow, with all inflation receiving by pension funds locked to maturity.
15 March 2010 70
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
A move to a “real” framework would boost hedging demand but the potential for this to
create a large domestic inflation market is uncertain. Given the liquidity of the euro HICPx
market and the absence of any other suitable alternative, pension funds have been used to
taking the basis risk created by long exposures to euro inflation. The FTK regulatory
framework furthermore has made it easy to receive euro area inflation versus liabilities linked
to domestic inflation, in contrast to the IAS accounting framework, which makes pension
funds in other countries reluctant to hedge domestic risk with euro HICPx. Additionally, while
the short-term basis risk between Dutch CPI and euro HICPx is significant, so too is the basis
risk between wages and Dutch CPI. Also, the absolute price level in the Netherlands has
remained close to the euro area since the start of monetary union, suggesting that euro HICPx
long term is a relatively effective hedge of Dutch inflationary risk. We highlight though that a
“real” framework would likely encourage the Netherlands to issue in the inflation-linked
market. This could be in Dutch inflation; the Frijns Commission’s report actually highlighted
the market for Dutch inflation as a missing one. Such issuance would significantly increase the
scope for the development of a domestic inflation swaps market.
Irish CPI
Ireland has a longer history of inflation paying than anywhere else in the euro area. Supply
comes mainly by housing associations and Public Private Partnership (PPP) deals, which are
usually swapped. Further inflation paying is likely to come from infrastructure upgrading
projects. Pension fund inflation liabilities are more explicit than in most other euro
countries, with a legal minimum of an LPI format, floored at 0% each year and capped at
4%. This, coupled with the longer history of the market, often results in relatively complex
structures. The size of the market is such that it is unlikely to affect valuations elsewhere,
but it may encourage innovation in other markets, particularly given the trend for pension
funds of multi-national companies to base themselves in Ireland. Given the demographic
structure in Ireland, pension fund demand for inflation tends to be in long maturities. Supply
from PPP was much less than expected in 2007, and this coincided with a wide divergence
15 March 2010 71
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
between Irish and euro inflation on the back of mortgage interest costs. The mortgage
component is a function of both interest rates and long-term house price growth, which has
been substantial since the mid-1990s. The ECB rate-cutting cycle therefore pushed Irish CPI
inflation much more negative than European inflation, and by the end of 2009 the y/y
differential was 6.1%.
Irish inflation is often perceived as being expensive due to a lack of supply. Despite recent
sharp price falls, the price level in Ireland is still significantly above the euro area’s
average, suggesting that Irish inflation should continue to fall relative to the euro HICPx in
the coming years. However, the fact that Ireland’s major trade partners are countries
outside the euro area means that price level convergence may not be very strong.
Consequently, from this point of view, Irish CPI inflation may continue to trade at rich
levels versus euro HICPx inflation, especially if the quantity of new supply continues to
disappoint versus expectations.
15 March 2010 72
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
UK
Chris Bettiss The UK Treasury has been issuing gilts whose principal value is index-linked to the
+44 (0) 20 7773 0836 Retail Prices Index (RPI) since 1981. By the end of 2009, linkers made up 24% of UK
chris.bettiss@barcap.com government bond debt by market value, over £212bn. Pension fund liabilities are linked
to RPI either directly or via Limited Price Indexation (LPI); hence, pension funds still own
the largest proportion of the gilt linker market, though life insurers hold almost as many
as they take on increasing pension exposures. Both use the RPI swap market to hedge
their inflation exposures, with the corporate linker market relatively moribund despite
about £37bn in non government RPI-linked bonds outstanding at the end of 2009.
In contrast to most consumer price indices collected internationally – including the UK CPI –
the RPI is constructed with arithmetic rather than geometric aggregation. As this
aggregation is based on the average of relative prices rather than a ratio of averages, it
produces an upward bias compared with a geometric aggregation. This statistical bias,
often referred to as the formula effect, has been worth about 0.5% in recent years (ranging
from 0.41-0.64% per annum throughout 2001-09). This was the basis on which the Bank of
England’s inflation target was adjusted from 2.5% RPIX to 2.0% CPI, with tolerance bands
remaining at 1% either side of the target.
Figure 35: Breakdown of RPI by major category Figure 36: UK RPI, RPIX and CPI inflation
15 March 2010 73
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Even though monetary policy is now focused on CPI inflation, the UK inflation-linked market
will remain almost exclusively linked to the RPI. This is because almost all inflation-linked
liabilities are based on the RPI. The majority of pension fund liabilities accrue on a limited
price indexation basis. This means that the liabilities increase each year by the rate of RPI
inflation capped at a certain level, usually 5% prior to April 2005, with an implicit floor of 0.
Apart from aggregation methodology, the main difference between UK RPI and CPI is in the
coverage, in particular the exclusion of owner-occupied housing costs. Currently 17% of the
weight of the RPI is linked to house price elements that are not included in the CPI. This is
mainly split between a depreciation component, based on a smoothed house price index
and mortgage interest payments, which move with the long-term average value of house
prices, as well as standard variable mortgage rates. The mortgage rate series was changed
from the standard variable rate to the average effective rate from the February 2010 RPI
data. Council tax makes up a further 4% of the RPI basket while not being included in CPI.
The other significant coverage difference is car prices, with RPI based on only used car
prices but the CPI based on new and used car prices. While the data comprising most
remaining series are the same for RPI and CPI, weights can vary significantly. RPI weights
are calculated based on a variety of sources, the most important of which is the expenditure
and food survey, whereas CPI weights are based on the household final consumption
expenditure component of GDP. The broader coverage of the latter measure means, for
instance, a higher weighting for airfares in CPI, while the use of imputed values for CPI
means a heavier weighting for financial services.
Ahead of the change in the monetary policy inflation target in 2003, it was widely estimated
(including by the Bank of England) that the long-term bias between RPI and CPI inflation
would be about 75bp. This was based mainly on the formula effect of approximately 50bp,
plus the expectation that in the long run housing costs would grow in line with wage
growth. In the following six years, the difference between RPI and CPI has averaged 0.5%,
albeit with a wide range of -3.5% to 2.3%, mostly due to house price components. This was
despite pressures in other coverage and weights tending to offset the formula effect, such
as sharply falling used, but not new, car prices. Trying to tie down a long-term steady state
for this basis is fraught with difficulty, particularly as components can be changed over
time. The standard deviation of the basis since 1997 has been 1.3%, and in periods of
declining house prices and Bank of England rate cuts, RPI inflation can be significantly
below that of CPI. Hence, in 2008 RPI inflation turned significantly negative even though CPI
never fell below 1%.
The switch to an average effective mortgage rate series from the Bank Rate sensitive standard
variable rate series implies a less volatile mortgage interest payment series. In the long run, the
change in methodology should smooth but not bias the RPI-CPI spread, but because it is being
introduced when rates are at a historic low it is a significant deflationary impetus through the
period of rate normalisation. Barclays Capital economists estimate a long-term RPI-CPI spread
of about 0.9%, but this could be reduced if owner-occupied housing costs were included in
CPI, which remains one of Eurostat’s long-term aims.
15 March 2010 74
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
The creation of a linker market was formally recommended by the “Committee to Review
the Functioning of Financial Institutions (1977-80)” (known as the Wilson Committee, after
its Chair Sir Harold Wilson). However, indexation of debt was not a new idea in the UK – the
UK Government’s National Savings department had been issuing inflation-linked savings
certificates for retail investors since 1975, while Keynes recommended the move as early as
1924. In the ten years prior to the launch of the first linker, annual RPI inflation moved
between 6% and 26%, prompting a strong demand for inflation protection, although for the
first year only pension funds were allowed to buy the new asset class. Issuance of indexed
debt contributed to the credibility of the government’s anti-inflationary rhetoric by
diminishing the incentive to debase the real value of the outstanding debt, but the
programme proved a windfall for government finances, as inflation fell. Early breakevens
were about 9%, an expected inflation accrual cost considerably higher than has been
realised; within two years of the inception of the market, RPI dropped below 5%, and with
the exception of the boom and oil price shock at the end of the 1980s and the sharp rise in
energy prices in 2008, has remained below 5% to date.
Other than a bond issued in 1983 that was convertible into nominal debt, the 2% 1999
dubbed “Maggie Mays”, all gilt linkers issued prior to 2005 had the same idiosyncratic
format that has not been copied elsewhere. Since September 2005, starting with the first
syndication by the UK government of the 2055 linker, a new curve has been built up of
Canadian-style bonds. By 2007, all new issuance was in Canadian-style bonds, with six
issues stemming maturities of 2017 and longer being launched by the end of that year and
ten by 2010. This rapid issuance of new bonds contrasts with the 2035 old style linker,
which was the only bond launched between 1992 and 2005. The prospectuses of linkers
issued before the 2035 bond contained “comfort language”, giving some protection against
adverse RPI measurement changes. In the event of changes to the coverage, or calculation
of the RPI, which the Bank of England (acting as “index trustee”) deems “materially
detrimental”, investors will be given the right to sell bonds back to the government at
indexed par (par, adjusted for inflation), although that is not of great comfort at present
because all stocks under this protection are trading well above indexed par. For the 2035
and new Canadian-style bonds issued by the Debt Management Office (DMO), the choice
of linking index is at the Chancellor’s discretion, with the proviso that there is consultation
with a body with “recognised expertise in the construction of price indices”. This choice will
be “conclusive and binding”.
Government funding plans are laid out annually in a “Gilt Remit” within the Treasury’s “Debt
and Reserves Management Report”. This generally coincides with the Budget, just ahead of
the beginning of the new fiscal year in April. The remit contains an estimate of the total size
of linker sales, by cash value, to be carried out in the new fiscal year. Planned auction dates
for the year are released at this time, and there is often guidance as to how plans might be
altered in the event of changes to the health of public finances. Formal remit revisions can
happen at any time, but two key times are early in the new fiscal year once the prior year’s
finances are known and in the final quarter of the calendar year, just after the Pre-Budget
Report is announced. Auction stocks are announced quarterly, although they are subject to
remit revisions, with size details announced on the Tuesday in the week before an auction.
The majority of issuance since November 1998 has been via auction, indeed almost all prior
to 2009. Linker auctions are single-price – ie, all successful bidders pay the same price, in
contrast to nominal gilt auctions, which use a multiple price mechanism. Since 2009 there is
an additional 10% of stock for each auction that may be sold to successful bidders at the
clearing price in a post-auction facility. The DMO also sells linkers via mini-tenders, which
are auctions of about half the size of regular auctions that are announced at shorter notice
and for which there is no post auction allotment. Institutional buying at auctions must come
via index-linked gilt-edged market makers (IL GEMMs). While in the past only some nominal
15 March 2010 75
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
market makers were IL GEMMs, in recent years all nominal GEMMs have also traded linkers.
These banks are also the route through which the DMO can conduct syndications, switch
auctions, taps and other market activities. IL GEMMs can also access the DMO’s standing
repo facility (10% of the Bank of England’s Policy Rate at the end of 2009), though with a
few exceptions, such as the IL32, it is rare for the linker repo to stray far from general
collateral rates for long.
Syndicated issuance of gilt linkers was first used in September 2005 for the launch of the
2055 linker, due to the extreme extension of the curve and the innovation of the new bond
format. Syndicated gilt linker issuance was not used again until 2009, when the DMO noted
in its response to the consultation on the supplementary methods for distributing gilts,
March 2009, that it would use syndicated issuance alongside the auction programme in
FY09/10 to issue larger volumes of gilt linkers and long nominals than it judges would be
possible via auction. The DMO subsequently announced in its FY09/10 gilt remit that it
would issue a combined £18bn of gilt linkers via syndication and mini-tender in FY09/10.
Although the DMO stated in its response to the consultation on the supplementary methods
for distributing gilts that it expected syndicated issuance to occur no more than once a
quarter, after the syndicated launch of the 2042 linker in July 2009, which saw £5bn
notional of the new bond issued, the DMO conducted two syndications in Q3 09: the launch
of a 2050 linker for £5bn notional in September 2009 and a new 2060 maturity nominal gilt
in October 2009. The 2050 linker was the first gilt linker to use the new March/September
coupon series, in contrast to older new-style linkers, which were all issued with maturities of
22 November.
While there has been some non-government issuance of sterling inflation-linked bonds
since the mid-1980s and RPI swaps have traded since the early 1990s, until the early 2000s
gilt linkers were by far the dominant feature of the UK market. In 2000, when the Minimum
Funding Requirement (MFR) encouraged pension funds to favour gilts, asset swaps were
sufficiently attractive for significant supranational swapped supply. This helped to kick start
what was until then a niche RPI swaps market. As equity markets declined in the following
two years, the life insurance industry began to focus on the potentially more accurate
liability-matching benefits of using inflation derivatives rather than bonds. In particular, the
PS16/04 regulation hastened the use of inflation swaps by the insurance sector. Initially,
they covered their life policy RPI exposure before focusing on immunising bought-out and
in-house pension portfolios. It was not until 2004 that pension specific, rather than
insurance business, became the main driver of the UK swaps market; but having been
spurred on by heavy swap activity around the time of the IL55 launch in September 2005, it
quickly grew to dominate end-user demand.
In the middle part of the decade inflation derivatives became the main source of inflation
liability hedging; however, index-linked gilts remain the main stock of assets held against
pension liabilities. Even with the government’s increased issuance, the market remains far
too small to address the liabilities completely. Fundamentally, there remain notably more
RPI-linked liabilities in the UK than available assets, though the imbalance is steadily
reducing given the sharp increase in government issuance. The Pension Protection Fund
(PPF) survey of 7800 pension schemes showed estimated private sector pension liabilities
to be £905bn at the end of December 2009. The PPF liability measure is more conservative
than most schemes report in their accounts with measurement based on nominal and
inflation-linked gilt curves. However, there are few new inflation liabilities accruing in the
private sector, given scheme closures and most remaining exposures being capped at 2.5%.
As recently as 2003 insurers and pension funds owned over 90% of all gilt linkers, but their
share of the total market has fallen significantly since, largely due to inflation hedging
increasingly occurring via derivatives. Insurance companies now hold almost as many gilt
linkers as pension funds, at 32% of the market compared with 34% in mid-2009 according
15 March 2010 76
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
to National Statistics data. This is mostly life insurers who are matching real annuity
obligations and increasingly pension fund obligations that have been bought-out (ie, the
risk of pension funds has been transferred to the insurer). Typically when a scheme is
bought out, the purchaser immunises as much risk as possible, typically with a much higher
asset allocation into inflation than before the liabilities are transferred. Buy-out demand was
probably the largest single source of end-user demand in the linker market in 2008,
whereas in 2009 a more important driver appeared to be the significant number of very
large pension schemes that were closing further accrual to existing members. When
schemes close they often also choose to de-risk, particularly if closure also involved an
injection of cash from its sponsor, as the uncertainty of their future liabilities is considerably
reduced without active members.
Traditionally most linkers were held versus an index benchmark rather than as directly
matching liabilities, but the percentage of the market held this way has been declining
steadily as investors have increasingly moved towards liability focused investment
strategies even in their bond portfolios. Over 5y indices are more widely used than all linker
indices, and it is not a coincidence that the yield on the FTSE over 5y gilt linker index used to
be the reference for MFR liability measurement. This means that when a bond drops below
five years, there can be a significant dislocation to the market, with the bond itself liable to
underperform due to forced selling, but the bonds remaining in the index are likely to be
supported due to an extension of the duration of the index.
Pensions’ regulation encourages recognition of the nature of pension liabilities and has been
a major factor behind increased inflation-linked demand since the 1990s. However, unlike
the MFR around the start of this decade, in recent years regulation has surprisingly not
prescribed how pension funds should address their exposures. There is no longer a
generally accepted discount curve dominating all others. This is likely to have encouraged
the use of swaps, which are likely to be closer than gilt linkers to the FRS17/IAS19
accounting definition. Buy-outs are typically priced off a gilt linker curve, with insurers
tending to have more regulatory flexibility if their base investment is in gilt linkers than
swaps. Neither the Pensions Regulator nor the Pension Protection Fund has been actively
pushing schemes backed by strong sponsors towards lower-risk or liability-driven solutions.
The Pensions’ Regulator has pushed for pension schemes to be safeguarded when firms are
taken over by lower-rated, or foreign, entities, which has led to significant capital injections
into pension funds, but a scheme’s trustees decide whether to use this injected money to
immunise risks. The Pension Protection Fund encourages risk reduction by schemes that
are in assessment for being taken over by the Fund, but it is not prescriptive beyond this
universe even though its Section 179 liability estimate is referenced off the gilt linker curve.
While there are few new RPI pension liabilities accruing in the UK, we expect nuclear liability
hedging to become a new source of demand for ultra long linkers in the coming years
following the announcement in 2008 that the government will allow new nuclear power
plants. The owners of new facilities will be required to set aside funds for decommissioning
costs. The Nuclear Liabilities Financing Assurance Board (NLFAB) will oversee investment of
these funds, which according to the original White Paper will be valued against an ultra long
real discount function. Details of how funds will be formally valued have yet to be
announced but they may be invested more conservatively than those held versus ultra long-
dated pension liabilities.
Mechanics of UK linkers
While new issuance is in Canadian-style linkers and most of the stock of UK linkers is now in
new-style issues, there are still many old-style bonds. The new-style issues have a virtually
identical framework to US TIPS, with the exception that they have no deflation floor (ie, they
15 March 2010 77
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
can be redeemed below par if RPI falls over the lifetime of the bond). Old-style linkers also
have no deflation floor, but they have already accreted a considerable amount of inflation.
The mechanics of new-style linkers are addressed earlier in this guide, but calculations for
old-style linkers are more complex and are set out below.
Instead of trading in real space, with a real price and with settlement amounts uplifted or
reduced to reflect the inflation experienced over the life of the bond, old-style linkers trade
in clean price cash terms (not real), with the traded price rising and falling to reflect inflation
that has occurred. An example of the difference in price evolution between the old-style
2.5% IL16 and the new-style 1.25% IL17 is shown below. As this demonstrates, in a positive
inflation environment, the clean price of the old-style linker tends to drift higher; hence,
linkers first issued in the 1980s now trade with prices well above £200. Since the price of an
old-style linker already includes accrued inflation, no index ratio is used to create the
settlement price, unlike for new-style linkers, and accrued interest is calculated on the cash
value of the coupon, paid on an actual/actual basis.
Figure 37: Example of difference in pricing styles between old-style and new-style linkers
Linker Clean price Index ratio Accrued Dirty price
interest
UK IL 2.5% 2016 (old-style) 290.493 N/A 2.79840 293.29
UK IL 1.25% 2017 (new-style) 103.72 1.11487 0.15014 115.78
Note: As of the close of 30 December 2009. Source: Barclays Capital
To trade in nominal space, it is necessary to know the inflated value of the next coupon so
that accrued interest can be calculated. As a result, the inflation indexation for the coupon
of an old-style linker is done with an eight-month lag (a coupon’s cash value will need to be
known six months before it is due, and it takes some time to gather and publish the price
information for the final month). Accrued interest is then calculated in the usual way for
gilts, but using the known inflated value.
For example, the eight-month lag means that the principal value of the 2.5% IL13, issued in
February 1985 and redeeming in August 2013, will be uplifted by the ratio of the RPI for
December 2012 versus June 1984. So, investors “gain” the inflation for the eight months
prior to the bond’s issue, but “lose” the inflation for eight months prior to the bond’s
maturity. This term mismatch is not especially large in a benign inflation era, but history
shows that, at times, the effect has been large on the realised return.
Using this methodology, the cash value of semi-annual coupons for old-style linkers are
calculated as follows:
i is issue month
⎛ RPI r −8 ⎞
Redemption value = 100⎜⎜ ⎟⎟
⎝ RPI i −8 ⎠
15 March 2010 78
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
A few extra stages are required to find the real yield to which the nominal price of the old-
style linker equates. To derive the yield metrics of a bond from its nominal price, we must
first know the nominal value of all of the bond’s cash flows to maturity. However, future RPI
prints define the value of a UK linker’s future cash flows, making the future cash flows of an
old-style linker uncertain, thus preventing a nominal yield calculation. This is irrelevant for
Canadian-style linkers because they are quoted at a real price, which is translated into a real
yield by using the same calculations as for a nominal bond and into a nominal settlement
amount by using their index ratio. However, for old-style linkers, we are required to use an
inflation assumption to determine the future value of the bond’s cash flows, thereby
allowing us to convert the nominal price of the old-style linker into a nominal yield (also
referred to as the “gross redemption yield” or “money yield”). This inflation assumption is
then removed from the “money yield” to create the real yield.
To arrive at the nominal yield or “money yield” of an old-style linker, it is assumed that the
RPI grows at an assumed rate beyond the last RPI print. By convention, this assumption has
been 3% per annum since the mid-1990s (prior to this, it was 5% and, originally, 10%).
Using this RPI assumption, the coupon payments and redemption value of an old-style linker
are then mapped out according to the assumed future path of RPI this creates. From these
cash flows, the internal rate of return, or “money yield”, is then calculated for any given dirty
price in the same way as the internal rate of return for a nominal bond.
Once the “money yield” is found, the inflation assumption is then removed to give the “real
yield” by using the following calculation, which is the convention:
2
⎛ y⎞
2 ⎜1+ ⎟
⎛ g⎞ ⎝ 2⎠
⎜1+ ⎟ =
⎝ 2⎠ (1+ f )
Where g is the real yield, y is the money yield and f is the RPI assumption.
15 March 2010 79
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
The DMO’s “Formulae for calculating gilt prices from yields”, 16 March 2005 update,
www.dmo.gov.uk/documentview.aspx?docname=/giltsmarket/formulae/yldeqns.pdf&page
=Formulae/Calc, gives a closed solution real yield formula. It also highlights the detail of Canadian-
style linker calculations. The yield formula, expressed algebraically, is daunting, a practical spreadsheet
calculation is less so.
The real yield formula covers traditional bonds with two or more cash flows left (when there is
only one coupon remaining, the bond has known nominal value and, hence, is best valued
using a money market yield). The term “quasi-coupon date” is the theoretical cash flow date
determined by the redemption date; weekends and holidays may mean true payment dates
differ.
Any errors of duplication are ours, and we have also trimmed and altered the wording of the
explanatory notes. Readers should refer to the above official publication to see complete
details.
⎡ acw 2 ⎤ r r r
+n
P = ⎢d1 + d 2 (uw ) + (1− w n−1)⎥(uw ) s + 100au s w s , for n ≥ 1
⎣ 2(1− w ) ⎦
Where:
d1 = Cash flow due on the next quasi-coupon date per £100 face.
d2 = Cash flow due on the next but one quasi-coupon date per £100 face.
1
w=
g
1+
2
RPIB = The Base RPI for the bond – ie, for the month eight months prior to issue date.
k = Number of months from the month whose RPI determines the next coupon to the month of
the latest RPI
2k
RPIL 12
a = u
RPIB
15 March 2010 80
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
y = NY + (3% – BE)
where NY is the comparable nominal gilt yield and BE is the market’s assessment of the
appropriate breakeven inflation rate. If exactly zero inflation were expected for an old-style
linker, a projected set of cash flows that is created using the 3% inflation assumption will
have to be priced such that the yield they would generate is 3% above that of the
conventional gilt for the yield on the inflation-linked bond to be at fair relative value. If an
average of 3% y/y RPI inflation was expected to prevail over the life of the old-style linker, a
projected set of cash flows assuming 3% inflation would have to be priced such that the
money yield of the linker would be equal to the yield of the nominal gilt.
Each time an RPI inflation print is released, it replaces one month of the 3% inflation
assumption with the actual value. Mechanically, if the price of the bond was unchanged, its
real yield on the above calculation would alter so that the m/m inflation rate would deviate
from 0.247% (ie, 3% annualised). For example:
Figure 38: Sample mechanical real yield adjustments on old-style linkers for an RPI print (bp)
Projected RPI (November
Real yield adjustment (14 December 2009)
2009)
Index m/m IL11 IL13 IL16 IL20 IL24 IL30 IL35
value
216.1 0.05% -11.1 -5.4 -3.1 -2.0 -1.5 -1.2 -0.9
216.2 0.09% -8.4 -4.1 -2.4 -1.5 -1.1 -0.9 -0.7
216.3 0.14% -5.7 -2.8 -1.6 -1.1 -0.8 -0.6 -0.5
216.4 0.19% -3.0 -1.5 -0.9 -0.6 -0.4 -0.3 -0.3
216.5 0.23% -0.3 -0.2 -0.1 -0.1 0.0 0.0 0.0
216.6 0.28% 2.4 1.1 0.6 0.4 0.3 0.3 0.2
216.7 0.32% 5.1 2.4 1.3 0.9 0.7 0.6 0.4
216.8 0.37% 7.8 3.7 2.1 1.4 1.1 0.8 0.6
216.9 0.42% 10.5 5.0 2.9 1.9 1.4 1.1 0.9
217.0 0.46% 13.2 6.3 3.6 2.4 1.8 1.4 1.1
Note: Example uses projection for November 2009 RPI print, which printed 216.6. Source: Barclays Capital
In this example, November RPI printed 216.6, 0.28% m/m, slightly higher than the 0.247%
m/m inflation assumption embedded in the bond’s cash flows. Hence, real yields on old-
style linkers mechanically adjust upwards immediately after the RPI release by the amount
of the difference between the actual data and the assumption. As Figure 38 shows, this
equated to a mechanical cheapening of the IL13 real yield by 1.1bp. The real yield on the
old-style linker may also adjust for changes to its market price, due to, for instance, a
change in inflation expectations brought about by the RPI release. However, all else being
equal, the real yield on an old-style linker will only adjust on an RPI release to reflect this
mechanical effect, and the price of the old-style linker will mechanically adjust to reflect
realised inflation.
15 March 2010 81
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
As Figure 38 shows, the yield effect on an unchanged price will be much greater on a short
bond than on a long one. In the event of a flat m/m RPI print, an unchanged price for a bond
with 1-year duration would mean a yield reduced 25bp, whereas the yield of a bond with a
duration of 25 years would be reduced only 1bp on an unchanged price. As a result, if the
month-on-month RPI print were zero and if the price of an old-style linker with exactly one
year to maturity is unchanged then mechanically the real yield would fall by 0.247%.
Alternatively, the price of the old-style linker would have to fall 0.247% for the yield to
remain unchanged. The discrete shifts in yield purely as a result of a new RPI release may
seem counterintuitive, as the real yield will appear to rise as the price also rises, but this is
simply a function of the difference between the “nominal” quoted price and the “real” quoted
yield, with the real yield adjusting for the deviation from the 3% inflation assumption and
the price adjusting for the realised inflation.
Figure 39: The effect of RPI releases on the assumed RPI schedule
216 2.16
215 2.15
214 2.14
2nd Cashflow
213 2.13
212 2.12
May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May
Although some investors may feel more comfortable with the gradual adjustments of the
mark-to-market value of the inflation-linked bond with respect to the price index (with the
3-month interpolated lag) that the Canadian model incorporates, old-style UK linkers have
an advantage in terms of the speed with which they compensate the investor for the value
of an inflation surprise. Instead of waiting up to six weeks for the effect of a large month-
on-month number to gradually filter through to the reference index ratio, the UK market can
react immediately and price in all available information as soon as it is published. However,
this argument does not offset the downside of the eight-month lag for coupon indexation.
Overall, the price adjustment of old-style linkers in response to a relatively high or low m/m
RPI is the UK linker market’s equivalent of inflation accretion. In terms of carry, there are clear
differences between the UK and Canadian methodologies that need to be controlled in order
to compare the relative merits of positions. The most significant difference to bear in mind is
the effect that the presence of an embedded 3% assumption has on carry and the market’s
reaction to carry. As noted, the assumed RPI schedule will only be lifted higher by a m/m RPI
print above 0.247%, and will fall for any number below it. Therefore, only if a m/m RPI print
is notably above 0.247% will a shift in yield be created that is worthy of a strong reaction
from market prices. To the extent that strong seasonals, be they negative or positive, are
predicted by the market, they add to mechanical yield volatility but not to price volatility.
As breakevens and actual inflation trend away from 3%, the use of this assumption can
become very misleading for shorter maturity bonds. In the final year of a bond’s life it can be
15 March 2010 82
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
difficult to meaningfully compare its screen real yield with the yield on its nominal
comparators or the real yield of longer-dated gilt linkers. In this situation, value for a
traditional linker is better assessed in an alternate way. It is fairly straightforward to evaluate
the nominal return for a short linker for any given percentage change in the RPI over the
remainder of the bond’s indexed life at the current price. Expressed as a spread to Libor, it
can then be compared with nominal gilts. The RPI level that offers a consistent Libor spread
compared with the nominal gilt then offers a more representative breakeven value.
We prefer a z-spread asset swap comparison for micro valuations of old- and new-style gilt
linkers because this measures the inflation elements versus a consistent curve. A more
straightforward, but less complete, method is to compare the yield of traditional linkers with
that of new-style linkers forward to the end of their known carry period because this leaves
the same known inflation data in the pricing of both bonds. While this is an appropriate
simple method to compare moves in real yields and breakeven, the distortions due to the
embedded 3% assumption in traditional linkers do not make the spreads between bonds
using this approach representative, especially at shorter maturities.
Taxation issues
UK gilt linkers have a more favourable tax treatment than either UK corporate or
international government bond issues. For most corporation tax purposes, an inflation tax
relief is granted based on the inflation experienced between tax year-ends. This relief is
deducted from the total return (calculated on a mark-to-market basis or on an accrual
basis, according to an election made by the investor), and the difference is taxed. This
means that index-linked enjoy a material tax advantage over nominal gilts – the aim and
effect is that investors are only taxed on their real return, not on inflation compensation.
This is almost the same as saying that the inflation increase in principal is not taxable. There
are two reasons it is not the same: if an investor’s tax year-end is, say, December, the relief
will be based on the RPI change from December to December, without a lag, whereas
indexation occurs with an eight-month lag; and the starting value in any tax year is unlikely
to be exactly indexed at par.
It is worth noting that most index-linked gilts are held by pension funds or within the
pension business lines of life assurance companies, which do not pay tax. Tax usually has a
material effect on the market only at sub-5y maturities where institutional investors are less
dominant. Private individuals who hold UK index-linked gilts only pay tax on income
accrued during the financial year, so they get all gains – inflation-linked, or otherwise – tax
free. This also means that losses, in the event of falling RPI, are not allowable against tax.
The relatively favourable tax treatment for gilt linkers coupled with a range of government-
backed national savings products means that a sterling structured inflation note market has
not developed, in contrast to other countries. This lack of structured product demand has
resulted in a less active inflation swaps market at the short end of the curve.
Investors in corporate inflation-linked bonds do not enjoy the inflation relief afforded to gilt
linkers. The inflation uplift is taxable – ie, no inflation credit is applied. The UK’s Inland
Revenue decided that since corporate issuers are allowed to offset the inflation uplift
against taxable income – in the year that it accretes – corporate linker investors should not
receive inflation relief. However, certain issuers might be able to obtain an exemption.
Corporate UK linkers
At the end of 2009, the market value of non-government bonds in the Barclays Capital
Sterling Index was over £23bn. However, non-government inflation-linked issuance from
anything other than government guaranteed Network Rail was extremely limited in 2008
15 March 2010 83
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
and 2009. This contrasts sharply with 2005-07, which saw unprecedented corporate
inflation-linked bond issuance. However, during this period, most new corporate linkers
were not bought by bond investors and, hence, are not eligible for inclusion in the Barclays
Capital Sterling Index. By the end of 2007, there were about £20bn of corporate inflation-
linked bonds that had been monoline-wrapped to transform them into AAA issues based on
utility, PFI and infrastructure projects. Of those issued between 2005 and 2007, the vast
majority had been absorbed by asset swap investors, particularly covered bond investors,
who provided for the inflation swap needs of pension funds by translating these into real
rate swap paying flow. However, as AAA insurers themselves lost their ratings, this demand
effectively ended.
Utility companies have been a major source of corporate linker issuance. Many regulated
utilities have a degree of RPI-based pricing restrictions and, hence, a clear inflation-linked
revenue link. Water companies, in particular, issued the majority of their debt in inflation
following the five-year fixing of water prices at the end of 2004. At its five-year fixing of
water prices in November 2009, the water regulator announced a £22bn investment
programme in the network. If there is demand, this may well prompt further issuance of
inflation-linked bonds, as the water regulator has indicated that the regulatory structure will
remain RPI-linked over the long term. However, if the bond route is not available, IAS39
accounting standards make it difficult for corporates to pay inflation without facing the
mark-to-market volatility of these positions on their balance sheets, unless they can prove
they have an expectation of closely matching cash.
Straight RPI bonds can be sold and accounted for on an accruals basis, but not LPI and
other more structured formats. Utility companies taken over by private equity and
infrastructure consortia may be able to use inflation derivatives with mark-to-market
considerations less of a concern. Yet, despite the improvements in the credit markets, the
overhang of old monoline wrapped supply may limit the scope for corporate linker supply
from utilities. The largest single utility issuer in corporate linkers is National Grid Plc, which
has electricity and gas prices linked to RPI. National Grid Plc had over £4bn in inflation-
linked bonds outstanding as at end-2009, with the largest bond £450mn notional, but less
than £200mn issued in 2008 and 2009.
Private Finance Initiative (PFI)-related deals have been another source of non-gilt issuance.
PFI deals involve a private company building infrastructure and being paid an income
stream until the asset comes under the ownership of the relevant authority. Most hospital-
related projects involve RPI-linked cash flows that will be paid to the financier once the
hospital is operational and, hence, are ideally suited to funding via inflation-linked issuance.
Thus, this kind of issuer has been numerous, although other large PFI projects have also
involved partial financing via inflation-linked. Almost all of these issues were wrapped with
credit guarantees to enable AAA ratings and, as with utility supply, most ended up with
asset-swap investors. Due to specific accounting restrictions, PFIs almost always issue
bonds rather than paying swaps. In duration terms, PFI financing has been a less important
factor than utilities in recent years as most, particularly hospital bonds, are amortising either
immediately or after a number of years when the building project is expected to be
complete and the offsetting of cash flows begins. Hence, while many utility linkers have
durations above 30, it is rare for that of PFI bonds to be much longer than 15 years.
Despite difficulties in sourcing PFI funding from 2008 due to adverse credit conditions, the
logjam of PFI projects was eased in March 2009 by the government offering backup funding
via the creation of the Treasury Infrastructure Financing unit. Recent PFI deals that have
gone through have also had a larger percentage of funding from the European Investment
Bank (EIB). For instance, almost half of the M80 road widening project was financed by EIB.
While PFI has been a source of significant supply over the past decade, the pipeline of deals
remaining in this format is limited, particularly those with inflation-linked revenues. In the
15 March 2010 84
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
absence of a wrapped bond market, only contractually RPI-linked revenue streams are likely
to be hedged by projects directly, via swaps.
While in recent years most long supply was absorbed by asset swappers, shorter on the
curve an active swaps market sometimes enabled natural AAA issuers to issue to real-
money investors and swap back themselves. Over £3.5bn of natural AAA bonds have been
issued by foreign entities since the start of 2000, but this is exaggerated by some
unswapped issuance in the past from EIB, when it was lending to PFI projects. This is no
longer seen as, in recent years, EIB has been able to access inflation swap markets directly
rather than hedging inflation-linked cash flows by issuing inflation-linked bonds.
The only major non gilt inflation-linked supply during 2008 and 2009 came from transport
infrastructure, specifically Network Rail, which issued £6.4bn over this period including
private placements. Network Rail is unusual in conducting much of its real rate funding on a
programmatic basis, having launched three benchmark issues at 20y, 30y and 40y in 2007
and continuing to re-open these in the following two years, in line with its aim to provide an
alternative to gilt linkers. As a natural AAA name backed by a letter of support from the
government, Network Rail has functioned as a very important source of non-government
inflation-linked supply in the absence of any other significant activity; even so, the increase
in gilt linker supply in 2008 and 2009 was barely enough to offset the overall absence of
non-government inflation-linked supply.
The early corporate linker issuance in the 1980s was mostly linked to property, and property
flows remain important today. For a property revenue stream linked directly to inflation,
straight paying of inflation or real rate swaps is possible. Property transactions are the main
source of LPI supply, as many contracts have explicit inflation floors and caps, although
overall LPI supply has reduced significantly in recent years. We believe there is substantial
potential for inflation flows stemming from property in the coming years, and
securitisations in the sector are likely to increase if the commercial property market
stabilises. However, if non-government bond supply does resume, we would not expect it
much beyond 20 years on the curve, given that the overhang of old monoline wrapped
issues is concentrated beyond the 30y sector.
15 March 2010 85
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
The absence of asset swap paying flow led to significant distortions between inflation bonds
and swaps, with gilt linker asset swap levels becoming much cheaper than nominals in Q4
08 and long forward real rates pushing into negative territory. Many of these distortions
have corrected significantly, and the inflation swap and gilt linker markets still interact via
asset swap activity, now concentrated in gilt linkers. The level of linker asset swaps fell in
2009 from the extremes reached in Q4 08 as pension funds and life insurers who had
already immunised inflation and duration were significant buyers of gilt linkers on asset
swap. This flow was accelerated following the gapping wider of relative z-spreads versus
nominal gilt asset swaps in March 2009 on the announcement of BoE gilt purchases. Hence,
despite the quantitative easing deliberately not focusing on linkers to avoid falling real
yields, causing problems to the pensions sector, in practice linker yields were affected very
significantly. The outperformance of bond over swap breakevens was also briefly helped by
PFI paying flows in mid-2009 after the logjam of projects was ended by the government
offering backstop funding via the creation of the Treasury Infrastructure Financing unit.
Short-dated linker asset swaps also benefitted from demand from bank treasuries.
While the underlying liquidity of the inflation swap market recovered significantly in 2009, it
is no longer consistently superior to inflation-linked bonds. The most frequently traded
maturity is still 30y. While 10y is relatively frequently quoted, liquidity to take positions in
short-dated RPI swaps remains poor, not helped by the lack of a structured note market,
which often drives sub-5y interest in other countries. While the most common format of
inflation swap in the UK market remains zero-coupon RPI swaps, real-rate RPI swaps can
also trade directly, both zero coupon and in a par-swap format, providing a consistent
comparison with nominal swaps. The standard initial lag for almost all swap transactions
other than asset swaps is a two-month calendar lag. While domestic pension funds provide
the main source of demand for RPI swaps, they are also traded by a variety of other
leveraged and real-money investors. Forward trading in inflation and real-rate swaps is
present, even at long maturities, and to some extent this aids liquidity and helps to reduce
relative value distortions across the curve. It is also possible, albeit rare, to see trading in
forward swaps versus other markets.
From a liability perspective, most pension exposures are linked to LPI rather than RPI.
Uncapped RPI linkages were relatively uncommon outside of the ex-public sector,
particularly between 1997 and 2005 when, under the Pensions Act of 1995, there was a
minimum statutory entitlement for new defined benefit liabilities to accrete annually at the
rate of RPI inflation capped at 5%, with an implicit floor at 0; LPI (0,5). Prior to 1997, LPI
(0,5) was the most common pension specification but exposures varied scheme by scheme,
with LPI (0,3) and LPI (3,5) other popular variants. From April 2005, under the Pensions Act
of 2004, the legal minimum for new pension liabilities is LPI (0,2.5). This leaves very few
new inflation liabilities accruing, with the 2.5% cap so far out of the money that it creates
no new hedging demand.
In 2005 a large proportion of pension demand in inflation swaps was in LPI (0,5), but in the
absence of significant supply in this format this drove LPI (0,5) swaps rich versus RPI, in
which there was regular issuance via non-government asset swaps. By Q2 06, LPI (0,5) no
longer traded at a discount to RPI swaps at a 30y maturity due to the supply/demand
imbalance, despite the cap level being much closer to spot RPI levels than the 0% floor. This
led to relatively limited trading in LPI (0,5). An exception was in H2 07 when higher and
more volatile RPI forwards drove a modest discount for LPI (0,5), which prompted renewed
pension interest. At the same time, increased property-related paying encouraged an
improved liquidity in LPI in general and LPI (0,5) in particular. Many property rental streams
15 March 2010 86
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
have an LPI-like structure, but often the annual growth is capped at levels other than 5%
and/or floored above zero.
The LPI market has helped define, albeit also heavily skew, the smile in RPI volatility to a
greater extent than seen in other inflation markets. In addition to pure LPI flows related to
LPI (0,3) and LPI (3,5), there is also outright interest in straight RPI caps and floors at 3%. As
LPI bond issuance does not offer the same accounting advantage as RPI supply, the
volatility market is set to remain concentrated in derivatives format. By contrast with LPI,
the inflation and real-rate swaption markets remain surprisingly subdued. This is partly due
to the lack of natural inflation volatility issuers and partly to a self-reinforcing lack of
perception of liquidity. However, there may be scope from regulation, which is only partially
RPI-linked to prompt conditional supply from regulated industries. Such supply, or a
significant further pick-up in activity in LPI swaps, could be sufficient to kick start the
swaption market, given how popular nominal rate and equity swaption strategies have
proved among pension funds. On the other hand, the trends since 2007 have been towards
simplification of risk and product, with the extreme and unanticipated realised volatility of
RPI inflation as well as market conditions discouraging non-linear risk taking.
Figure 40: UK linkers: Historical performance and risk Figure 41: Return/risk linkers versus nominals and equities
18% 4 Equity
UK IL Returns UK IL Ann Mthly Vol IL
16%
3 Nominal Bonds
14%
12%
2
10%
8% 1
6%
0
4%
2%
-1
0%
-2% -2
1999 2001 2003 2005 2007 2009 1999 2001 2003 2005 2007 2009
15 March 2010 87
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Japan
Stefan Liiceanu The Japanese Ministry of Finance (MoF) has been issuing inflation-linked JGBs (JGBis)
+81 3 4530 1554 since March 2004. Within five years there were 16 bonds with total capitalization of nearly
stefan.liiceanu@barcap.com JPY10trn (USD96bn). However, unbalanced ownership, heavily tilted toward overseas
leveraged investors, coupled with the 2008 global financial market turmoil, forced a
suspension of the program, with the last auction in August 2008. Since then, the MoF’s
efforts have been concentrated on absorbing excess supply from the secondary market
via buyback auctions. 40% of JGBis originally issued were retired by the end of 2009,
helping 10y breakevens to rebound to around -50bp from under -300bp. The MoF has
penciled in a JPY300bn JGBi auction for FY 10, with the likelihood that new bonds will
include a principal floor to stimulate demand from domestic market participants.
The Japanese CPI is calculated based on the Laspeyres method, which compares the prices
of goods and services according to their weights at the time of the base year (currently
2005) with the y/y change in those figures. Under the Laspeyres Price Index, the weight
(quantity) of various goods and services is fixed at the base year. Accordingly, goods that
have undergone steep price drops in recent years, such as IT goods, do not reflect any
increases in real volume accompanying such price drops (for the calculation of the price
index for such items the MIAC uses the Hedonic adjustment method based on the
correlation between the price of products and their performance derived from available
sales data). Goods undergoing price increases, on the other hand, tend to be overestimated
in the calculation because there is an increasing impact due to the fixed weightings. As a
result, the Laspeyres method tends to show an upward bias, which increases the further
one moves away from the base year. To address this issue, from January 2007 the MIAC
decided to begin releasing CPI numbers based on the chain-weighted calculation method as
a supplement to the fixed-weight CPI, where weights are adjusted each year to reflect
changes in spending patterns. To date, the upward bias in the inflation rate based on the
fixed-weight index has not been particularly high – on average 0.2% larger than the inflation
rate as measured by the chain-weighted core CPI.
15 March 2010 88
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Figure 42: Major items in the Japanese core CPI (2005-base index)
Misc., 6.1%
Recreation, 11.5% Food, 22.7%
Education, 3.8%
Transp. &
Telecom, 14.5%
Medical Care,
4.7% Housing, 21.3%
Clothes &
Fuel, Light &
Footwear, 4.8% Furniture &
Water, 7.1%
Household
Utensils, 3.6%
Source: Ministry of Internal Affairs and Communications (MIAC), Barclays Capital
Meanwhile, base revisions to the index are conducted once every five years for years ending
in “5” and “0”. The last such adjustment took place on 25 August 2006, when the CPI data
were rebased to the year 2005 (ie, the July 2006 nationwide CPI had the year 2005 as base
and 2005-base data were made available retroactively from January 2005). A total of 34
new items were added to the coverage, including digital electronics such as flat-panel TVs
and DVD recorders. The next rebasing will occur in mid-2011, when the index’s base year
will change to 2010. Further details on the Japanese CPI in the English language are
available from the MIAC Statistics Bureau (www.stat.go.jp/english/data/cpi/index.htm).
In the context of monetary policy, the Japanese core CPI naturally plays an important role,
being the BoJ’s “target” index, along with the corporate goods price index (CGPI). However,
unlike some other central banks, the BoJ’s definition of desirable long-run price stability based
on the core CPI is rather loose. According to the Bank, Monetary Policy Board members
understand medium- to long-term price stability as “an approximate range between zero and
2%,” with “most members’ median figures falling on both sides of 1% 2”. Moreover, the BoJ
stated that this view of price stability is flexible, “reflecting changes in the economic structure
such as further progress in globalization and innovations in information and communication
technologies” and that “as a rule Policy Board members will review it annually”.
2
See for example the BoJ Quarterly Bulletin, May 2006.
15 March 2010 89
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
As deflationary pressures eased and borrowing needs remained high, a year and half later, in
March 2004, Japan issued its first inflation-indexed government bond (JGBi1), with a 10y
maturity and a small pilot size of JPY100bn. The bond came at a low 15.5bp BEI (spot inflation
was -0.1% y/y at the time), but shortly thereafter, market perceptions vis-à-vis the new
instrument turned favourable and breakevens subsequently rallied to a high of 94bp during the
same year (for details please refer to Figure 44). Although the bond’s maturity was not
particularly long compared with the structure of other developed linker markets, and although
Japan had had a prolonged deflationary experience since the late 1990s, this first linker issue, as
well as all subsequent JGBis, did not feature a deflation floor, as per the initial proposal brought
forward by the issuing authorities. This initially left a notable gap between supply and demand
for deflation protection. The main reason that a floor was not included was to make the real yield
a relatively representative value of expected long-run inflation. Otherwise, the value of the
embedded floor would have been a significant part of the value of the new bond, given that
inflation had averaged -0.3% y/y throughout 2003 and market perceptions generally were not
particularly inflationary. The key product specifications of JGBis are illustrated below.
Maturity: 10 years
Type of Issue: Coupon-bearing bonds
Coupon Frequency: Semiannual
Minimum face value unit: JPY100,000
Issuance Method: Public offering
Auction Method: Yield-competitive auction/ Dutch-style auction
Recent Issuance Frequency: Bimonthly (even months)
Reference Index: Japan nationwide CPI ex-fresh food (Japan Core CPI)
Reference Index frequency: Monthly
Reference Index Seasonality Adjustment: No seasonal adjustment
Indexation Lag: 3 months
Indexation Style: Canadian Model (linear interpolation to the 10th of the month)
Floor: No floor
Strippable: Not strippable
Source: MOF
Since 2004 and until the global financial shock in 2008, the market evolved rapidly, both in
terms of size and depth. By August 2008, the MoF had issued a total of 16 bonds and the
overall market size had reached almost JPY10trn (about USD96bn, or nearly 1.4% of total
JGBs outstanding), placing Japan ahead of other inflation-issuing countries with longer
experience with the product, such as Australia and Sweden. JGBi auctions typically were
held on a bimonthly basis, with issuance amounts of JPY500bn per auction (for details
please see Figure 46). Up until October 2008, when issuance was temporarily suspended,
JGBi auctions were Dutch style, with bids being filled from the lowest yield (highest price)
until the entire amount offered has been raised. Under this approach, the auction clears at a
highest accepted yield and all bidders pay the same price for the bonds. Re-opening
auctions in FY 06 also were Dutch style, but from FY 07 all re-openings (three held to date)
became price-competitive auctions. The auction process itself has been identical with that
of nominal bonds, with the issuing authorities first announcing the exact details of the
offering (linker coupon and size) at 10:30 on the morning of the auction, at which point the
bidding begins. The bidding ends at 12:00 and the auction results are announced at 12:45 3.
3
News providers such as Reuters and Bloomberg disseminate the information in real time (for Bloomberg for instance
please refer to <NI JPB>).
15 March 2010 90
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Meanwhile, JGBis became eligible for MoF’s buyback operations in January 2007. Until April
2008, the MoF held five buyback operations, on each occasion retiring a modest JPY40bn to
JPY50bn worth of linkers. From April 2008 the Ministry stepped up its JGBi buybacks
(auction sizes of approximately JPY80bn) in response to the sudden decline in linker prices
that occurred around March (the on-the-run 10y linker’s BEI briefly touched the -2bp level,
as illustrated in Figure 44), a time when deleveraging by overseas investors distorted many
asset prices. As the global financial shock reached its climax during September-October
2008 and JGBi BEIs sank to unseen levels, the MoF cancelled two JGBi auctions planned for
October 2008 and February 2009, and increased the size and frequency of buybacks. From
October 2008, monthly JGBi buybacks averaged about JPY215bn, with two auctions of
JPY100bn to JPY120bn. The precise buyback amounts have usually been determined on a
quarterly basis, with the choice between retiring linkers, floating-rate JGBs or fixed-coupon
JGBs made following MoF hearings with the primary dealer community (since October 2008
the weight has been placed on linkers rather than floating-rate or regular JGBs).
Overall, the MoF held a total of 39 JGBi buyback auctions during 2007-09, retiring
JPY3.74trn worth of linkers from the secondary market. This represents nearly 40% of the
total JGBi issuance since the program’s inception and leaves the market size a more modest
JPY6trn as of the end of 2009. Some issues, such as JGBi2 and JGBi6, have been aggressively
retired, with a current outstanding market size of less than c.40% of the original amount
issued. At the time of writing, of the existing 16 JGBi issues, four are especially large –
namely, JGBi8, 12, 14 and 16, each accounting for at least 10% of the broad JGBi market. It
can be said that the original purpose of JGBi buybacks – to stabilize the market – has been
largely achieved. Long-dated BEIs recovered to near -50bp by the end of 2009 versus a low
of -323bp in mid-December 2008, and offer-to-cover ratios at buyback auctions have
declined significantly, falling to 2x or less from around 4x in early 2009 (Figure 45). Simply
put, buybacks helped address distressed selling and thus absorb excess supply in the
secondary market. Since the middle of 2009, most buyback auctions have been orderly,
with tight tails and average prices higher than market levels prior to the auction.
Note that in addition to MoF buybacks, from the end of 2008 JGBis have also become
eligible for BoJ outright purchase (Rinban) operations, under which the Bank buys
JPY1.8trn/month of government securities directly from dealers. The BoJ’s outright JGBi
Figure 44: History of 10y JGBi breakevens (Mar 2004-Jan 2010) Figure 45: Bid-to-cover ratios and auction tails in JGBi buybacks
150 6 +250
100 5 +200
50 4
+150
0 3
-50 +100
2
-100 1 +50
-150
0 +0
22 Jan 07
20 Aug 07
18 Feb 08
19 May 08
04 Aug 08
24 Oct 08
14 Nov 08
12 Dec 08
29 Jan 09
20 Feb 09
13 Mar 09
24 Apr 09
19 May 09
25 Jun 09
16 Jul 09
24 Aug 09
15 Sep 09
23 Oct 09
13 Nov 09
17 Dec 09
-200
-250
-300
-350 Buyback Tail (sen, RHS)
2004 2005 2006 2007 2008 2009 2010 Offer-to-Cover Ratio (x, LHS)
Note: from Mar 2004 to mid-Sep 2005 on-the-run JGBi BEI, thereafter seasonally Source: MoF, Barclays Capital
adjusted CMT 10y BEI derived from parametric curve fitting.
Source: Barclays Capital
15 March 2010 91
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
purchases have been more modest than the MoF’s buybacks, being held on a bi-monthly
basis (usually odd months) in amounts of JPY40bn. As of the end of 2009, the BoJ’s total
JGBi holdings amounted to JPY454.4bn, or just 13% of total JGBi buybacks by the Ministry
of Finance.
Future supply of JGBis will depend on how fast market conditions normalize and allow the
MoF to resume its regular auctions. The FY 10 JGB issuance plan announced at the end of
2009 does incorporate a JPY300bn JGBi auction (and total buybacks, including linkers, of
JPY3trn), which is to be implemented if the secondary market environment is favorable, ie,
breakevens return to positive territory. Interestingly, the official minutes of regular meetings
between the issuing authorities and dealers as well as domestic investors have revealed
strong support for this asset class, particularly if it is restructured to include a principal floor.
Although there has been divergence in domestic market participants’ opinions with regard
to the optimal timing to resume JGBi issuance (the minutes showed than some preferred
issuance of floored bonds to restart only after all current JGBis would have been retired,
while others argued for earlier supply of JGBis), overall domestic market participants have
argued that unlike floating-rate JGBs, inflation-linked bonds are a global asset class and
urged MoF not to terminate the program.
From the issuing authorities’ perspective, it also makes sense to foster the linker program.
Japan’s overall debt levels have been rising sharply, with its gross debt-to-GDP ratio now
near 200%. Rollover bond issuance is likely to soar in the next decade (official estimates for
yearly rollover JGB supply until FY 20 are between JPY104trn and JPY140trn), and budget
deficits of at least 3% of GDP are also likely to persist. In the past few years, domestic
appetite for floating-rate bonds has deteriorated sharply, as did interest from Japanese
households for retail JGBs. This leaves Japan’s overall funding channels at a disadvantage,
particularly as overseas investors currently own less than 6% of total JGBs outstanding. As
rising government bond supply volumes are usually accompanied by the necessity to
diversify funding channels, inflation-indexed JGBs should be an appropriate choice of
issuance in the medium to longer run.
Meanwhile, as is the case in other developed linker markets, trading volume in inflation-
indexed JGBs has naturally been lighter compared with nominal bonds, but liquidity did
improve significantly up until the middle of 2008, with bid/offer spreads in the 10y sector
tightening from 20-25 sen to around 10 sen between 2005 and 2008. Liquidity tended to be
somewhat lower in the front end of the JGBi curve, as issues JGBi1 and JGBi2 were originally
small (JPY100bn and JPY300bn, respectively), but longer-dated linkers featuring larger sizes
(eg, JGBi8, JGBi10, JGBi12, JGBi14 and JGBi16 all were originally issued in JPY1trn size)
exhibited better liquidity. Following the 2008 market turmoil and the MoF’s aggressive JGBi
buybacks, liquidity deteriorated to 2005 levels, with bid/offer spreads between 20 sen and
30 sen, and daily turnover of about JPY5-10bn, compared with JPY50bn or more at the peak
of the market.
15 March 2010 92
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
First, the daily reference CPI (DRI) for day N in month M is:
1) If N = 10, the reference CPI is the index three months previously, ie, CPIM-3
15 March 2010 93
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
(N-10)
DRI = CPIM-3 + (CPIM-2 – CPIM-3) x
No. of days from 11th of month M to
the 10th of month M+1
Next, the CPI ratio for any given day N is calculated as DRIN/Base CPI
An important feature of JGBis is that due to rounding conventions, inflation accrual does
not develop smoothly across the month. As mentioned above, the ratio of the reference
CPI on settlement to the base CPI on which the settlement price is based is only rounded
to three decimal places (and the CPI itself is published to one decimal place, unlike other
countries such as the US), in contrast to other markets where it is usually rounded to five
decimal places. Because of this, carry on a 10y JGBi jumps almost 1bp when the rounded
index ratio changes by 0.001, a situation not seen in other inflation markets. The number
of these “carry jumps” exhibited by linkers’ CPI ratio in any given month obviously
depends on the magnitude of the month-on-month change in the CPI.
JGBis are traded in real price terms – ie, without incorporating inflation adjustment. In the
broker market, linker prices move in 5 sen increments, and daily closes are also rounded to
the nearest 5 sen. While this is not a strict rule, it is an appropriate degree of accuracy given
the approximately 10 sen jumps in nominal price terms that occur with inflation accretion.
Settlement and day-count conventions applied to the nominal market are also used for
JGBis (T+3, ACT/365). The simple bond calculation between price and yield is used, and
therefore, real yields are determined by the rate that equates the traded price with the sum
of the bond’s cash flows discounted to present value.
15 March 2010 94
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Income tax is payable on JGBi coupon payments. The increase in the inflation-adjusted
principal, if any, at the redemption date is considered to be an interest payment and so, as
such, taxable. No tax is applied when the principal falls in value due to a decline in prices. In
the domestic market, “designated financial institutions” are not subject to Japanese
withholding taxes. Most overseas investors were excluded from directly owning JGBis before
April 2005, mainly because the tax status of the new bonds was not set. Even when holding
restrictions were relaxed, there was still uncertainty as to the treatment for investors usually
exempt from withholding tax for nominal JGBs. The National Tax Agency formally clarified
at the end of 2005 that JGBis’ interest and the gains or losses on principal would be exempt
from withholding taxes if they are held by “entities entrusted to manage corporate pensions
approved under the tax systems in the UK and the US, provided the bonds are held in book
entry form”.
In Japan, the typical inflation derivative transaction has been the zero-coupon inflation
swap, although in recent years other structures such as year-on-year inflation swaps,
forward-starting swaps, total return swaps and asset swaps have become increasingly
traded. Moreover, while transactions in inflation structured notes are lagging the
European and US markets, many of the popular structures in these areas have been
priced, including the classical “additive structures” and “multiplicative structures” (eg, the
note’s payoff is determined by y/y core CPI growth plus a spread and a floor, or by a
multiple of inflation with a floor of 0% or higher). The main demand for structured
products involving Japanese inflation has been in the 5y and 10y sectors. Furthermore,
since 2008, more innovative 5y inflation-linked term deposits have been offered by
regional financial institutions.
All of the derivatives mentioned above feature the same three-month indexation lag and
interpolation to the 10th of the month. Regarding the zero-coupon swap market in Japan,
liquidity was initially concentrated in the ~6-10y zone (bid/offers similar to JGBis), but
since early 2006, investor interest picked up both in shorter and longer maturities. In the
short end, all tenors including 1y inflation swaps have traded, while in the longer end
both 15y and 20y inflation swaps have traded. Further out the curve, 30y inflation swaps
are regularly priced but activity has been limited. Forward and relative spread trades have
also become increasingly important to the pricing of the JPY swap breakeven curve. While
the inflation element of 10y notes has been straightforward to hedge, there have been
times when the 5y sector of the swaps curve has been driven rich, prompting significant
interest in buying 5y5y forward breakevens.
Pricing the belly of the Japanese CPI curve has been particularly complicated by
expectations of consumption tax (VAT) hikes. One of the supportive factors for 4-5y
15 March 2010 95
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
swaps into the start of 2007 was increased expectations of consumption tax hikes sooner
rather than later, given the apparent strength of the German economy through its 3%
VAT hike (indeed as of March 2007, major Japanese think tanks expected two hikes, one
as early as 2009 and the second one around 2012-13). The direct impact on the Japanese
CPI basket of a hike of a similar magnitude would likely be to boost consumer prices by at
least ~1.5-2.0% (the CPI pass-through at the time of the 3% sales tax implementation in
FY 89 and its subsequent hike by 2% in FY 97 has been at least 70% within the first two
months), which would be worth 30bp on a 5y swap, so the shorter end of the swap curve
has generally been extremely sensitive to the expected timing of a VAT increase.
Expectations for a VAT hike naturally weakened as business conditions deteriorated
following the 2007-08 global financial turmoil, but with the Japanese government moving
back to an expansionary fiscal policy and tax revenues currently lower than yearly JGB
issuance associated with the budget deficit, a VAT hike appears to be only a matter of
time. This risk is likely to be again priced into the JPY inflation swap curve.
Since 2007, JGBis started to trade on asset swap as spreads have been tighter versus Libor
relative to nominal asset swap spreads of similar maturities. Prior to 2008, long-dated
JGBi asset swaps were quoted at Libor-4bp to Libor-8bp, mid-market, with no significant
maturity differentials, a level about 12bp cheaper relative to nominal asset swaps. As has
been the case in other inflation-linked bond markets, the major asset price disruptions in
2008 left JGBis exceptionally cheap on an asset swap basis, with the long end of the JGBi
curve at times indicated at Libor+80bp. However, as the market’s cheapness gradually
corrected throughout 2009, asset swap margins normalized as well, currently hovering
around Libor+15bp to Libor+24bp across most issues.
Figure 47: Japanese linkers – historical performance and risk Figure 48: Return/risk linkers versus nominals and equities
5% 2
0% 1
-5% 0
-10% -1
-15% -2
2005 2006 2007 2008 2009 2005 2006 2007 2008 2009
15 March 2010 96
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Canada
Michael Pond Real Return Bonds (RRBs) were first issued by the Canadian government in December
+1 (212) 412 5051 1991, and there are currently five issues outstanding. Issuance has always been
michael.pond@barcap.com concentrated in the 30y sector to facilitate pension fund demand. As of February 2010,
with an adjusted principle amount of $34.2bn, RRBs made up 6.4% of total marketable
Canadian government debt, 9.5% of Canadian government bonds and 31% of those with
longer than 10 years to maturity.
The index is weighted to reflect typical spending patterns. The weights are determined
based on family expenditure surveys that are conducted periodically. The current weights
are based on the 2005 survey. As Figure 49 shows, the index comprises eight major
components. The component with the highest weight is the shelter component, which
includes owner-occupied and rented accommodation. The CPI includes only consumer
items and excludes personal income taxes, consumer savings and investments etc. The
index uses geometric means at the first-stage aggregation of collected price data, making
quality adjustments where possible. The fixed basket price index is an arithmetic average of
price relatives for all single commodities contained in the basket. The index attempts to
capture innovations in final prices, which include any changes in the Goods and Services
Tax as well as provincial retail sales taxes.
Alcoholic
beverages and Food
tobacco products 17%
3%
Recreation,
education and
reading Shelter
12% 27%
Health and
personal care
5% Household
Transportation operations,
20% Clothing and furnishings and
footwear equipment
5% 11%
Note: Reflects 2005 basket at April 2007 prices. Source: Statistics Canada
15 March 2010 97
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
40%
30%
20%
10%
0%
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Given its maturity profile and size, the Canadian inflation-linked bond market has been
largely the domain of pension funds. However, from time to time, international investors
have taken advantage of real yield differentials versus other more heavily traded
international markets such as the UK and US, and RRB issuance can have a valuation impact
on the long-end of other markets.
The Bank of Canada (BOC) acts on behalf of the Department of Finance for the purpose of
managing the financing program. The BOC currently operates under a quarterly funding
schedule with one 30-year RRB auction every three months. Similar to the US, RRB issuance
has been relatively constant despite a recent increase in total borrowing needs (Figure 51).
We expect the quarterly issuance pattern to continue as the government appears
committed to using the program as a cost-effective way to diversify its investor base. At
present, the Bank of Canada issues about C$500-700mn in RRBs each quarter, up from
C$300-400 before 2007.
15 March 2010 98
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
120.0
80.0
60.0
40.0
20.0
0.0
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Source: Bank of Canada
This new methodology became known as the “Canadian model”, and has been generally
followed by all subsequent major issuers, including all newly issued UK inflation-linked gilts. The
change in methodology allowed for simpler valuation and has assisted in the relative value
analysis of the product versus conventional bonds as well as cross-currency real yields. The
concepts of forward real yields and forward breakevens have become determining factors in the
relative valuation of international markets that have adopted this calculation method.
Calculation methodology
A reference CPI value is calculated for every day based upon the CPI values for three months
and two months prior to the month containing the settlement date. The reference CPI for the
first of each month is the index value of three months previously. The reference CPI for any day
during the month is calculated by linear interpolation. Unlike some other countries that issue
based on the Canadian model, RRBs do not have a par floor on the inflation adjusted principal.
(d − 1)
(CPI t −2 − CPI t −3 ) + CPI t −3
m
d = day of the month (eg, the 1st implies d=1)
The indexation factor is the reference CPI for the settlement date divided by the reference
CPI for the base date. Coupons are accrued on an actual/actual basis and paid semi-
annually. The gross settlement price is calculated as follows:
15 March 2010 99
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
CPI t
( p + c)( )
CPI base
c = real accrued
Real return bonds are taxable for residents but are not subject to withholding tax for non-
residents. For residents, RRBs’ income received and accrued is taxed in a given year while
the inflation accretion on the principal is also taxed. Capital gains are not taxed until
realised. For non-residents, the Canadian Treasury is not ordinarily required to withhold tax
from interest or principal paid on RRBs. However, the Treasury’s website provides more
detail on these conditions: www.fin.gc.ca/invest/taxtreat-e.html.
Demand for a Canadian inflation derivatives market is growing from pension funds, which
already hold most of the outstanding Canadian linkers, seeking exposure through receiving
inflation in swaps. Many Canadian investors have considered going to the US market as a
proxy for Canadian inflation due to the lack of an inflation swap market in Canada.
However, as the interest for Canadian inflation swaps grows, we look for a reasonable two-
way market to begin to develop. Despite the lack of government issuance, there appear to
be other parties interested in paying inflation to reduce the exposure they get through
governmental contracts or explicit inflation links.
Figure 52: Canada RRB – historical performance and risk Figure 53: Return/risk versus nominals and equities
5 IL
15% Nominal Bonds
4
10% 3
2
5%
1
0
0%
-1
-5% -2
1999 2001 2003 2005 2007 2009 1999 2001 2003 2005 2007 2009
Sweden
Stefan Liiceanu The Swedish government has been issuing inflation-linked bonds since 1994 and has a
+81 3 4530 1554 long-run target that this should make up 25% of central government debt including
stefan.liiceanu@barcap.com expected inflation accretion. With the share of linkers at this target and Swedish
government debt levels barely growing, issuance is extremely limited, but there is an
Chris Bettiss explicit government commitment to continue fostering the linker market. The non-
+44 (0) 20 7773 0836 government linker market remains embryonic, accounting for about 7% of total
chris.bettiss@barcap.com inflation-indexed bonds outstanding. The sector owning the majority of the market,
54% as of Q3 09, is insurance companies, which own linkers versus liabilities.
Sweden’s monetary policy is based on underlying inflation (CPIx) rather than the headline
CPI measure, in order to exclude mortgage interest payments and direct effects of changes
in indirect taxes and subsidies. From 1993, the Riksbank adopted a 2% annual inflation rate
target, with symmetrical tolerance of 1%. Official Swedish statistics also include the release
of HICP and this provides a consistent comparison to other European countries. On average,
there has been no bias between HICP and CPI over the long run, though the baskets differ
(Figure 55). For instance, HICP includes childcare costs and care for the elderly. Owner-
occupied housing costs carry a weight of around 8% in the CPI, but are not included in
Figure 54: Swedish CPI breakdown by major category, 2009 Figure 55: Swedish inflation – CPI, CPIx and HICP
Miscellaneo Food, 5%
Furnishing
us beverages &
& household 4%
6% tobacco
goods
6% 17% Communica 3%
tion 2%
3%
Housing & 1%
utilities Transportat
27% ion 0%
14%
-1%
Clothing & Restaurants -2%
footwear & hotels 1998 2000 2002 2004 2006 2008 2010
Recreation
6% Healthcare 6%
3% & culture Headline CPI Underlying Inflation (CPIx) HICP
12%
Source: Statistics Sweden Source: Statistics Sweden
HICP. Therefore, in a tightening cycle, this leaves a bias towards higher CPI than HICP and
CPIx, but CPI will tend to be lower than the other measures as mortgage rates fall. The
Swedish National Debt Office (SNDO) made it clear before the last euro referendum in
September 2003, that there would be no change to the measurement index for domestic
inflation-linked bonds, even if Sweden were to adopt the euro.
Meanwhile, seasonality in Swedish inflation is more extreme than in larger countries, resulting
into relatively volatile carry. Most Swedish electricity is generated by hydroelectric power,
causing headline CPI to have a higher sensitivity than normal to the weather, particularly very
cold winters or extended dry spells. Most Swedish linkers have very favourable seasonality into
redemption, redeeming at the start of December (for details please refer to Figure 57) and so
benefiting from September CPI, which is the most favourable month for Swedish CPI
seasonality (the September seasonal factor has been 1.0034 on average in the past decade, ie,
seasonality shows inflation higher by 0.34% higher in that month). The linker 1% April 2012
and the 0% coupon April 2014 bond, however, are issues that suffer from maturing at the
start of April, which means their final accrual is based on the very poor end-of-year and
January seasonal factors.
Figure 56: Swedish CPI monthly seasonal factors (NSA CPI/SA CPI ratio)
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Average 0.9936 0.9957 1.0013 1.0029 1.0032 1.0019 0.9976 0.9969 1.0030 1.0034 1.0010 0.9999
Min 0.9923 0.9947 0.9990 1.0016 1.0014 1.0011 0.9967 0.9961 1.0023 1.0025 0.9986 0.9990
Max 0.9947 0.9962 1.0026 1.0041 1.0053 1.0028 0.9988 0.9979 1.0036 1.0042 1.0041 1.0005
St. Dev. 0.0009 0.0006 0.0015 0.0009 0.0014 0.0006 0.0008 0.0008 0.0005 0.0007 0.0022 0.0006
Note: based on data sample from November 1999 to November 2009. Source: Statistics Sweden, Barclays Capital
The 5y breakeven is commonly quoted in the Riksbank’s quarterly Monetary Policy Report
as a measure of inflation expectations, although the 10y point on the curve has traditionally
been the more liquid sector of the linker curve. This is due to the 10y sector of the nominal
curve being relatively slow to develop, making the 10y breakeven a less accurate measure,
rather than because Swedish inflation is traditionally more volatile than in the euro area. In
practice, the Riksbank examines a range of maturities when assessing medium to long-term
inflation expectations, including the 5y5y forward breakeven.
Historically once can distinguish three separate regimes in the development of the Swedish
linker market. The first one, lasting roughly from 1994 until 1996 was a period when the
program was poorly understood by investors and auctions frequently went
4
“Ten years with inflation-linked bonds – a new asset class has been established,” SNDO, 2004.
undersubscribed. During the second phase, lasting from around 1997 to 2001, Sweden’s
inflation-linked government bond program underwent numerous reforms that helped
develop the market further. Lastly, from around 2002, the program reached a mature phase,
with linkers enjoying stable demand from investors while evolving into a full-fledged debt
management instrument for the issuing authorities.
3106 27-Sep-05 01-Apr-12 6.5 2.2 Yes 1.0 25,189 26,986 7.1
3001 01-Apr-94 01-Apr-14 20.0 4.2 No 0.0 4,017 4,924 22.6
3105 26-Apr-99 01-Dec-15 16.6 5.9 Yes 3.5 52,075 61,059 17.3
3102 01-Dec-95 01-Dec-20 25.0 10.9 No 4.0 42,992 52,692 22.6
3103 01-Dec-97 01-Dec-28 31.0 18.9 No 3.5 3 4 22.6
3104 19-Apr-99 01-Dec-28 29.6 18.9 Yes 3.5 44,783 52,509 17.3
Total 169,060 198,174 17.2
Note: Outstanding amounts as of end November 2009. Source: SNDO
In April 1994, upon consultations with the government and the central bank, the SNDO
decided to launch a program of government inflation-linked bonds and started with a zero-
coupon 20y instrument linked to the Swedish consumer price index, the SGIL 0% 4 Jan
2014 bond, auctioned Dutch-style (single price auction). The SNDO judged that a large
portion of the bids were at too low prices and consequently it issued only SEK1.2bn of the
bond versus original plans for SEK3.5bn. The lack of investor enthusiasm at this first auction
was a harbinger of limited acceptance of the new product in the following three years or so
and indeed, while the balance of linkers rose from SEK3.1bn in 1994 to SEK73.4bn in 1996,
at times the SNDO was forced to cancel auctions and breakeven inflation rates were on the
decline (please refer to Figure 62).
Figure 58: Amount outstanding of govt. inflation-linked Figure 59: Linker amounts sold at auctions and bid-to-cover
bonds and share in government debt portfolio ratios (quarterly)
Govt. inflation-linked bonds outstanding (LHS, SEK bn) Amount auctioned (LHS, SEK mn)
Share of linkers in central govt. debt (RHS) Bid-to-cover ratio (RHS, x)
Note: Inflation uplift included. Source: SNDO Source: SNDO, Barclays Capital
The long duration of the first bond (20y versus 8y for the longest-maturity nominal bond) is
one of the reasons why the reception for this new instrument was chilly and the decline in
spot inflation, from around 2.5% in mid-1994 to 0% by August 1996, also likely
contributed. To deal with this issue, in January 1995 the SNDO issued SEK500mn of a 9y
zero-coupon bond (0% April 2004 issue), a bond that it continued to sell throughout the
year in small quantities at weekly auctions but in this bond’s case too, it had to be
withdrawn from auction on more than one occasion.
The second stage of the Swedish inflation-linked government bond market proved to be
even more difficult as the CPI dipped into negative territory from September 1996 to May
1997 (average -0.4% y/y) and from July 1998 to February 1999 (average -0.7% y/y), due to
sharp falls in the owner-occupied housing sub-component of the CPI (mortgage interest
payments were on the decline as the Riksbank cut rates) and fundamental core CPI
disinflation. In addition, equity markets were overall bullish, central government finances
were stronger, leading some to doubt the SNDO’s long-term commitment to foster the
inflation market, and thus perceptions vis-à-vis the merits of linkers moved to the back
burner. However, the SNDO took a proactive and flexible stance, tackling the problem from
various angles. The linker market-related reforms implemented during this stage can
chronologically be summarized as follows:
1. The SNDO allowed private investors to purchase inflation-indexed bonds through primary
dealers after each auction (June 1995) and later in 1997 it introduced retail-oriented
inflation-indexed bonds (available to individuals and small companies and organizations).
Figure 60: Maturity comparison of bonds exchanged in Figure 61: Linker amounts sold in switch auctions and bid-
linker switch auctions to-cover ratios
35 14 9 4
30 12 8 3
25 10 7
3
6
20 8
5 2
15 6
4 2
10 4
3
5 2 1
2
0 0 1 1
0 0
05
05
06
08
09
07
06
07
p
p
ec
ct
ct
ct
ay
ar
Se
Se
O
08
09
05
05
06
07
06
07
M
D
p
p
ec
ct
ct
ct
ay
ar
Se
Se
O
O
M
D
M
Maturity of bond bought back (LHS)
Maturity of bond issued/tapped (LHS) Amount sold in the market (RHS, SEK bn)
Maturity ratio bonds offered/bonds bought back (RHS) Bid-to-cover ratio (LHS)
3. The addition of deflation floors on the principal payment to all inflation-linked bonds
that would be issued in the future (April 1999), a feature officially motivated by the need
for the “international harmonization 5” of the Swedish inflation-linked market, although
the deflationary experience was likely another objective reason behind this move. The
first bond with a deflation floor was sold in April 1999 (the 3.5% Dec 2028 bond), being
the destination of the 3.5% Dec 2028 non-floored bond issued in the previous year.
Thus, both bonds had the same coupon and maturity and terms were set at even yields.
A second switch auction was held five days later, enabling conversion from the
seasoned 0% 2014 bond into a new issue, the 3.5% Dec 2015 bond. In other words,
within the space of one year, the SNDO designed switch auctions both from non-floored
zero-coupon bonds to non-floored coupon-bearing bonds, and from non-floored bonds
to floored bonds. At present, of the SEK169.1bn of outstanding inflation-linked bonds
(without inflation uplift), approximately 75% is floored and the rest non-floored (see
Figure 63). Regarding the addition of deflation floors, we can make two noteworthy
observations. First, the adoption of floored bonds did not negatively impact the
performance of non-floored bonds but rather helped the overall stability of the market.
Indeed, as illustrated in Figure 62, around 1998-99, when Sweden experienced deflation,
breakevens turned negative (the 2y BEI reached nearly -50bp while the 10y BEI was just
18bp above zero) but following the issuance of floored bonds, BEIs were stable at around
2% in early 2004 despite the arrival of another deflationary episode. Second, the price
differential between floored and non-floored bonds of similar maturity suggests the
market has been consistently discounting the existence of the floor although it did not
attach too much value to it, not surprising given the Riksbank’s inflation target and the
bond’s generally long duration. Figure 64 illustrates the monthly average BEI spread
between the 3.5% Dec 2015 bond (floored issue) and the 0% April 2014 bond (non-
floored bond) as well as the evolution of spot inflation. Between May 1999 and November
2003, average inflation was 1.7% y/y, close to the central bank’s target and the average
spread between the two bonds’ BEIs was a low 1.3bp. From December 2003 to February
2006, inflation averaged a low 0.5% y/y and the BEI spread was 6.1bp on average. Again,
more recently, from August 2008 to November 2009, the average BEI spread between the
two bonds was 6.2bp versus an average realized inflation rate of 0.7% y/y.
5
“Swedish Debt Policy,” Erik Thedéen, February 2000.
-1
-2
Jan 97 Jan 99 Jan 01 Jan 03 Jan 05 Jan 07 Jan 09 Jan 11
4. Lastly, to further promote the inflation-linked bond market and especially increase investor
awareness, the SNDO introduced a new dealer and commission system from 2000. This
system was more demanding for the newly authorized inflation-linked bond dealers,
requiring that they submit detailed business plans about how they intend to broaden the
investor base. However, the other side of the coin was that levels of compensation were
higher – in any given year, these reflect dealer performance in the previous year, consisting of
a fixed payment of SEK1mn per dealer and a payment based on dealers’ share of the SNDO’s
activities in inflation-linked bond issuance, buybacks and switch auctions. Furthermore, the
authorized dealer status is valid for one year and the following year, new dealers can be
appointed if one or more of the previous year’s dealers fail to perform.
The third stage of the Swedish inflation-linked market has been one where investor
acceptance of the product matured and the asset class secured a stable place in the
country’s debt portfolio. During this third stage, the SNDO implemented other facilities to
further foster the market, including: 1) occasional buybacks to provide liquidity when
needed; 2) flexibility regarding the amount of linkers sold at each auction, with a target
interval rather than a fixed amount announced ahead of auctions; 3) secondary auctions
enabling dealers to buy an additional 20% of the allotted volume in the primary auction; 4)
linker repo for dealers up to SEK200mn at rates 25bp below the central bank’s overnight
rate; and 5) online linker sales to the retail sector permitting individuals to purchase directly
in the auction at the average yield level set after bidding.
The outstanding balance of linkers more than doubled from December 2001 to its recent
peak in September 2008, moving from SEK94.2bn to SEK227bn, and the share of linkers in
total government debt rose steadily (Figure 59). The SNDO has a long-run target that 25%
of its debt should be inflation-linked, once adjusting for expected future inflation accretion
(the 2009-2011 Central Government Debt Management Proposed Guidelines indicate that
“inflation-linked SEK debt should be steered in the long-term towards a percentage of 25
per cent of the central government debt”). With government debt levels relatively stable and
linkers’ share in the debt portfolio having been above the target level, the outstanding
amount of inflation indexed-bonds has been on the decline since 2007 (for example, in real
terms the size of the market was SEK192.14bn in January 2007 while in November 2009 it
was SEK169.1bn; the redemption of bond 3101 in December 2008 largely contributed to
this decline). With the debt stock in line with the long-run target at the start of 2010,
issuance was scheduled to be SEK10bn a year in 2010 and 2011, though this was revised
down to SEK9bn in the funding update published in March 2010.
Figure 63: Share of floored linkers in total Swedish govt. Figure 64: BEI spread of floored bond vs non-floored bond of
inflation-linked bond market similar maturity and spot inflation
70% 20 3
60% 15 2
50% 10
40% At the end of 2009, nearly 75% of 5 1
Swedish govt. inflation-indexed debt
30% 0 0
featured pricipal floor, with the rest
20% of 25% being non-floored . . . -5
-1
10% -10
0% -15 -2
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 99 00 01 02 03 04 05 06 07 08 09 10
Note: No inflation uplift included. Source: SNDO, Barclays Capital Note: BEI spread data are monthly averages of non-seasonally adjusted BEI spread.
Source: Statistics Sweden, Barclays Capital
Although important in themselves, liquidity measures in recent years have been less
instrumental that those implemented at the end of the 1990s, which clearly helped the
SNDO restructure the debt portfolio and stimulate domestic demand. In June 2002, after
reviewing the track record of switches and auctions held in the previous few years, the
SNDO concluded that “there is no strong need for further restructuring 6” given that most
inflation-indexed issues had reached satisfactory volumes and liquidity. Indeed, as
illustrated in Figure 66 since 2004 secondary market turnover of linkers has been above 4%,
at times reaching nearly 10% of total government bond turnover and in addition the repo
market has been functioning well.
On the demand side, the local insurance industry has been a key player, boosting its share
of total government linkers outstanding from around 35-40% in the early 2000s to above
50% in 2009 (Figure 66). The part-privatization of the national pension system and the
conversion of cash balances to government bonds at three state entities in 2002 (the
Swedish Nuclear Waste Fund, the Premium Pension Authority and the Deposit Guarantee
Board) also brought inflation-linked bonds closer to the spotlight. Furthermore, the explicit
indexation of the national pension scheme to wage inflation led to linker purchases by
several of Sweden’s national pension funds. For example, the financial statements of the
first three national funds (so-called AP1, AP2 and AP3) reveal that between 2004 and 2007
they owned, on average, some 14.5% of linkers outstanding. As mentioned above, the
SNDO has facilitated the purchase of this product for retail investors as well (ownership
around 5%).
6
“Central Government Borrowing: Forecast and Analysis,” SNDO, 18 June 2002.
Figure 65: Secondary market turnover of Swedish govt. Figure 66: Inflation-linked govt. bond holdings of Swedish
inflation-linked bonds and repo transactions insurance comps as percentage of market size
10% 100%
55%
8% 80%
50%
6% 60%
4% 40% 45%
2% 20% 40%
0% 0%
2004 2005 2006 2007 2008 2009 2010 35%
Note: Turnover represents linker spot transactions as % of aggregate spot Source: Statistics Sweden, SNDO, Barclays Capital
government bond turnover. Source: Riksbank, Barclays Capital
Minimum of [d − 1, 29]
(CPI t −2 − CPI t −3 ) + CPI t −3
30
This convention affects daily valuations, but all coupons are paid on the first of the month.
Interest accrues on a European 30/360 basis. Bonds issued since 1998 have deflation floors,
with the 2020 bond the only benchmark that does not. The settlement convention for
Swedish linkers is T+3. The clean nominal price (ie, after inflation accrual) of coupon bonds
is rounded to three decimal places before adding on accrued interest. The settlement price
is then rounded to the nearest krona.
While conceptually the well-defined bond market means that there is plenty of scope for a
liquid inflation swaps market to develop, particularly given the presence of supra-national
supply, the Swedish CPI swaps market still remains surprisingly underdeveloped. Quotes
remain infrequent across the curve, with the 10y the only maturity in which they are not
uncommon. However, as the liquidity of the nominal swaps curve has improved in recent
years, particularly at the long end where it was previously practically non-existent, there is
scope for the inflation swaps market to develop.
Figure 67: Swedish linkers – historical performance and risk Figure 68: Return/risk versus nominals and equities
10% 3
8% 2
6% 1
4% 0
2% -1
0% -2
1999 2001 2003 2005 2007 2009 1999 2001 2003 2005 2007 2009
Australia
Stefan Liiceanu The Australian government actively issued inflation-linked bonds (commonly called
+81 3 4530 1554 Treasury Indexed Bonds) from 1985 until 2003, when it suspended its TIB programme
stefan.liiceanu@barcap.com due to ongoing budget surpluses. In 2009, however, issuance resumed, and the size of
the market expanded by a sharp 66% to AUD10bn. The Australian inflation-linked
government bond market is thus poised to revive as the Commonwealth government
explicitly expressed its commitment for continued supply, possibly across several
maturity sectors. Besides the government IL bond market, in recent years several state
governments and infrastructure companies have issued inflation-indexed securities,
helping to fill the gap left by the government. Non-government inflation-indexed bonds
stood at AUD12.4bn at the end of 2009, slightly larger that the size of the government IL
market. Moreover, a significant local inflation swap market has been gathering pace,
supported by large institutional participants such as asset managers and pension funds.
Meanwhile, the CPI basket is divided into 11 major groups, each representing a set of goods
and services (Figure 69). These groups are further divided into 33 subgroups, and the
subgroups into a total of 90 expenditure classes. As is the case in other developed countries,
the composition of the basket and other features of the CPI are reviewed “from time to time to
ensure that it continues to meet community needs”. 8 The ABS undertakes these reviews at
approximately five-year intervals with the timing generally dependent on the availability of
results from the Household Expenditure Survey. The reference base period for the CPI is also
updated, but at less frequent intervals (currently the base is year 1989-90=100).
In the comprehensive review of CPI calculation methodology undertaken in 1998, the ABS
decided that the index would be modified from a measure of the change in living costs of
employee households to a general measure of price inflation for the household sector. As a
result, the population covered was expanded from wage and salary earning households to
include all metropolitan households. Weights were revised to reflect new expenditure
patterns and the expanded population coverage. More recently, the fifteenth series CPI was
introduced in September 2005, with item weights being revised in line with 2003-04 HES
expenditure patterns and “Financial & insurance Services” included in the basket 9.
7
The eight cities are the six state capital cities (Sydney, Melbourne, Brisbane, Adelaide, Perth, Hobart) plus Darwin and
Canberra. The ABS estimates that the individual consumer population in these cities represents about 64% of
Australian private households.
8
Australian Bureau of Statistics, “Australian Consumer Price Index: Concepts, Sources and Methods”, 2005.
9
For further details on the Australian CPI please refer to the Guide to the Consumer Price Index published by the ABS
(www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/6440.02005?OpenDocument).
Financial &
Insurance
Services, 9.31 Food, 15.44
Education, 2.73
Alcohol and
Recreation, 11.55 Tobacco, 6.79
Transportation,
13.11 Housing, 19.53
The Australian CPI is an important economic indicator not only for the bond market but also
for the central bank, which introduced inflation targeting in 1993. The Reserve Bank of
Australia’s monetary policy aims to achieve, over the medium term, a target for consumer
price inflation of 2-3% as the bank judges that preserving the value of money is the
“principal way in which monetary policy can help to form a sound basis for long-term
growth in the economy.” The central bank’s commitment to containing inflation helps
explain the fact that, with the exception of a brief period around the 2008 global financial
turmoil, Australian breakeven inflation rates have been trading around a long-run average
of 250bp (in Figure 70, note that the rebound from the 1% to the middle of the range in
effect over the past decade or so has also been swift).
3.0
2.5
2.0
1.5
1.0
0.5
0.0
96 97 98 99 00 01 02 03 04 05 06 07 08 09
Being released on a quarterly basis rather than on a monthly basis as is the case in the US, UK,
and the euro area, seasonality of the Australian CPI is less relevant. That said the index exhibits
seasonal patterns to a certain extent, as illustrated in Figure 72. The graph shows the average
q/q percentage change in the index over two periods – ie, 2000 through 2009 and, from a
longer-run perspective, 1990 through 2009. Inflation has averaged 2.8% y/y since 1990 (Figure
71), levels comparable with those in the US (2.9%) and the UK (3.3%). Prices tend to increase
most in the second and third quarters of the year, with Q3 recording the fastest pace of
consumer price appreciation. In contrast, Q4 exhibits the slowest increases in consumer prices.
Figure 71: The Australian CPI (y/y) Figure 72: Seasonality patterns – average q/q changes
14% 1.2%
1990-2009
12% 2000-2009
1.0%
10%
0.8%
8%
6% 0.6%
4%
0.4%
2%
0.2%
0%
-2% 0.0%
81 83 85 87 89 91 93 95 97 99 01 03 05 07 09 Q1 Q2 Q3 Q4
However, the government resumed its index-linked issuance programme in 1993, with
supply being “tailored to identify market demand”. 10 Between 1988 and 1993, the domestic
inflation market had become more sophisticated, helped by issuance by a number of state
governments and growing demand for long-term linkers from the emerging
superannuation or pension fund industry. Under the new programme, only capital-indexed
bonds were brought to the markets, all with long maturities. The size of linker supply picked
up considerably, increasing on average by about AUD640mn a year in 1993-2000, from
AUD1.6bn to AUD5.9bn (Figure 73). However, TIB liquidity was generally low as the bonds
were issued largely to buy-and-hold investors. Annual turnover averaged a modest
AUD12bn during 2001-06 versus AUD410.3bn for nominal Commonwealth Government
Securities (CGS). In other words, TIB trading volume was less than 3% of that of regular
government bonds despite the size of the total market being around 10% of the nominal
market 11. The number of TIBs issued was not large (the inflation-indexed programme never
10
Commonwealth Debt Management Report, 1996.
11
Australian Financial Markets Association, “2006 Australian Financial Markets Report.”
had more than five bonds), and additional supply came in the form of re-opening seasoned
bonds on average 5-6 times a year.
6 8%
6%
4
4%
2
2%
0 0%
86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Note: Mid-year levels except 2009, where Dec levels are shown. Source: Australian Office of Financial Management (AOFM)
Meanwhile, the Australian Treasury announced the suspension of the Treasury Indexed
Bond programme with the publication of its budget in 2003. This announcement followed a
one-year period of analysis and consultation with more than 90 domestic and foreign
market participants, which sought to determine whether the CGS market was a viable going
concern given the sharp fall in the Commonwealth government’s financing requirement
over the previous few years and the abundance of cash available from the sale of
government assets (the Australian government’s net debt had fallen from 18.5% of GDP in
1995-96 to 1.3% of GDP by 2004-05). On purely economic grounds, the very favourable
fiscal conditions implied no need for active issuance of government securities. On the other
hand, there was concern about banks’ disproportionately large role within the financial
markets and hence it was argued that the CGS market should be maintained. The
government’s review concluded that an interest rate market completely dominated by
banks and corporate paper would be vulnerable to economic shocks, thus posing significant
threats to financial stability and the accessibility of refinancing capital for corporates.
Because of this, the decision was taken to support government debt liquidity, and was
structured in such a way that it supported the 3y and 10y Treasury bond futures contracts.
Rather disappointingly, however, there was no room for a continuation of the Treasury
Indexed Bond Programme.
In 2009, however, the inflation-indexed government bond programme was revived after a
six-year hiatus. In an announcement regarding the issuance of Commonwealth government
bonds made public in May 2009, the Australian Office of Financial Management (AOFM)
stated that resumption of linker supply “could assist in the debt financing of long-term
infrastructure, since Treasury Indexed Bonds (TIB) would serve as both a pricing benchmark
and a risk management tool… indexed financing can be attractive for those infrastructure
projects whose revenues are linked to inflation… in addition, indexed instruments have
advantages for investors with inflation-linked liabilities”. One can think of at least two more
reasons for the resumption of inflation-indexed bonds not explicitly stated in the AOFM
note. First, similar to other countries affected by the 2007-09 global financial crisis, the
government’s expenditure base had increased sharply, while tax revenue plunged, leading
to Australia’s first budget deficit in seven years and the largest one on record (AUD57.6bn,
or nearly 5% of GDP in 2009). Second, double-digit negative returns on Australian
households’ mandatory superannuation (pension) funds owing to financial market volatility,
combined with proposals to raise the age at which individuals can access these funds, had
been steering debate on asset management in the direction of safer investment guidelines,
and linkers obviously fit the description.
Following consultation with various market participants, the AOFM decided to issue the
new linker at the beginning of October 2009 via syndication, with an announced supply size
of “at least” AUD1bn. The actual amount sold was four times larger, coming at AUD4bn. As
a result the Australian government inflation-linked bond market expanded sharply, by 66%
(besides the auction of the new 2025 issue, in mid-November 2009 the AOFM also issued
via an AUD300mn tap of the 4% Aug 2020 bond). The AOFM expressed its commitment for
additional issuance that would contribute to the benchmark status of the new linker (TIB
issuance is officially expected to be AUD1.5-2bn until the middle of 2010).
As shown in Figure 74, the maturity of the new bond is 16y, redeeming on 20 September
2025, the longest-dated government inflation-linked bond in the Australian market. The
AOFM’s notice added that “it is also planned, subject to investor demand, to extend the
Commonwealth inflation indexed yield curve beyond 2025 at an appropriate time”. Such a
move, however, would also require the launch of nominal bonds of similar maturity to
enable the observation of BEIs (currently the longest-dated nominal bond is the 5.75% 15
May 2021 issue). The structure of the new IL security is unchanged, being identical to the
three existing capital-indexed bonds: indexation is relative to the Weighted Average of Eight
Capital Cities – All Groups Index (the Australian CPI), both principal and coupon payments
have a deflation guarantee, and coupons are payable on a quarterly basis.
change in the Consumer Price Index over two quarters ending in the quarter, which is two
quarters prior to that in which the next interest payment falls”. This means that the bonds
have a six-month indexation lag compared to eight months in the UK. The interest on
Australian linkers is accrued on an actual/actual basis, and the bonds are quoted on a yield
basis. Interestingly, while conventional bonds pay semi-annual coupons, inflation-indexed
pay coupons on a quarterly basis. Australian linkers trade ex-dividend for seven days prior
to the payment date. Furthermore, as is the case with other developed linker markets, these
securities contain an embedded put at maturity that protects against deflation over the life
of the bond. Unlike other markets that offer an inflation floor, however, capital-indexed
bonds protect both coupon and principal from deflation over the life of the bond.
The calculation of interest and principal payments for Australian index-linked bonds is
significantly different than US, Canadian, Euro and Swedish bonds. The settlement price for
AUD100 face value of Australian government inflation-linked bonds is provided by the
following formula.
− V = 1/(1+i), with “i” being the annual percentage real yield (quoted real yield)
divided by 400. For example, if the annual yield is 4.65%, then “i” is equal to
4.65/400, or 0.0116.
− “f” is the number of days from the date of settlement to the next interest payment date.
− “d” is the number of days in the quarter ending on the next interest payment date.
− “g” is the fixed quarterly interest rate payable (equal to the annual fixed rate divided
by 4).
− “an” is the sum of the power series or V, with the highest power being “n” (the
number of full quarters between the next interest payment date and the date of
maturity). Mathematically,
− “Kt” is the nominal value of the principal at the next interest payment date (whether
or not there is an interest payment due). Kt can also be expressed as
Kt = Kt-1*(1+p/100), where Kt-1 is the cash value at the previous payment date. If
there has been no previous payment date, Kt-1 is equal to AUD100. Note that Kt and
Kt-1 are rounded to two decimal places.
− “p” is the average percentage change in the CPI over two quarters ending in the quarter
which is two quarters prior to that in which the next interest payment falls; for example
if the next interest payment is in November, then “p” is based on the average movement
in the CPI over the two quarters ended in the June quarter preceding. Mathematically
expressed, “p” is (100/2)*[(CPIt/CPIt-2)-1], where CPIt is the CPI for the second quarter
of the relevant two-quarter period, and CPIt-2 is the CPI for the quarter immediately
prior to the relevant two quarter period. “p” is also known as the “Australia CPI factor
average change (ACIF)” and is regularly calculated by the Reserve Bank of Australia
(RBA). These figures are also available on Bloomberg as ACIF Index.
Interest payments for Australian linkers are calculated as g*Kt/100, where “g” and “Kt” are
the variables defined in the calculation of settlement prices above. Interest payments are
rounded to the nearest cent (ie, 0.50 being rounded up). Moreover, no interest payment is
based on a nominal value of less than AUD100. If the nominal value of the principal falls to
below AUD100, then the interest payment would be based on a nominal value of AUD100.
Subsequent interest and/or principal payments will in such cases be reduced by the
difference between the fixed interest payment that was paid in the period and the payment
that would have been made under the above formula except for this provision.
Taxation
Up until 2009, income from Treasury Indexed Bonds derived through interest or discount or
through capital accruals throughout the life of the bonds was taxed according to the laws of
the Commonwealth and states. However, on 21 August 2009, the government announced it
will legislate to make Commonwealth Government Securities eligible for exemption from
non-resident interest withholding tax. This follows the removal of the interest withholding
tax on publicly issued corporate bonds some 10 years earlier and on state government
securities in 2008. The Treasurer of the Commonwealth of Australia argued that such a
move would closer align Australia’s tax treatment of securities to those systems in other
developed markets and that this would also lead to Australian bonds, TIBs included, being
perceived as more appealing by overseas investors.
More recently, the Treasury Corporation of Queensland issued AUD268mn of 15y capital-
indexed bonds in mid-2006, following the exact format of inflation-linked CGS. This was
followed by a total of AUD460mn worth of linker supply (maturities of 18y and 28y) at the end of
2007 by the Treasury of New South Wales. Illustrating robust investor demand for inflation-
indexed bonds in August 2009, the Treasury of New South Wales issued AUD800mn of a new
3.75% November 2020 capital indexed bond at an outright real yield of 3.75% (a spread of
nearly 60bp to the 2020 Commonwealth government TIB) on a book-build basis. Hints of such
pick-up in inflation supply could be seen in the entity’s 2008 Annual Report, commenting that
“the potential to expand the use of CPI linked debt by a broader range of clients, particularly for
PTEs (public trading enterprises) with revenue streams linked to the CPI, is currently being
explored… this may lead to further client demand for CPI linked funding”. The Queensland
Treasury Corporation, the third largest issuer of inflation-linked bonds in Australia, also
announced in its borrowing programme for 2009-10 that it would fund half of the estimated
deficit of AUD22.5bn via domestic bonds including the 2030 capital indexed bond.
Thus, while the Commonwealth Government ceased to provide liquidity to the Australian
inflation markets during 2003-2009, local government were quick to pick up the slack. Inflation-
indexed security supply related to infrastructure projects has also surged in recent years. This
type of issuance has been sporadic but has featured hefty deal sizes. Australian governments
responsible for building infrastructure projects have been turning to consortiums typically made
of construction companies and financial institutions (a recent example is the consortium
charged with the construction of nine public schools under the so-called “New Schools Privately
Financed Projects,” a public-private partnership). Once the project is finalised, the government
enters into a long-term lease contract with the consortium, with payments adjusted to the rate
of inflation. Therefore, the consortium has incentives to issue inflation-linked bonds with
maturities similar to that of the lease contract in order to achieve a better asset-liability profile.
As the projects feature amortising cash flows, in many cases the structure of these bonds is
Credit Foncier type, paying cash flows consisting of both principal and interest. When
demand from end-investors for such bonds is not strong enough, the consortium typically
issues only one part of the debt linked to inflation and the rest in nominal form with an
inflation outlay, ie, paying inflation via the swap market. A good example is Reliance Rail,
which at the end of 2006 financed the renewal of the Sydney rail network (a fleet of 78
trains) partly via AUD300mn worth of 29y CPI-linked annuity bonds and partly via
AUD1.6bn floating-rate bullet bonds with maturities between 9.75y and 14.75y; the nominal
bonds were issued along with paying CPI swaps to the tune of AUD1bn. The pick-up in non-
government inflation-linked bond issuance has thus clearly fostered the development of an
AUD inflation swap market. However, trading has been more or less sporadic, with activity
in the swap market spiking around the time of these large infrastructure projects.
Figure 75: Australian linkers – historical performance and risk Figure 76: Return/risk linkers versus nominals and equities
0% -4
1999 2001 2003 2005 2007 2009 1999 2001 2003 2005 2007 2009
Brazil
Marcelo Salomon Brazil has been issuing inflation-linked bonds since 1964. Having been the dominant
+55 (0) 11 5509 3295 form of local debt through hyperinflation periods, linkers fell to a small share of national
marcelo.salomon@barcap.com debt until 2003. Subsequently, there has been a major move towards increased linker
issuance to replace floating and foreign currency debt. At end-2009, Brazil was the
Guilherme Loureiro fourth-largest issuer after the US, UK and France, with the uplifted notional of IPCA-
+55 11 5509 3295 linked bonds equivalent to US$200bn.
guilherme.loureiro@barcap.com
Inflation targeting
The Brazilian inflation targeting (IT) regime has been consistently put under pressure
since it was implemented in 1999. However, it was during its tenth anniversary, when
the Brazilian economy weathered the global financial meltdown surprisingly well, that
we believe the regime finally reaped reward from its investment. For the first time the
Brazilian Central Bank (BCB) was able to implement countercyclical monetary policy,
helping to minimise the downturn of economic activity at a time when capital flows
were drying up dramatically and the BRL depreciated substantially. We believe this
action signals the way for smoother business cycles and that it is a new step in the real
interest rate convergence story in Brazil.
Figure 77 plots the IPCA rate (official inflation target rate, observed and forecast) along with
the bands and mid-point targets of the Brazilian IT regime since it was implemented. After a
volatile initial few years, reflecting both domestic and external shocks, inflation converged
towards the current 4.5% mid-point of the target, floating not only in the upper but also in
the lower range of the bands.
Each July the National Monetary Council (chosen by the Finance and Planning Ministers
along with the BCB President) sets the two-year ahead mid-point of the target along with
the tolerable range of fluctuation (currently at 4.5% +/- 2%). The BCB’s only mandate is to
implement monetary policy in line with these targets. It enjoys a significant degree of “de-
facto” (not “de-jure”) independence, with its governor being nominated by the Brazilian
President and the board of directors sanctioned by the Senate. There are no fixed mandates
for the board or the governors.
Figure 77: The Brazilian inflation targeting regime Figure 78: IPCA weights (%), January 2010
20 Housing
Comuni Food and
18 Goods
Forecast cation Beverages
16 4%
6% 23%
14 Clothing
12 7%
10 Education
7%
8
6 Personal
4 Services
Transport
10%
2 19%
0
Personal
Jan 99 Jan 01 Jan 03 Jan 05 Jan 07 Jan 09 Jan 11 Care Dwellings
IPCA (% Y/Y) Inflation target
11% 13%
Lower bound Upper bound
While the crisis crowned the Brazilian IT regime as a very successful framework of monetary
policymaking, it also introduced another layer of uncertainty: how will monetary authorities
deal with asset/credit bubbles? In our view the Brazilian IT will be enhanced by the use of
non-interest rate tightening mechanisms to rein in such bubbles. Specifically, further
prudential and regulatory measures should start to emerge as part of the BCB’s toolkit. In
addition, while the BCB is already a benchmark for regulatory and prudential regulation, we
believe it will continue to lead the pack by introducing new mechanisms to contain
exaggerated credit expansion.
Indexation
Brazil developed several inflation indices during its high inflation period. The most
important are IPCA (BZPIIPCA <Index> on Bloomberg) and IGPM (IBREIGPM <Index> on
Bloomberg). The IPCA (December 1993 = 100), calculated by the national statistics
agency (IBGE), is the official national consumer price index (and the measure of inflation
targeted by the central bank). IBGE publishes the IPCA around the 10th day of each
calendar month covering the period of the previous calendar month. In other words, the
February index, released in March, reflects average prices during the month of February,
and the m/m change for February represents the full-month February average compared
with the January full-month average.
Current weights are based on the consumption survey taken in 2002 and 2003 and reflect
the consumption patterns in households with incomes of 1-40x the minimum wage. As
standard in CPI index calculations, weights implicit in the index level are fixed. However,
IBGE-reported weights move slightly each month, reflecting the changes in relative prices.
Figure 78 shows the weights in January 2010, with food/beverages, transportation and
housing, having the largest share. The importance of the food and transportation
components is exacerbated by their high level of volatility, as they are affected by the
swings in international food and energy prices. Geographical coverage comprises 11
metropolitan areas and these indices are aggregated to form the national one, with São
Paulo having by far the biggest weight of 33%.
Figure 79: IPCA and IGP-M consumer price component (% m/m) Figure 80: IGP-M wholesale and consumer components (% y/y)
3.5% 50%
3.0%
40%
2.5%
2.0% 30%
1.5% 20%
1.0%
0.5% 10%
0.0% 0%
-0.5%
-10%
-1.0%
Jan 02 Jan 04 Jan 06 Jan 08
Jan 02 Jan 04 Jan 06 Jan 08
IGP-M Consumer Component
IGP-M Consumer Component IPCA IGP-M Wholesale Component
Source: IBGE, FGV, Barclays Capital Source: IBGE Source: FGV, Barclays Capital
The IGP-M Index (August 1994 = 100), published by the private Getulio Vargas Foundation,
measures a broader set of prices. It consists of three components: a measure of wholesale
prices (60% of the total); a measure of consumer prices (30% of the total); and a measure
of construction costs, both materials and labour (10% of the total). The IGP-M is also
published monthly, but around the 30th day of each month. Instead of calendar months, it
covers a period from the 21st day of a given month through to the 20th day of the following
month. The consumer price component of the IGP-M tends to behave similarly to the IPCA
but the wholesale price component is considerably more volatile, partly because of the
prevalence of raw foods in the index and partly because of exchange rate movements,
which can affect tradable goods prices significantly (Figure 80). As a result, inflation, as
measured by the IGP-M, fluctuates more widely than IPCA inflation. BRL changes have a
faster and higher pass-though into IGPM than into IPCA.
Both inflation indices It are updated only once a month and evolve as a step function. To
adjust the current price of the bonds correctly (see next section) one needs to account for
the accrued inflation from the last date the index was updated to the settlement date of the
bond (generally T+1). The market convention is to use the official ANBIMA’s (Brazilian
Association of Financial and Capital Market Companies) inflation forecast and pro-rata it to
define an index value, I ,so that we have:
t
I t' = I t (1+ i A )
n/ N
where n is the number of Brazilian business days between the evaluation date and the last
day of the previous IPCA period coverage; N is the number of Brazilian business days
between the last day of the previous IPCA period coverage and the last day of the next IPCA
period coverage; and i A is the inflation projected by ANBIMA. Special attention should be
paid to the dates when there are releases of other inflation indices, since they are generally
correlated with IPCA or IGPM or both. ANBIMA’s inflation projection might change
according to other data releases, affecting the value of the linkers.
Figure 81: NTN-B market information by maturity Figure 82: IPCA-linked debt duration and NTN-B (May15) yield
nov/11
mai/11
ago/12
mai/13
ago/14
mai/15
mai/17
ago/20
ago/24
ago/30
mai/35
ago/40
mai/45
ago/50
indexed to IPCA. They usually have semi-annual payments of fixed-rate coupons on the
indexed principal, though there have been some zero coupon bonds issued with indexed
principal. NTN-Cs are similar to NTN-Bs, but with principal indexed to IGP-M.
Until 2005, the market for NTN-Cs was substantially larger than that of the NTN-Bs.
However, demand for IPCA-linked securities has been increasing, given IPCA’s central role
within the inflation-targeting regime. Furthermore, from a supply perspective, the
combination of the 1999 devaluation and the high FX pass-through of the IGPM, led the
Treasury to shift towards IPCA-linkers. In the past few years, NTN-Cs issuance has been
extinguished and, as the end of February 2010, NTN-Bs account for about 90% of the total
outstanding amount of inflation-linked securities (BRL410bn), with the curve extending out
to 2050. Issuance in recent years has been twice monthly, with one auction a month
featuring all on-the-run issues and the second comprising issues up to the 10y benchmark.
In February 2010, the National Treasury issued three new NTN-B bonds maturing in August
2030, 2040 and 2050. The strategy was mainly to increase the maturity of the public debt,
to establish new benchmarks on the longer ends of the IPCA yield curve and to broaden the
alternatives of inflation liked bonds.
Liquidity is often poor for NTN-Cs, as a sizeable portion of the outstanding bonds are held
by buy-and-hold pension funds, which also have long-term IGPM liabilities acquired in the
past. NTN-Bs, on the other hand, trade currently (1st bimester of 2010) on average
US$1.1bn on a daily basis (Figure 82). NTNs are quoted on a yield basis using the Brazilian
Business/252 day count convention and annual compounding. All IPCA linkers have 6%
real coupons. However, due to the local convention, the effective coupon c’ (paid every 6m),
( )
1
is given by: c ' = 1 + c 2 − 1, where c is the annual coupon.
Linkers have their principal indexed from a base date that does not coincide with the
issuance date. The base date is set at 15 July 2000 to all NTN-Bs. Therefore, newly-issued
bonds start up with a large inflation adjustment and a nominal invoice payment necessary
to acquire the bond that is materially above par. The yield to price formula is given by:
I 't ⎡ n c' 1 ⎤
Pt = 1000 ∑ +
I 0 ⎣ i =1 (1 + Y ) (1 + Y )tn ⎥⎦
⎢ ti
Where t corresponds to the settlement date, ti refers to the period in years from the
settlement date to each coupon payment date and tn is the maturity of the bond in years,
( )
1
with the Brazilian Bus/252 convention used; c' = 1 + c 2 − 1 is the effective coupon, I is
0
the inflation index at base date, I t' is the current index level, Y is the quoted yield and Pt is
the current price of the bond.
fixed-income and inflation-linked component moved to 60% from around 20%. From 2005-
09, the average duration of the public debt increased to 3.5 years (from 2.8 years), while the
average (12-month) implicit cost of public debt declined from 14.2% to 9.4%.
The debt management gains affected more than just public finances. Reducing the FX/Selic
component of the debt enhanced the perception of fiscal solvency in periods of stress. In
previous crises, the stop-and-go cycle of foreign capital forced the depreciation of the BRL,
leading the BCB to start a tightening cycle to rein in inflation (both observed and expected)
and stem the process of currency depreciation. A larger share of fixed and inflation-linked
debt (smaller FX and Selic) broke the cycle where a weaker BRL and higher Selic rate would
lift the debt-to-GDP ratio, worsening fundamentals and further weakening BRL. Along with
large international reserves this helped to pave the way for counter-cyclical monetary
policymaking during the recent financial global meltdown.
Focusing on the evolution of the inflation-linked component of the debt, we note that its share
grew to nearly 27% of debt in 2009 from 10% in 2003 (Figure 83 and Figure 84). From 2005-
09, its average duration increased to 6.3 years from 5.8 (with the total debt maturing in more
than 5 years rising to 41% from 34%). In the same period, the IPCA-linked bond (NTN-B
maturing in May 2015) yield declined to 6.7% from 8.9% (Figure 82).
According to the 2010 Annual Borrowing Plan (ABP 2010), the Government’s goal is to
consolidate the share of fixed-income bonds and inflation-linked bonds at 55-65% of debt.
Although, in relative terms, inflation-linked debt should remain broadly unchanged, in
absolute terms we anticipate further issuances of inflation-linked bonds as the gross public
indebtedness should continue to expand in the next few years. The forecast of the ABP 2010 is
that the internal public debt expands to BRL1600-1730bn in 2010 (from BRL1497bn in 2009)
which is compatible with new net issuances of NTN-B reaching about BRL100bn in 2010.
Taxation
Local residents in Brazil pay a withholding tax of 15% or more on the income from bonds.
The tax rate ranges from 15-22.5%, with the precise bracket depending on the holding
period (22.5% if held for less than 180 days; 15% for periods above 720 days; and
intermediate rates for holding periods in between).
Figure 83: Inflation-linked bonds (% federal government Figure 84: Debt composition
debt) and target (annual borrowing plan)
35% 100%
30%
80%
25%
60%
20%
15% 40%
10%
20%
5%
0%
0%
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Source: National Treasury, Barclays Capital Source: National Treasury, Barclays Capital
Since early 2006, the withholding income tax rate applicable to sovereign local bonds has
been set to zero for foreign investors who are not in tax havens (see
www.receita.fazenda.gov.br for a list of these). Thus, qualifying foreign investors are exempt
from withholding taxes. Non-resident investors domiciled in tax havens are taxed at the
same rates as local investors.
It is important to point out that a zero tax rate is different from a non-existent tax. Any new
tax in Brazil has to be approved by Congress and the Senate, with all the political cost
associated with that. Therefore, it is much simpler for the Executive to raise a tax rate when
the tax already exists. The IOF tax on foreign capital flows is a good example of this. In
October 2009, the government raised it from 0% to 2%. The tax is now levied on all inflows
(equity and fixed income) except foreign direct investment and it is collected only on the
way in. There is no discrimination between long-term and short-term flows. This measure
was aimed at containing BRL’s appreciation as the government seemed concerned about an
eventual de-industrialisation process. At current levels of BRL we do not expect new FX
measures. Risks, however, increase if USD/BRL eventually starts to trade below 1.70.
Despite being relatively small in size and suffering from unequal taxation, new instruments
continue to emerge in the private fixed income market. One of them is the bank financial
note (Letra Financeira), which was just approved by the government and which should be
Figure 85: Non-government linkers breakdown Figure 86: Inflation-linked debentures (% total debentures)
4.2%
Others CD
11% 15%
4.0%
3.8%
3.6%
3.4%
3.2%
Debentures
74% 3.0%
Dec 05 Dec 06 Dec 07 Dec 08 Nov 09
formally regulated by the monetary council in early 2010. It is a new longer-term debenture-
like instrument that can be issued by banks, potentially representing a new investment
alternative for global and domestic investors. Once its details are announced and the first
issuances approved, we will be able to gauge the significance of this new instrument.
On the derivatives side, onshore inflation-linked swaps over IPCA and IGPM are available,
but with very limited liquidity and concentrated on short tenors (up to 2y). Longer tenors
are available, albeit with very large bid/ask spreads or coinciding with the maturity dates of
bonds. These swaps can be registered in CETIP (OTC) or BM&F (the main local exchange)
and are usually traded as zero coupon, but can also be coupon bearing. The leg of the swap
linked to inflation pays the changes in the inflation index plus a real rate coupon that is
quoted at the onset of the trade. The other leg of the swap is usually the accumulated
overnight rate (CDI), as is also the case for the IPCA futures contract (WLA <Index> on
Bloomberg), but it can also be a fixed Libor plus spread or other formats.
Offshore total return swaps are common investment vehicles for foreign investors looking
to circumvent the burden of opening and managing local investment accounts in Brazil.
Dealers with an onshore presence buy the bonds on their books and pass the total return to
investors offshore though an ISDA swap. The funding leg of the swap may be the local
overnight rate CDI or Libor, in case the client wants to keep the FX exposure along with the
local interest rate exposure. The cost of the pass-through service charge by the dealers has
declined to a few basis points, while the reimplementation of the IOF tax has increased
interest from international investors in total return swaps. Regular offshore swaps against
CDI or Libor may also be found on a limited basis and with fairly wide bid/ask spreads.
Since the local inflation swap market is not well developed, the few dealers quoting the
offshore swap need duration hedges using the local linkers, leaving their books with basis
risk and cash flow mismatches.
Figure 87: Brazilian linkers – historical performance and risk Figure 88: Return/risk – linkers versus nominals and equities
Mexico
Roberto Melzi Mexico has been issuing inflation-linked bonds for over 30 years, with around 15% of
+52 55 5241 3260 outstanding government debt denominated in the inflation-linked UDI Index at the end
roberto.melzi@barcap.com of 2009 (UDI98bn, c.USD16bn equivalent). The UDI is unusual in that it fixes off a CPI
Index that is published twice monthly rather than monthly. Pension funds are the
Jimena Zuniga largest holders of government UDIBonos, with foreigners holding less than 3% as at the
+1 (212) 412 5361 end of 2009, despite no restrictions or withholding tax for international investors.
jimena.zuniga@barcap.com
Indexation
Mexico’s current inflation-linked bond market started in May 1996 as a result of the ‘Tequila
crisis’ in late-1994, as higher inflation led to accelerated amortisation of loans in real terms.
This created an incentive to issue credit in UDI (‘Unidades de Inversion’) to preserve their
real value. The central bank (Banxico) adopted the Inflation Targeting (IT) regime in 1999
with an initial 13% aim for that year, but with the objective of bringing inflation down to 3%
by 2003 and beyond. The latter has remained the target with a tolerance range of +/- 1%.
The CPI (‘Indice Nacional de Precios al Consumidor’ – INPC) is calculated bi-weekly by Banxico,
which releases the data on its web page (www.banxico.org.mx) in accordance with an annual
calendar. The release occurs one day before publication in the official gazette, which is due on
the 10th and 25th day of each month (or previous day if not a business day) for the previous
half-month period. The CPI considers fixed weights and is based on the consumption basket of
2000. Price information is collected in 46 cities and metropolitan areas. The housing component
has the heaviest weight in the index, followed by the food component. Banxico also reports a
breakdown of the headline index into core (69.6% of the total) and non-core (30.4% of the total)
items. While core inflation is fairly stable in time, non-core prices inject considerable volatility into
the index, particularly those of perishable food items.
The UDI Index (MUDI <Index> on Bloomberg) is released twice a month by Banxico, it is a
function of bi-weekly inflation and has been published since 4 April, 1995. By day 10 of each
month, Banxico publishes index values for the period between days 11 and 25 of the
previous calendar month. On day 25, it publishes values for the period between day 26 of
the previous month and day 10 of next month. In each period, the UDI Index changes by the
Other
6.9% Food 20
Educ.
11.5% 22.7% 18
16
14
12
Transport Clothes 10
13.4% 5.6% 8
6
Health 4
8.6% 2
Housing 0
Furniture
26.4% 1999 2001 2003 2005 2007 2009
4.9%
daily geometric equivalent of the corresponding bi-weekly inflation rate according to:
UDI t = UDI t −1 × (1 + π )1 / n
where π is the most recent reported bi-weekly inflation rate and n is the number of days
during the period to which the inflation rate corresponds.
UDIBonos are quoted in real yields, with a typical bid-ask spread of 5bp. Liquidity as of
January 2010 is concentrated in the on-the-run Jun12 and Nov35 issues, while the 10y on-
the-run Jun19 is less actively traded. The average trade ticket is MXN100mn. Given the
‘bullet’ structures of UDIBonos, the real yield calculation is simply the “yield to maturity” of
the bond quoted in UDI. The total return will depend on the realisation of the UDI Index,
which will affect the interest accrued and principal of the bond. Hence, other than the half-
monthly inflation periods and short lag, conceptually the calculations are similar to those of
the Canadian-model.
Domestic investors hold 94% of the total outstanding amount of UDIBonos, with local
pension funds being the single key holders (52% of the total). Foreign investors, who are
not subject to withholding taxes or any other local Mexican taxes on purchasing
government securities, hold just 3% of the outstanding amount of these bonds.
Figure 92: Mexican linkers: historical performance and risk Figure 93: Return/risk of linkers versus nominals and equities
10% 2
8% 1
6%
0
4%
-1
2%
0% -2
2004 2005 2006 2007 2008 2009 2004 2005 2006 2007 2008 2009
Argentina
Guillermo Mondino Argentinian inflation linkers are an endangered species. Inflation under-reporting has
+1 212 412 7961 significantly reduced liquidity in this market and the government no longers issues. In
guillermo.mondino@barcap.com addition, in 2009, a series of swaps aimed to extend maturity and switch to nominal
bonds also reduced the amount outstanding.
Sebastian Vargas
+54 (0) 114850 1230
sebastian.vargas@barcap.com
CER Inflation-linked Index
Inflation-linked bonds issued were issued in 2002 for the first time in the recent past and are
indexed to consumer prices through the CER Index (Coeficiente de Estabilización de
Referencia). The CER Index, launched in February 2002, is published daily by the central
bank and is calculated using the geometric mean of the changes in the consumer price
index (CPI) with a one-month lag. The CPI is calculated by the Instituto Nacional de
Estadistica y Censos (INDEC), while the CER is calculated by the central bank.
The CPI methodology was changed in May 2008 to introduce a new CPI (base 2008) to
replace the old CPI (base 1999). The new methodology was introduced in the midst of
controversies over inflation under-reporting by the INDEC. This new CPI replaced the old
index, which was discontinued. The chosen basket is intended to represent the
expenditures structure of the population considered in the National Survey of Household
Expenditure (ENGH) of 2004-05. Despite the government’s efforts, the new methodology
was unable to restore confidence in the new CPI. The wedge between genuine inflation and
the CPI inflation is still significant.
The index is broken down into the following nine groups (weights in parentheses): food and
beverages (37.9%); transportation and communications (16.6%); housing and basic
services (12.1%); medical attention and healthcare expenses (5.6%); leisure (5.1%);
household equipment and maintenance (4.9%); apparel (7.3%); other goods and services
(6.3%); and education (4.3%). In the first 10 working days of each month of the year, the
INDEC makes public the index corresponding to the previous month.
Since 2007, there has been significant controversy over the CPI Index and, hence the CER.
There is a widely-held belief by investors that inflation is being understated. This has,
among other negative consequences, significantly affected both domestic and global
demand for inflation-linked bonds. Recent promises to restore INDEC credibility made by
Minister Boudou have not yet materialised and the development of this asset class in
Argentina seems highly dependent on future political developments.
It is important to note that the Argentine government stopped issuing inflation-linked debt
in 2005, which, coupled with the ongoing controversy about the measurement of the
Consumer Price Index, has detracted from the appeal of this market among institutional and
other long-term investors at whom these assets were originally targeted. In addition, the
nationalisation of pension funds in October 2008 will potentially reduce the traditional
demand for this kind of instruments. In addition, two swaps performed during 2008 aimed
to extend maturity and switch currency denomination (to nominal bonds) have reduced the
amount outstanding of CER denominated debt even further.
Bonos del estado nacional are the BODEN (CER-linked) and BONAR (USD- and ARS-
denominated) bonds. BODEN were largely issued to compensate for losses incurred by
individuals and financial institutions as a result of the compulsory peso conversion adopted
by the government during the 2002 economic crisis. Banks and depositors were
compensated with BODEN 2007 and BODEN 2012, while public sector workers and
pensioners received BODEN 2008. The central bank received BODEN 2011 and BODEN
2013. Pension funds were compensated with BODEN 14.
BODEN 2011 CER 30 Apr 03 30 Apr 11 Sinking fund 2%* Monthly; 30/360 Am: 84 equal monthly installments
(first 83 of 1.19%, last 1.23%), starting
5/30/04
BODEN 2014 CER 30 Sep 04 30 Sep 14 Sinking fund 2%*(initial CER = Am: 8 S/A equal installments of 12.5%,
1.5178) S/A; 30/360 starting 3/31/11
Note: *Principal is adjusted for inflation using CER Index (T-10 business days)/Initial CER (note this varies for the different BODENs). Source: Mecon
In February 2005, Argentina extended a global exchange offer to the holders of its defaulted
debt. Of the $82bn of eligible debt, about 76% was tendered by holders, who, in exchange,
received PAR, DISCOUNT, or QUASI-PAR bonds denominated in ARS, USD, EUR and JPY.
DISCOUNT and QUASI-PAR bonds carried a haircut of 66.3% and 30.1%, respectively. All of
these bonds incorporated a GDP-linked unit (or GDP warrant, as is usually referred to in the
market), which began to trade separately in November 2005. The government also issued
PAR/DISC and QUASI PAR regulated by the Argentine local law and linked to inflation. Both
the PAR and the QUASI PAR were targeted at long-term local investors. As a result, trading
in those bonds is less liquid than in the DISC (particularly in the QUASI PAR, which was
customised for buy-and-hold private pension funds).
PAR CER 31 Dec 03 31 Dec 38 Step-up cpn/sinking 0.63% first 5yrs, 1.18% Am: 20 equal S/A instalments
fund 6-15yrs, 1.77% 16- starting 6/30/29
25yrs, 2.48%
thereafter*. S/A;
30/360
DISC CER 31 Dec 03 31 Dec 33 Step-up cpn/sinking 2.79% first 5yrs, 4.06% Am: 20 equal S/A installments,
fund/capitalised 6-10yrs, 5.83% starting 9/30/24 - Cap: 3.04% first
thereafter*. S/A; 5yrs, 1.77% 6-10yrs, 0% thereafter
30/360*
QUASI CER 31 Dec 03 31 Dec 45 Step-up cpn/sinking 0% in the first 10yrs, Am: 20 equal S/A installments,
PAR fund/capitalised 3.31% thereafter*. starting 6/30/34. - Cap: interest fully
S/A; 30/360 capitalised in the first 10yrs
Note: *Principal is adjusted for inflation using CER Index (T-10 business days)/Initial CER (1.4549). Source: Mecon
An earlier series of BOCON was issued as compensation to families of victims who were
jailed or disappeared during the military dictatorship. Re-openings of BOCON
subsequently took place opportunistically for amounts within those originally authorised
in 2002, or for additional amounts authorised through new decrees. Individual re-
openings are not reported. The following table summarises the structure of some of the
most liquid BOCON issues:
PRE 08 CER 3 Feb 02 3 Jan 10 Sinking fund/ 2%* monthly; 30/360 Am: 48 equal monthly instalments.
capitalised The first 47 of 2.08% the last ones of
2.24%, starting: 2/3/06 – Cap: thru
1/3/06
PRE 09 CER 15 Mar 04 15 Mar 14 Sinking fund 2%* monthly; 30/360 Am: 72 equal monthly instalments of
1.35% except the last two of 2.75%,
starting: 4/15/08
PRO 11 CER 3 Feb 02 3 Dec 10 Sinking fund 2%* monthly; 30/360 Am: 106 equal monthly instalments
the first 105 of 0.95% the last one of
0.25%, starting: 3/3/02
PRO 12 CER 3 Feb 02 3 Jan 16 Sinking fund/ 2%* monthly; 30/360 Am: 120 equal instalments. The first
capitalised 119 of 0.84% the last one of 0.04% -
Cap: thru 1/3/06
PRO 13 CER 15 Mar 04 15 Mar 24 Sinking fund 2%* monthly; 30/360 Am: 120 equal monthly instalments of
0,83% except the last one of 1.23%,
starting: 4/15/14
Note: *Principal is adjusted for inflation using CER index (T-10 business days). Source: Mecon
Issued by the Fondo Fiduciario de Desarrollo Provincial (a trust), BOGAR bonds were used
to restructure the debt of a number of provinces. Payment is secured by a government
guarantee, which, in turn, is secured by a pledge of up to 15% of the province’s share in
shared tax revenues and, beyond that ceiling, is additionally guaranteed by the central
government, the financial intermediation tax and remaining fiscal resources (net of what
corresponds to the state-managed social security system).
BOGAR18 ARS 4 Feb 02 4 Feb 18 Sinking fund 2%* monthly; Am: 156 monthly rising payments, starting
Actual/365 3/4/05 - Cap: thru 9/4/02
BOGAR20 ARS 4 Feb 02 4 Oct 20 Partially capitalised 2%* monthly; Partially capitalised: first 3yrs of the 2% cpn:
Actual/365 60% will be paid in cash, 10% will be
capitalised and 30% represents the haircut.
From 2/4/05-8/4/05 the 2% coupon will be
capitalised. Paid in cash thereafter
Note: *Principal is adjusted for inflation using CER index (T-5 business days)/Initial CER (0.9999). Source: Mecon
CER are units of account, whose value in pesos is indexed to Argentine CPI. The CER index
was fixed as 1 ARS on 2 February 2002 and tracks the Argentine CPI with a one-month lag:
(
a) Ft = CPI j − 2 CPI j − 3 )
1/ k
for days 1-6 of each month, the CER is based on the geometric
mean of CPI variation, between the second and third month previous to the current month;
(
b) Ft = CPI j −1 CPI j − 2 )
1/ k
for days 6 to the last day of each month, the CER is based on
the geometric mean of CPI variation during the previous month:
where k is the number of days in the current month and j is the current month.
The CER time series can be viewed on Bloomberg using the code ACERCER <Index>.
To calculate the real yield to maturity for CER-linked bonds, the first step is to set up the
entire real cash flows structure (including amortisation and capitalisation). Since these real
cash flows are essentially deterministic to calculate the real yield to maturity, it is sufficient
to solve for R in the equation:
n
c(t )
Pr ice = ∑
t =1 (1 + R / q )
qt
Where c(t ) is the cash flow at time t (in years) and R is the real yield to maturity q -
compounded.
The only complication in this formula is to adjust the quoted all-in peso market price to take
into account the change in the CER Index:
Mkt Pr ice
Pr ice =
CERT − d / CER0
where Mkt Pr ice is the peso price quoted, CERT − d is the index level for T-d (where T is
the payment and d specific number of days) and is the base index level (fixed at inception of
the specific bond). The following table summarises the change in CER calculation for the
most liquid CER-linked bonds:
Figure 99: Argentinian linkers – historical performance and risk Figure 100: Return/risk of linkers versus nominals and
equities
100% 2
50% 1
0% 0
-50% -1
-100% -2
2004 2005 2006 2007 2008 2009 2004 2005 2006 2007 2008 2009
Chile
Roberto Melzi Few countries have indexation embedded within their financial markets as much as
+52 55 5241 3260 Chile. Most domestic debt is inflation indexed, using the CPI-linked UF unit, while UF-
roberto.melzi@barcap.com linked derivatives are also traded. The largest issuer is the central bank, but the
government also issues debt using the central bank as agent, with a combined uplifted
Jimena Zuniga notional size of CLP8.57tn ($16.9bn) at the end of 2009.
+1 (212) 412 5361
jimena.zuniga@barcap.com
Indexation
Chile has a long tradition with price indexation. It is present not only in financial markets,
(including the vast majority of mortgages), but also in labour and house rental contracts.
This fact partly explains why Chile escaped from the widespread dollarisation observed in
other Latin American economies that, like Chile in the 1970s, suffered from very high
inflation. CPI-indexed bonds were first issued in 1966 by the central bank, although indexed
deposits to both wages and CPI occured as far back as 1959.
The central bank runs an inflation-targeting regime, with a 3% operational objective (in a +/-1%
tolerance range) within a two-year policy horizon. Being a highly open economy with few
regulated prices, international developments quickly transmit into domestic prices. This is
particularly noticeable with oil prices (despite the operation of a stabilisation fund) and with
foodstuff, which are related to soft commodity prices such as grains and milk. Despite CPI
inflation volatility (mainly due to food and fuel prices), the central bank enjoys ample credibility,
with survey inflation expectations well-anchored at 3% over the 2y policy horizon.
The CPI is calculated by the National Institute of Statistics (INE), which releases CPI for the
previous month during the first eight days of the month within a pre-announced annual
calendar. Since February 2010, the index is based on a nationwide geographical coverage
including all the region’s capital cities (as opposed to the previous’ index focus on Gran
Santiago only). The new CPI also adopted the entire average of 2009 prices as a base, in
contrast with the previous index, whose base was one month (December 2008). CPI
releases are not subject to revisions nor are they seasonally adjusted. Since 1967, indexation
has been almost exclusively based on the ‘Unidad de Fomento’ (UF), with daily re-
adjustments based on the previous month’s CPI inflation prevailing since 1977.
Education, 10
Food, 18.9
6.0 8
Entertain., Drinks, 2.0
6
7.5 Clothing, 4
Com munic., 5.2
2
4.7
Housing, 0
13.3 -2
Transport,
19.3 Furniture, -4
Health, 5.4
7.5 2000 2002 2004 2006 2008
Inflation target mid-point CPI % y/y
Source: Barclays Capital Source: Central bank, Barclays Capital
The UF Index (CHUF <Index > on Bloomberg) is calculated and published by the central
bank, with CPI released early in each month to enable continuous calculation of the UF as
each release fixes the index to the 9th of the following month. The same readjustment factor
is used between the 10th day of the current month and the 9th day of the following in the
daily equivalent of inflation calculated geometrically. In other words, for day t between the
10th of month m and 9th of month m+1:
where d is the number of calendar days between those two dates and round (π m −1 ,1) is
the monthly CPI inflation of month m-1 rounded to the first decimal point.
In addition, the government (with the central bank acting as its agent) issues BTUs (‘Bonos
de la Tesoreria General de la Republica en UF’; CHILBT <Govt> on Bloomberg), UF-
denominated standard bullet bonds. Government supply of debt in general depends on its
fiscal situation and is contingent on the ‘structural balance’ rule. When the economy is
growing below/above potential the government runs a cyclical/temporary deficit/surplus,
implying higher/lower net issuance of debt (including BTUs). Given that the financial crisis
increased the need to finance higher deficits in coming years, some additional supply from
the Treasury could materialise. Yet, we do not foresee its debt growing significantly given
the structural balance rule. Thus, the central bank is likely to continue to have more
outstanding debt paper than the Treasury.
BCUs are more liquid than BTUs, with the market for both being dominated mainly by local
pension funds and banks. Foreign investors do not participate much due to taxation issues
and the complexity of trading them (a local custodian account is needed). The average
ticket size is UF100k, with a typical bid/ask spread of 2bp and an average traded volume
(which is quite volatile, depending on pension funds participation) of around UF2m per day.
Taxation
Chile has a general flat 4% tax rate for interest earned by foreigners in the domestic market.
This rate applies to the BCU rate, not to earnings due to indexation (in general the Chilean
tax system considers tax bases that are CPI deflated). Capital gains, however, have a more
cumbersome treatment. The general rule for bonds is that they pay the general income tax,
which is 35% for foreigners. However, if the investor is domiciled in a country with which
Chile has a double-taxation treaty, the tax is zero.
Regarding derivatives, two main types of swaps are traded OTC. The UF/Camara swap is a
fixed-for-floating contract with a counterparty paying/receiving a fixed UF rate and the
other receiving/paying a floating nominal rate that depends on the ‘Indice de Camara
Promedio’ (ICP). The ICP Index represents the funding cost of local financial institutions and
is published daily by Chile’s banking association (www.abif.cl). The index depends on the
‘Tasa Camara Interbancaria Promedio’, which is the average O/N interbank interest rate set
by the central bank.
The other key swap traded OTC is the UF/Libor, an offshore, fixed-for-floating cross-
currency swap, where a counterparty pays/receives a fixed UF rate and the other
receives/pays the 6mth USD Libor floating rate. In both swaps, coupons are exchanged
semi-annually in an ACT/360 day-count convention, with notional amounts exchanged at
the start of the swaps. Finally, it is also possible to trade floating real rate versus fixed real
rate swaps, with the floating real rate determined by deflating nominal ICP with the UF
Index. Activity is mainly concentrated in the interbank market, with foreign hedge funds
participating comparatively more in the nominal swap market. Liquidity is concentrated in
the 1y, 5y and 10y segments of the curve (typical bid/ask of 5bp/10bp), although quotes
can be obtained up to 20y. Local corporations use these markets for liability management.
Finally, there is also a liquid market of inflation rate forwards (up to 1y), where investors can
trade their views on inflation rates in the future directly. In combination with positions in nominal
swaps, these ones provide a means to synthetically trade UF-denominated swap rates.
Figure 104: Chilean linkers – historical performance and risk Figure 105: Return/risk of linkers versus equities
8% 2
6% 1
4% 0
Equity
2% -1
IL
0% -2
2003 2004 2005 2006 2007 2008 2009 2003 2004 2005 2006 2007 2008 2009
Colombia
Roberto Melzi Inflation-linked bonds were first issued in 1967 and the market has grown steadily since
+52 55 5241 3260 then. Around a quarter of government debt is linked to CPI via the UVR indexing unit,
roberto.melzi@barcap.com but taxation limits interest from international investors.
Jimena Zuniga
+1 (212) 412 5361
Indexation
jimena.zuniga@barcap.com The central bank (Banrep) conducts monetary policy within an inflation-targeting framework.
In October 2009, Banrep set the 2010 inflation target to be the long term-goal of 3% (in a +/-
1% tolerance range), culminating a successful process of gradual target reductions and
inflation convergence since the inception of the inflation-targeting regime. Although inflation
averaged 6.2% in 2001-08, it plummeted to 2.0% by December 2009 and is widely expected
to end 2010 within the central bank's new target range. As in other emerging countries, food
prices are critical drivers of inflation volatility. Energy prices are relatively heavily regulated and
typically reflect international prices more smoothly and with a lag.
Since August 2000, the inflation indexing unit has been the ‘Unidad de Valor Real’ (UVR
<Index> on Bloomberg), which is calculated/published by Banrep based on DANE’s CPI
inflation rate for the previous month. The UVR Index is adjusted daily between the 16th day
of the current month and the 15th day of the following month. For any day t within this
period, the UVR is given by:
where
UVR15 is the index value in day 15 of the previous month,
π m −1 is the previous
month’s m/m inflation rate and d is the number of days in the period.
Others 11
Communic.
6%
4% 10
Food
28% 9
Transport.
15% 8
7
Entertain. 6
3% 5
4
Education
6% 3
2
Health Apparel Housing
2% 5% 31% 2000 2002 2004 2006 2008
Inflation target mid-point CPI % y/y
Source: DANE, Barclays Capital Source: Central bank, Barclays Capital
The UVR/Libor OTC is a swap that used to trade but is practically non-existent now. This swap
was an offshore fix-for-floating cross-currency swap, where a counterparty pays/receives a
fixed UVR rate (semi-annual, Act/360) and the other receives/pays 6mth USD Libor floating.
When and if prices are provided, bid/ask spreads are very wide. When inflation-linked
corporate deals occur, activity in this market rises, although this is sporadic.
Taxation
Taxation issues limit foreign investors’ participation in the UVR-linked bond market as they
are subject to withholding tax on both income and capital gains. The rate ranges from 7% if
the bond matures within five years to 4% if it has a longer maturity. In addition, UVR bonds
are subject to a 0.4% financial transaction tax.
Figure 108: Linkers – historical performance and risk Figure 109: Return/risk of linkers versus nominals and equities
20% 3
15% 2
10% 1
5% 0
0% -1
2003 2004 2005 2006 2007 2008 2009 2003 2004 2005 2006 2007 2008 2009
Uruguay
Sebastian Vargas Uruguay resorted to a CPI-linked market in the aftermath of the 2002 financial crisis, as
+54 (0) 114850 1230 a means of creating a local currency market in the context of widespread financial
sebastian.vargas@barcap.com dollarisation and with both high and uncertain inflation expectations. Owing to the
government’s effort, the market expansion has been sustained.
The creation of an inflation linked bond market is part of a consistent effort to provide a
dollar substitute and is continuing at a growing speed. To foster this market, in 2002 the
Uruguayan government replaced the old Unidad Reajustable (UR), which was adjusted with
a wage index, with a newly-created Unidad Indexada (UI), which was adjusted with the CPI.
The UI was introduced in June 2002, with its level set at 1 Uruguayan peso per UI, and is
calculated and published by the INE. The level of UI is calculated on a daily basis from the sixth
day of the month up to the fifth day of the following month, using CPI levels from the previous
month. Specifically, the daily factor UId,,M is computed according to the following formulae:
d + DM −1 − 5
⎛ CPI M − 2 ⎞ D M −1
a) UI
d ,M = UI 5, M −1 ⎜⎜ ⎟⎟ from the first day of month M up to the fifth day;
⎝ CPI M −3 ⎠
d −5
⎛ CPI M −1 ⎞ DM
b) UI
d ,M = UI 5, M ⎜⎜ ⎟⎟ from the sixth day of month M to the end of the month,
⎝ CPI M − 2 ⎠
where DM represents the amount of calendar days in month M and CPIM corresponds to the
CPI level in month M.
Inflation-linked bonds
Since the creation of the UI, the government has issued several types of bonds linked to the
index, in particular Letras, Notas BCU and Notas del Tesoro, with initial maturities ranging
between three and 10 years. Recently, there has been notably larger supply in UI-
denominated global bonds whose cash flows are UI-linked but paid in USD. Three
benchmarks have been issued since 2006 with 2018, 2027 and 2037 maturities and now
make up almost half of the government’s inflation-linked debt.
Israel
Arko Sen Israel has the oldest continuous sovereign linker market, with CPI-linked government
+44 (0) 20 3134 2839 bonds first issued in 1955 and a wide spectrum of corporate inflation-linked bonds. This
arko.sen@barcap.com was in reaction to years of very high inflation rates. Although inflation has been below
10% since 1998, inflation-linked bonds are still an important instrument in Israel’s
Daniel Hewitt financial markets.
+44 (0) 20 3134 3522
daniel.hewitt@barcap.com CPI
The inflation reference for linked bonds in Israel is the consumer price index (CPI). The index
is calculated and published on the 15th of each month by Israel’s Central Bureau of
Statistics. This is calculated from a monthly survey of 1,320 goods and services, by
numerators or by phone, of c.2,700 stores, business and households in c.100 localities. The
weightings in the basket are based on the household expenditure survey and are changed
every few years. No schedule has been given for the next change, but weighting changes
tend to be modest.
The Israeli central bank targets 1-3% CPI growth per annum, and was successful in
achieving this target range from 2004 until 2008, when inflation exceeded 5% at one point.
Since then, the financial crisis has helped pull inflation lower, but new indirect taxes and
house price increases have kept inflation near or above the upper end of the target range.
Food accounts for only a small share of the CPI basket, and at this stage one of the key risks
to inflation seems to be from housing, which accounts for 21% of the CPI basket and has
remained buoyant throughout the global crisis. FX indexation in the Israeli economy has
fallen substantially. The BOI estimates that the proportion of rental contracts quoted in USD
has fallen from 90% to 15% over the past two years. As a result, the pass-through from the
exchange rate to the housing component has fallen, from 90% to 48% in the short term
(one month), and has disappeared over a longer term (year and above) where previously it
was fully transmitted.
0.0
Health ,
5.22 Housing , -2.0
Education , 20.69
12.53 -4.0
Dec-99 Dec-01 Dec-03 Dec-05 Dec-07 Dec-09
Furnishing Housing
, 3.75 maintenanc
Israel CPI
e , 10.62
Source: Haver Analytics, Barclays Capital Source: Bloomberg, Barclays Capital
Market development
The real curve in Israel has a greater maturity than the nominal curve (2036 against 2026
for the longest maturity nominal local government bond). However, the government’s
emphasis on developing the unlinked curve has been slowly paying dividends. The share of
the tradable local government bond market, which is unlinked (either to CPI or FX), has
risen from over 8% in 1995 to about 58% by end-2008. In 2009 this trend paused as the
share of CPI-linked issuance increased to 27% compared to 21-24% over 2006-07.
The reform of Israeli pension funds, which started in January 2004, marked an important
change in the local debt market: the issue of non-tradable government debt to these funds
virtually ceased and these funds became more active in the tradable local debt market –
both in linked and unlinked bonds. Initially, their buying interest was concentrated in CPI-
linked bonds, but in recent years their buying of linked and unlinked bonds has stabilised.
Domestic pension funds’ share of total holdings in tradable government bonds rose from
2% in 2002 to 13% by end-2008. Provident funds have generally declined in importance
and hence so have their share of the holdings, with individuals overtaking these funds as the
group holding the largest share of inflation-linked bonds (Figure 113).
Calculations
The annual coupon on Galil bonds are fixed real rates, while the value of principal is
enlarged by the rate at which CPI has increased between the base date and the latest
inflation observation. Indexation has a one- to a one-and-a-half-month lag: for example, the
base CPI for a Galil bond issued on 20 August 2001 is the CPI for July 2001. Both the coupon
and redemption value of Galil bonds in Israel have an implicit floor against deflation. By
Figure 112: Distribution of tradable debt by type (%) Figure 113: Distribution of total bond holdings (%)
100% 100
90% CPI-linked USD-linked Unlinked
80
80%
60
70%
60% 40
50% 20
40%
0
30% 2005 2006 2007 2008 2009Sep*
20%
10% Public Mutual fund Provident fund
Source: Ministry of Finance, Barclays Capital Source: Ministry of Finance, Barclays Capital
contrast, the value of the principal of the new inflation-linked bonds can rise and fall with
the change in CPI relative to the base CPI (ie, the latest CPI when the bond was first issued).
Another difference between Galil bonds and the new CPI linked bonds is the method of
computing interest payments. For Galil bonds, a multiplicative formula is used:
Where R = interest rate for the interest period; r = the fixed interest rate and T= number of
days in the interest period.
For the new CPI linked bonds, the interest payment is computed more simply as:
R = r*T/365
The key difference between Israeli inflation-linked bonds and those of most other markets is
that there is no official daily referenced index, so the principal and coupon can have step
moves according to the release of the latest CPI report.
Where M1 = latest CPI at the time of payment and M0 = base CPI, or CPI known when the
bond was first issued.
The quoting convention is to use uplifted dirty prices. The latest CPI report is used to
compute the index ratio from time of issue – this index is used to deflate the uplifted price
and hence derive the implied real yield. As mentioned above, this induces volatility in the
yield levels around CPI release dates.
The market for inflation derivatives is still in the early stages of development with the
interbank market not yet very active. It is, however, quite possible to trade inflation swaps
on request, and much of the current activity in this sphere is undertaken by Israeli
corporates hedging their balance sheet exposures. One of the more common products is
real rate swaps where the floating leg is 3mth Telbor and the fixed leg is the real rate
uplifted again by the ratio of the most recent CPI index over the CPI index at start. This
market is likely to grow over the coming months/years as local corporates and banks adopt
a more sophisticated approach to liability management and foreign interest in creating
global inflation exposure increases.
Turkey
Arko Sen Turkey issued its first Canadian-style linker in February 2007, a 2012 issue. Since then it
+44 (0) 20 3134 2839 has continued to build up the market for inflation linked bonds, especially in 2009 when
arko.sen@barcap.com linked issuance leapt to 16% of total supply. Even with issuance likely to increase further
and a sharp fall in real yields, inflation risks still make inflation-linked bonds attractive.
Christian Keller
+44 (0) 20 7773 2031
christian.keller@barcap.com
Consumer price inflation
Inflation-linked bonds are referenced off CPI, which is calculated and published by the
Turkish Statistical Institute. The index is published on the third day after the reference
month and has a large coverage span of 26 regions, 81 provinces, covering 447 items.
Generally, prices are collected twice each month (four times each month for vegetables,
fruits, and petroleum products). The prices include any relevant taxes. Weightings and the
item basket are updated annually, using continuous household budget surveys. The data
are not seasonally adjusted.
Inflation targeting by the central bank formally began at the beginning of 2006. Previously
monetary policy targeted inflation reductions, but with informal annual targets. After the
initial inflation target of 4% had been missed in 2006 and 2007 – and seemed unattainable
in the foreseeable future – the official targets were changed in mid-2008 to 7.5% for 2009,
6.5% for 2010 and 5.5% for 2011, with the uncertainty band of +/- 2 unchanged.
From a structural perspective, Turkey’s inflation challenges remain: 1) the inertia in service
price inflation, although there have been some breakthroughs in recent years – eg, in housing
rent inflation; 2) the high exposure to external food and energy prices; and related to this 3)
the pass-through from exchange rates (c.20% over a six-month horizon, according to our
estimates, and a total of 30% over the longer term). Thus, headline inflation in Turkey can
change significantly as a result of global commodity price developments combined with
changes in the exchange rate – ie, rising global food and energy prices combined with a
weaker TRY can drive up headline CPI even as demand pressures remain relatively muted.
CPI inflation ended 2009 at 6.5%, well below the official 7.5% target but well above the
Central Bank of Turkey’s projection of 5.5% in the quarterly inflation report. While demand
conditions remained weak, y/y headline inflation was driven higher by food price
Figure 114: Composition of Turkey’s CPI basket Figure 115: Inflation targeting is still evolving
Household Clothing 2
goods and and
Housing
services footwear 0
19%
8% 7% Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Jan 09 Jan 10 Jan 11
Source: Turkish Statistical Institute Source: Central Bank of Turkey, Bloomberg, Barclays Capital
developments. The government has announced a series of fiscal measures coming into
effect in early 2010: increases in excise taxes on gasoline, tobacco and alcohol, highway
fees, and various other taxes. In addition, the introduction of formula-based quarterly
energy price adjustments since 2008 also implies further energy price increases for end-
consumers in 2010, as a consequence of the rebound in oil and gas prices. Taken together,
and coupled with base effects following the sharp disinflation from December 2008, this
implies headline inflation rising toward 10% in the first half of 2010. With base effects
turning more favourable in the final months of the year and the output gap likely remaining
large, we expect inflation to moderate toward the end of the year, ending 2010 near the
upper end of the tolerance band. The challenge building for monetary policy makers will
also be to contain the deterioration in inflation expectations vis a vis the 5.5% inflation
target for 2011. Given that inflation expectations tend to be highly adaptive in Turkey, the
rise in headline inflation rate in H1 10 is likely to also raise expectations for 2011.
Market development
The Turkish Treasury re-entered the linker market in February 2007 with a new instrument
that followed the international standard for CPI-linkers (see below). A small amount of
inflation-linked bonds (albeit of a different format) had been issued in the 1990s, but were
redeemed in 2000 and never enjoyed great popularity. The initial launch of the February
2012 10% CPI-linked bond (TRY4.0bn, ~US$3.0bn sold) was met with strong demand and
quarterly taps of the 2012 issue raised the nominal amount outstanding to USD5.4bn by
end-2007. In 2008, the Treasury held only three auctions for a total sum of TRY2.4bn in
2012 and 2013 bonds. It was in 2009 that the pace of issuance really accelerated as the
Treasury held six auctions at a rate of up to one in a month, for a total amount of TRY23bn,
or about 16% of the year’s gross issuance. Two 2014 bonds were issued in 2009 with the
amounts outstanding being increased to TRY8bn and TRY6.3bn over the year. In 2010, a
new 5y bond was issued in February, which as of March had an amount oustanding of
TRY7.8bn. At the end of January 2010, the total amount of Lira-denominated debt (free
float) amounted to TRY263bn of which CPI indexed debt amounted to TRY27bn. It seems
clear that the Turkish Treasury remains committed to maintaining and building a real curve.
One of the key reasons behind the surge in inflation-linked issuance over 2009-10 is the
newfound interest for such bonds among local banks. They had been reluctant to buy these
bonds when nominal yields were high but, with the collapse in policy rates and nominal
yields, breakeven inflation had reached startlingly low levels. Partly as a function of this and
partly due to some fears of an inflation spike following the central bank’s sharp policy
easing, CPI linked bonds found renewed favour among local investors. Anecdotal evidence
also suggests that much of the inflation linked paper was originally bought for banks’ hold
to maturity portfolios.
Turkey has historically offered one of the highest real yields in EM markets, which was
reflected in the strength of foreign demand for Turkish linkers. This is, however, no longer the
case with real yields hovering around the 3-4% mark at the start of 2010. Cyclically, both real
yields and nominal yields are poised to rise, but structurally too it is perhaps not hard to argue
that Turkey will continue to offer one of the higher real yield levels in the EM space – as a
function of its current account and fiscal deficit and also due to the still comparatively low
monetary policy credibility under the relatively new inflation targeting framework.
To date, all new inflation-linked bonds in Turkey have been issued with a 5y maturity, but it
would not be a surprise if the Treasury chose to extend the curve longer. The liquidity in the
market has improved over the years but is still poor relative to the nominal yield curve, on
which a 10y bond was issued for the first time in 2010. Taking pure inflation views is still
difficult as the repo market is not developed enough to take short positions. Artificially
structured positions are possible by using cross currency swaps but these too are illiquid
and investors would also need to assume the asset swap spread risk. As yet there has been
no significant activity in inflation swaps.
Calculations
Inflation-linked bonds in Turkey follow a Canadian format that investors are familiar with: a
fixed annual coupon bond, with the principal value changing by the reference index, linked to
CPI with a two-to-three month lag. The Treasury updates the daily reference series for linkers
after a new CPI report is published. Turkish linkers are protected against deflation by a par
floor, although this likely has only a modest effect on pricing considerations, given that
inflation is still relatively high (8.7% average since 2004) though it has reached new lows
following the financial crisis (5.5 in November 2009). The format of the bond is as follows:
Existing bonds have a coupon of 10% in real terms paid semi-annually. They are indexed
against inflation as follows:
Coupon payment = (Reference index coupon date/Reference index issue date) * 100 * Real
coupon rate
Daily reference index = CPI a-3 + (G-1 / AG) * (CPI a-2 – CPI a-3)
Where CPI a-2 = CPI of (a-2) month and G = number of past days in month “a”
Where CPI a-3 = CPI of (a-3) month and AG = number of total days in month “a”
Notably, the market convention in Turkey has changed and CPI bonds are now quoted on a
clean price basis similar to other markets.
Non-government bonds
Both the nominal and real corporate bond markets in Turkey have shown slow progress.
This is due, in part, to the dominance of borrowing via an extensive local banking system
and recently because of the appeal of external borrowing. The problem has arisen because
of the high nominal yields compared with developed markets and also as, since 2002, the
general outperformance of the lira against the FX forwards has encouraged external
borrowing by Turkish companies. There is little indication that this will change, especially as
the government’s full balance sheet is likely leading to significant crowding out of the
domestic bond market.
South Africa
Jeff Gable Inflation-linked bonds have been available in South Africa since 2000, with swaps
+27 (0)11 895 5368 trading developing subsequently. Following a surge in inflation during 2008-09,
jeff.gable@absacapital.com medium-term inflation breakevens remain within the central bank’s 3-6% inflation
target. Surveys of business/labour inflation expectations show above-target results,
Bulent Badsha however. Since early 2009, the pace of linker issuance by the National Treasury has
+27 (0)11 895 5323 increased significantly, in line with the need for an extended period of deficit financing.
bulent.badsha@absacapital.com This additional issuance, along with the continued development of an active repo
market, has helped lead to a noticeable improvement in liquidity. Inflation and real rate
swaps are also increasingly active, helped by asset swapping of government bonds.
In 1989, the South African Reserve Bank (SARB) adopted a strong anti-inflation stance as
annual average CPI declined from a high of 18.6% in 1986, to a low of just 1.4% in 2004. The
inflation-targeting regime was announced in 2000; the first target year was 2002. The 3-6%
target has been unchanged since the introduction of the regime, helping to entrench the
disinflationary trend and reinforce the credibility of monetary policy. Inflation had been in the
3-6% band from September 2003 to March 2007, followed by a period of sharply rising
inflation throughout much of 2007 and 2008, reaching a peak of 13.6% y/y in August 2008.
This experience highlights that, despite a solid policy commitment to low inflation, South
Africa remains vulnerable to episodes of high inflation volatility. Part of the difficulty is due to
the currency, which historically has been subject to periods of significant movement.
Prices of goods and services included in the CPI are collected in the first seven days of the
month. While most prices are collected monthly, some are collected quarterly, semi-
Figure 116: Weight of CPI in basket Figure 117: CPI inflation history
,
Culture,
Food, 16 CPIX (previous MPC target)
Hotels & Other
Alcohol and CPI (new basket)
Restaurants 9% 14
Tobacco
7% SARB target upper bound
20% Clothing 12
and SARB target lower bound
Education,
footwear 10
Comms and
Insurance 4% 8
13% Other
housing and 6
utilities
4
10%
Transport OER 2
19% Households 12%
0
contents
2004 2005 2006 2007 2008 2009
6%
Source: Barclays Capital Source: Barclays Capital
annually, or annually. The collection of prices depends on the frequency with which they
tend to change. Every five years Stats SA conducts a Survey of Income and Expenditure of
Households. This information is weighted according to the population census figures in
order to represent all households spending patterns in South Africa accurately. The survey
is used to identify the goods and services purchased by a typical consumer, or household,
and which should be included in the basket of goods and services to monitor price changes.
From this survey, weights are determined for specific products in the basket. The weighting
of each product stays the same for the five-year period – ie, until the results of the next
Survey of Income and Expenditure of Households become available.
The most recent CPI basket reweighting took place in January 2009, based on the 2005
expenditure survey. In addition to the usual re-basing and re-weighting, the revised
methodology also involved a major reorganisation of the basket to “democratise” it by
removing luxury items, such as airline tickets, that might have a high expenditure value in
the basket but are only used by a small minority of South Africans. This resulted in the
number of items in the basket falling from 1,124 to 386. There are four other changes of
note, the most critical being the replacement of mortgage rates as a proxy for housing costs
(as per the previous CPI basket) with an estimate of Owner Equivalent Rent (OER). This is
likely to mean a smoother trajectory for inflation through the interest rate cycle as changes
in the policy rate under the previous CPI basket automatically induced a jump in the
inflation series via mortgage interest costs. OER has a weight of 12.21% in the current CPI
basket. The three further important changes all relate to weights, with the weight on
foodstuffs reducing to 14.27% from 20.99% previously, the weight of vehicle purchases
(new and used) rising to 11.25% from 5.95% earlier, and the weight of electricity prices
reducing from 2.97% to 1.68%. The lower weight of electricity in the current CPI basket will
be important in preventing the very large planned electricity price increases of the next few
years from passing directly into inflation.
The shift to the new basket affected a decline of about 1.4 percentage points on measured
CPI inflation in the January 2009 figure, but as with other inflation markets, holders of
inflation-linked bonds saw a direct index adjustment on an index rebase and so did not get
affected by the revised y/y rate. The seasonality of this new series is expected to differ from
that of the previous series, but over a longer period the new series is not expected to have a
significant effect on measured CPI. The next survey revision is due only in 2014.
Inflation returned to within the upper inflation band in October 2009, but breakeven
inflation measures suggest that the market expects inflation to remain firmly within the 3-
6% inflation target (albeit towards the upper end). The recent episode of high inflation has
had a damaging effect on surveys of inflation expectations, however, and though the
average of analyst expectations for 2010-11 inflation is within the inflation band, the
expectations of business and labour representatives are well above there.
Government bonds
Market development
The government first issued inflation-linked bonds in March 2000, launching the CPI-linked
6.25% March 2013, known as the R189, citing confidence that inflation would decline over
the medium term and underlining a policy commitment to low and stable inflation. The
government has underscored its commitment to the inflation-linked market by issuing
inflation-linked bonds on a regular basis. The frequency of amount of supply has increased
significantly, with auctions now being held every Friday in linkers across the curve (there are
currently four bonds: R189 (6.25% Mar ‘13), R197 (5.5% Dec ‘23); R210 (2.6% Mar ‘28);
and R202 (3.45% Dec ‘33)). In addition, to improve liquidity, the government introduced a
reverse repo facility in June 2006, which allows the SARB to lend up to ZAR1bn of inflation-
linked bonds to the market for fixed periods. Since demand for inflation linked bonds has
been strongest in the short end of the curve, the government announced in the 2010
Budget that two new ILBs maturing in 2017 and 2022 will soon be introduced.
Liquidity in the market is improving, albeit from very low levels. In 2009, the turnover ratio
(total turnover divided by the nominal outstanding issue of bonds at year-end) for
government linker bonds was 3.2, versus 27.2 for government nominal bonds. In part, this is
due to the small size of the market: linkers make up only 23% of the notional amount of
government bonds issued, and the largest linker issue, the R197, is only ZAR28.8bn, versus
ZAR72.0bn for the largest nominal issue, the R186. This understates the importance of
linkers, however, given that linkers have accrued significant inflation, with the R197
principal having increased to 1.67 times its original size by the start of 2010. The nature of
the investor base is also an important limiting factor for liquidity as linkers tend to be
bought by buy-and-hold domestic institutional investors, rather than by trading accounts.
However, as issuance has increased and real yields rise, there may be an increase in interest
from foreign participants, including hedge funds.
Calculations
South African government inflation-linked bonds carry a principal deflation floor and are
quoted on a real yield basis, with inflation indexation calculated using a slightly augmented
Canadian methodology. For settlement on the first day of any calendar month, the CPI from
four months previous is the reference CPI for that date. This means that South African
linkers have a lag that is a month longer than those in Canada, the US or the euro area.
Each day has its own distinct reference index. The first day of each month has a reference
index equal to that of the CPI of four calendar months earlier – eg, for 1 December 2009, the
CPI is for August 2009 and for 1 January 2010 the CPI is for September 2009. Reference
indices for intervening days are calculated by straight-line interpolation.
This formula is used to calculate a reference CPI for the official original issue date, or “Base
Reference index”. For settlement date or cash flow payment date “t”, a reference CPI is then
calculated. Both the reference index and the base index are rounded to 15 decimal places.
These two indices provide an index ratio for the value date:
For settlement amounts, real accrued interest is calculated as for ordinary South African bonds.
Dirty price and accrued are each multiplied by the index ratio to arrive at a cash settlement
amount. For coupons paid, the (real) semi-annual coupon rate is multiplied by the index ratio,
and likewise for the par redemption amount (with the cash value subject to the par floor).
Taxation
South African inflation-linked bonds pay interest on a semi-annual basis, with a 10-day
“books closed” ex-dividend period. However, international investors are not subject to
withholding tax, making investment relatively straightforward. For domestic investors, South
African CPI-linked bonds fall under section 24J of the Income Tax Act of 1962. Interest on
bonds is taxed on a yield-to-maturity basis. Coupon payments and the difference between
the acquisition cost and the nominal value of the bond are defined as interest and are liable
for income tax. Basically, inflation-linked bonds are treated like floating-rate instruments. The
tax liability is determined annually, taking into account any adjustments in the principal
amount and the coupon payments as a result of changes in the CPI.
Non-government bonds
As of December 2009, the Johannesburg Stock Exchange listed 54 non-government
inflation-linked bonds, with an outstanding value of ZAR36.7bn. Though the first issue in
the non-government market dates back to 2001, much of the issuance has occurred most
recently, with 2009 seeing total outstanding issuance double. Banks were the largest issuers
in this market during 2009, responsible for ZAR14bn in inflation-linked debt, bringing their
total outstanding figure to ZAR21bn. Much of the bank issuance remains driven by tailored,
structured notes, however. Parastatal (state-owned companies) inflation-linked issuance
remains disappointing. In 2009, ZAR1.5bn in new paper came to market from the likes of
ACSA (ZAR500mn) and Eskom (ZAR1bn). The sector’s reluctance to issue inflation seems
likely to continue into 2010 despite the significant infrastructure requirements that need to
be funded over the medium term. Among concerns such as large refinancing risks and cost
effectiveness, the fact that the market remains highly illiquid seems a key deterrent (about
3% of issue size was traded over 2009).
When swap trading began, it quickly developed into a more liquid market than bonds, as the
swap market was mainly an inter-bank market rather than a buy-and-hold one. While the
swaps market does not have the consistent weekly liquidity injection of auctions, these do
create swap market activity if there is asset swap flow. Recently, liquidity in the swap
market has dried up, with more frequent trading in the bonds. There is less of a structural
constraint on taking long/paid real yield positions in the swap market – but the repo market
is developing, with the aid from the SARB’s reverse repo facility enabling investors to take
on short positions in the bonds to some degree. Liquidity is notably worse than the nominal
swaps market. Even though the size of underlying flows has grown significantly as the curve
has become better defined, it is still mostly a market between five market-making banks.
For this reason, bid-offer spreads remain quite wide, ranging from 10-20bp.
Figure 118: ZAR real bond and swap curves Figure 119: Composition of the SA government bond market
2.5
2.0
Nominal
bonds,
1.5 442.7, 77%
R189 (2013) R197 (2023) R210 (2028) R202 (2033)
Real bond-swap spreads are currently trading at historically wide levels, leading to increased
interest from institutional investors to pay inflation linked swaps and thereby take
advantage of the wide spreads on offer. Although this has the potential to push real swap
yields higher, increased government ILB supply could have the same effect on bond yields
as well. The local pension fund industry’s liability driven investor flows will typically drive
real yield swaps lower relative to government real yields, while the lack of significant
development in the corporate side of the market makes the market rather one-sided.
Figure 120: ZAR linkers – historical performance and risk Figure 121: Return/risk versus nominals and equities
15% 2
1
10%
0
5%
-1
0% -2
2001 2002 2003 2004 2005 2006 2007 2008 2009 2001 2002 2003 2004 2005 2006 2007 2008 2009
Poland
Daniel Hewitt Poland inflation-linked bonds comprise a small market. The government has only issued
+44 (0) 20 3134 3522 two bonds – the first in 2004 and the second in 2008 – and they only account for 1.7% of
daniel.hewitt@barcap.com government debt. This is understandable as inflation was very low during 2002-06,
averaging just 1.9% per annum, but it has since gained momentum and averaged 4%
Arko Sen during 2007-09. The index is dominated by the 30% weight of food and beverages.
+44 (0) 20 3134 2839
arko.sen@barcap.com CPI
Inflation-linked bonds are referenced off the non-seasonally adjusted m/m CPI inflation, which is
published by Poland’s central statistical office around the middle of the following month. CPI is
calculated using the data gathered from 209 research regions (ie, a town, or a part of the large
city) across the country. The prices of 1,800 consumer goods and services are surveyed between
the first and the 25th day of each month. Fruit and vegetables are priced three times per month.
Prices of the other goods and services are collected once a month. About 292,000 individual
prices are collected each month. Since 1990, the weighting system has been based on a
household budget survey. The weightings are revised annually but importantly, and unusually,
the statistical office only publishes m/m and y/y changes in CPI – not the underlying index.
Poland adopted inflation targeting in 1999. The NBP (National Bank of Poland) inflation target
has been 2.5% +/- 1% since 2004. During 2002-06, inflation averaged only 1.9% per annum.
During 2007-09, inflation doubled to 3.9% per annum. The main factors accounting for this
increase were a tripling of food and beverage prices (30% of the index, albeit from 33% three
years previously); a doubling of housing prices (20% of index); and a doubling in energy prices
(11% of the index). Clearly, factors outside the control of the MPC – eg, global trends in food
and commodity prices – have a large impact on the path of inflation. The MPC pursues its
inflation targets seriously and expects to keep inflation within the inflation target band.
However, this will depend on global trends in prices.
Market development
The Polish linker market has grown slowly, with only one significant sovereign issue, a 3.0%
2016 bond, first auctioned on 18 August 2004, and built up by 2008 to about PLN9.1bn
outstanding. It is unlikely to be tapped again. At time of issue, domestic investors were ill-
placed to deal with inflation-linked bonds and the bulk of the issue was place outside Poland.
Education, 12.0
Others, 5%
1%
Food and 10.0
Entertainment Beverages,
14% 30% 8.0
6.0
Transport
and 4.0
Communication
14% 2.0
Apparel, 5%
Health, 5% 0.0
House Dec-99 Dec-01 Dec-03 Dec-05 Dec-07 Dec-09
maintenance Housing,
and equipment, 19%
Poland CPI
5%
Source: Haver Analytics, Barclays Capital Source: Bloomberg, Barclays Capital
Since then only one new series has been started – the 2.75% of 2023, which was first
auctioned in July 2008 but has not been tapped since and only has PLN 450mn outstanding.
The issuer’s original intention was to provide an inflation-linked asset to the growing pension
industry, but there has been little progress here. As a result of the modest growth in this
market, there have been no issues of corporate inflation-linked bonds, although in the past
when the bond traded cheaply, there were euro-structured notes with the government linker
as the underlying asset.
Trading volumes in the sovereign linker market are low (even compared with other EM
linker markets), and the amount outstanding of the 2016 bond has been eclipsed by other
new entrants to the EM linkers market. In 2010, there are plans for only very modest linker
issuance, thus the outlook for trading liquidity looks poor in the near term, although further
developments in the pension fund industry could bring back government interest in this
market over the medium term.
Local pension funds have about USD60bn of assets under management and have grown quickly,
having first emerged from the government reforms of the previous “pay-as-you go” pension
system in 1999. These pension funds hold very few inflation-linked bonds. There are several
possible factors behind this, which might change. First, the majority of pension funds will only
start to make payments in 2010, suggesting that hedging inflation liabilities could gradually
become a more pressing concern. Second, initial regulation of the new Polish pension funds
encouraged total returns-focused investment strategies, which a change in regulation could
alter, although the Polish Ministry of Finance has yet to announce any formal changes.
Calculations
The Polish government’s inflation-linked bonds carry a principal deflation floor and are
quoted on a price basis, with inflation indexation calculated according to the Canadian
methodology. The only difference from standard Canadian methodology is that there is no
CPI Index, so the month-on-month CPI changes are combined in order to calculate an index
that is based to July 2003. For settlement on the first day of any calendar month, the implied
CPI Index from three months previously is the reference CPI for that date, with linear
interpolation during the month. The reference coefficient for month “T” is published to five
decimal places and in the case of the August 2016 linker bond is 100 times the product of
the monthly changes. For settlement amounts, real accrued interest is calculated with
reference to the index ratio, with coupons paid annually.
Figure 124: Polish linkers: historical performance and risk Figure 125: Return/risk versus nominals and equities
8% 1.0
0.5
6% 0.0
4% -0.5
-1.0
2%
-1.5
0% -2.0
2005 2006 2007 2008 2009 2006 2007 2008 2009
Source: Barclays Capital Source: Thomson Datastream, Barclays Capital
Iceland
Alan James Iceland issued its first government inflation-linked bond in 1964, but the market is
+44 (0) 20 7773 2238 dominated by the Housing Finance Fund mortgage agency. Mortgage debt was partially
alan.james@barcap.com indexed as long ago as 1955, with the market consolidated in 2004 into benchmark
bonds that are similar to other linker markets.
The Central Bank of Iceland adopted a monetary policy inflation target of 2.5% on the CPI in
March 2001. The Central Bank’s inflation target takes priority over all other objectives, and it
has a +/- 1.5% tolerance band (at initial adoption, the upper tolerance limit was 6%, falling
to 4.5% at the end of 2002). However, the challenges of inflation targeting for such an open
economy were highlighted by inflation moving above 8% in mid-2006 following currency
weakness. Inflation peaked above 18% at the start of 2009 following the sharp depreciation
of the krona through 2008. In its November 2009 inflation projections, the Central Bank did
not envisage CPI inflation dipping below the 4% upper tolerance bound before Q4 10 or
below 2.5% before 2012.
Figure 126: Components of Icelandic CPI Figure 127: Icelandic inflation targeting has been unsuccessful
Food, drink 20
Hotels and Others CPI
8% and tobacco 18 Central Bank target
restaurants
18% Upper tolerance bound
5% 16
Health and Clothing Lower tolerance bound
14
recreaction and
footwear 12
4%
6% 10
Communica 8
tions
6
4% Housing
Transport 4
and utilities
13% 24% 2
Furnishings
0
8%
Mar 01 Mar 03 Mar 05 Mar 07 Mar 09
Source: Statistics Iceland, Barclays Capital Source: Statistics Iceland, Central Bank of Iceland, Barclays Capital
Since 2004, HFF has issued bonds dated 2014, 2024, 2034 and 2044, though the shortest
issue is no longer re-opened and is now notably smaller than the other bonds. As of the end
of November 2009, there were ISK526bn notional of benchmark HFF bonds, with a market
cap of almost ISK800bn (~US$6.5bn). Issuance has slowed in recent years due to the
weakness of the domestic housing market, though this has been offset by the increase in
the uplifted size of the market due to inflation, with the reference CPI at the start of 2010
more than 50% higher than when the benchmarks were issued during 2004.
South Korea
Wai Ho Leong South Korea issued its first inflation-linked bond in March 2007, using a standard
+65 6308 3292 Canadian format, albeit without a par floor on the principal. As of January 2010, there is
waiho.leong@barcap.com only one issue, the KTBi Mar 2017, which has not been re-opened since July 2008. There
are plans for resumed issuance starting June 2010. The issue is linked to headline
Matthew Huang Korean CPI. The central bank targets inflation on both the headline and core CPI
+65 6308 3093 measures, with a current target range of 2.0-4.0%.
matthew.huang@barcap.com
CPI
The reference index for computing the inflation adjustment factor for the first South Korean
linker is the South Korean Headline Consumer Price Index (CPI), which is not seasonally
adjusted. Since 1998, the Bank of Korea has applied inflation targeting, and set the target in
consultation with the government. The inflation target for 2010 onward is 3 percent, with a
tolerance range of plus/minus 1 percentage point around this target, in terms of the 12-
month rate of change in the consumer price index. Previously the tolerance band was plus
or minus 0.5%. The Bank reviews the performance of inflation targeting annually and makes
a public announcement of the results. The target horizon is three years (2010~12). Since
inflation targeting was institutionalised in 1998, inflation in Korea has since been better
contained and more stable. For instance, inflation receded from 6.2% in the last 30 years to
just 2.9% in the last 10 years – with a standard deviation of 1.2%.The Korean National
Statistics Office publishes CPI data each month, usually the first working day of the
following month, eg, the headline CPI for January 2010, released on 1 February 2010.
Although the CPI has been used as a measure of South Korean inflation since 1936, there have
been considerable changes in South Korea since its inception. These changes largely reflect
the national shift from agriculture to heavy industry in the 1970s and 1980s, before the
growth of the service sector in the 1990s. The CPI is measured monthly by the Consumer
Price Survey, which takes the form of a questionnaire. Approximately 80,000 price quotes are
obtained each month from 13,000 outlets, and the quotes are obtained by personal interview.
The major categories of goods and services and their weights are shown in Figure 128.
The index Is revised every five years to reflect the changes in consumption structure of
Figure 128: Breakdown of CPI by major category Figure 129: CPI inflation and the breakeven of KTBi
Misc 7.0
Recreation Food
5% 6.0
6% 14%
Health 5.0
5% 4.0
Cooked
Education food 3.0
11% 13% 2.0
1.0
Clothing 0.0
6%
-1.0
Housing Transport May 07 Nov 07 May 08 Nov 08 May 09 Nov 09
23% 17%
BoK policy bad Breakevens to Bonds
CPI % y/y Breakevens to Swaps
Source: Korea National Statistic Office Source: Bloomberg, Bank of Korea, Barclays Capital
urban households. In January 2002, the results of revised index as of 2000 were released. In
April 2003, the Nationwide Chain Index as of 2002 as a supplementary index was compiled
and released. The indices have been revised on a five-year basis to reflect the changes in
consumption structure of urban households. In December 2006, the results of revised
indexes were released with 2005 prices based.
Government bonds
Market development
The government first issued inflation-linked bonds in March 2007, when it launched the
CPI-linked 2.75% March 2017, known as the KTBi, with a view to expanding the Treasury
bond market base to resident investors and foreigners. The government continues to
express a commitment to expand its bond markets including the KTBi; however, the
financial crisis has led to a pause in new issuance of linkers. The last monthly auction was
held in July 2008 for KRW61bn (c.USD61mn as of July 2008). A new issue was scheduled for
September 2008, but was postponed due to the financial crisis, with KRW900bn of the
existing issue bought back by the Treasury in Q4 08. As economic conditions in Korea and
globally continue to stabilise, we expect the priority for developing the linkers market will
again come to the forefront of government priorities, leading to continued support for the
nascent Korean linkers market. The finance ministry stated in its 2010 issuance strategy
that it intended to resume inflation-linked issuance in June 2010.
At present, market liquidity for the sole outstanding issue remains poor, with minimal
trading in secondary markets. Currently there is KRW1.77trn outstanding of the bond, less
than 1% of the total outstanding government bond market. The investor base has tended to
be foreign buy-and-hold institutional investors rather than trading accounts, limiting the
amount of secondary trading. For reference, bid-ask spreads for linkers are 5-10bp
compared to 1-3bp for 5y and 10y KTBs. We anticipate that liquidity will improve upon the
government’s opening of a new issue in June 2010. Furthermore, heightened concern over
rising inflation following historical amounts of monetary easing through 2009 could also
provide demand for inflation-linked products auctioned in June.
Investment regulations will also help demand for Korean bonds. Investing in Korean linkers
has been helped by the removal of withholding tax for foreign investors and the addition of
Korean bonds into Euroclear’s settlement systems in 2009. The addition into Euroclear is
still at an early stage with limited trading at present, but with time this is likely to improve
accessibility tremendously as any counterparty with a Euroclear account would have access
to trade the bonds, reducing the need for an onshore custodian account and reducing
associated paperwork for foreign investors.
Calculations
South Korean government inflation-linked bonds are quoted using the standard Canadian
model, with no floor on the principal value. For settlement on the first day of any calendar
month, the CPI from three months previously is the reference CPI for that date. A reference
CPI value is calculated for every day based on the CPI values for three months and two
months prior to the month containing the settlement date. The reference CPI for any day
during the month is calculated by linear interpolation.
(d − 1)
(CPI t −2 − CPI t −3 ) + CPI t −3
m
d = day of the month eg, 1st implies d=1
m = number of days in that month
The indexation factor is the reference CPI for the settlement date divided by the reference CPI
for the base date. Coupons are accrued on an actual/actual basis and paid semi-annually.
Taxation
The tax treatment for the KTBi is the same as that for KTB; however, there is now a
differentiation between taxes passed for resident and non-resident investors. As of 21 May
2009, non-resident investors were made exempt from taxes on interest and capital gains
taxes through a tax law change by the Korean parliament. The change was made amid
concerns of insufficient demand for KTBs to fund Korea’s supplemental budget. The
removal of taxes was made in a bid to increase foreign participation in Korean Treasury
markets. Prior to the change, withholding and capital gains taxes were assessed subject to
levels determined by tax treaties. Standard rates of 15.4% withholding tax on interest
income and 24.2% on capital gains were assessed if not specified in a treaty otherwise.
Taxes were not removed for Korean resident investors and a 15.4% withholding tax and a
24.2% capital gains tax applies. This too is a change as financial institutional investors were
exempt from withholding tax prior to 1 January 2010.
INFLATION INDICES
US TIPS (Series-L)
INSTEP
INSPIRE
Real income
In this section of the publication, we offer overviews of our flagship inflation-linked indices
(benchmark and other index products), with details on specific index rules and return
calculations.
A complete database of bond and index returns in local currency plus returns hedged
and unhedged in most major currencies.
Bond and index analytics including forward yields, breakeven, carry and an interactive
carry calculator.
Forward index information with projected index constituents at the next rebalancing
date available for the majority of indices to help investors manage against their
benchmark.
Real time indices and analytics for the US, euro area and UK markets.
Barclays Capital inflation-linked indices are also available via Bloomberg, and a growing
number of other third-party data providers. Please contact the Index, Portfolio & Risk
Solutions team for more details or for requests for customised indices.
Introduction
The Barclays Capital Inflation-Linked benchmark family includes flagship government
benchmarks for developed and emerging markets, broader benchmarks for the euro and
sterling markets that include both government and non-government linkers, and
comparator indices that consist of the nominal comparator bonds for government linkers.
In addition to these traditional bond indices, other inflation-linked index products that can
be used as benchmarks include breakeven inflation indices that measure the total return of
a breakeven inflation position and inflation-swap indices that can be used for Liability
Driven Investing (LDI) indices.
Barclays Capital launched its first global benchmark index for this asset class with the World
Government Inflation-linked Bond Index (WGILB) in October 1997, and has covered the
expansion of the international inflation-linked bond markets with successive benchmark
index launches as either stand-alone indices or additions to existing flagship indices. The
most notable expansions were the launch of the Emerging Markets Government Inflation-
Linked Bond Index (EMGILB) and the Universal Government Inflation-Linked Bond (UGILB)
Indices. Currently, the Barclays Capital Inflation-Linked Index family tracks 19 countries and
over 200 inflation-linked securities.
Dec 2009 Greece excluded from World Government, Euro and Euro Govt Inflation-Linked
Indices following ratings downgrade to BBB+
Feb 2009 Barclays Capital BESA South Africa Government Inflation-Linked Bond Index is
rebranded as Barclay Capital / ABSA South Africa Government Inflation-Linked
Bond Index and adopts ABSA capital market makers pricing for index
calculation
Oct 2007 Universal and EM government inflation-linked bond indices are launched
Jun 2006 Inclusion of Greece into GILB, EGILB and EILB. Index methodology changed
such that the returns are calculated month-to-date and intra-month coupons
are held in cash and accrue at the local market monthly deposit rate
Mar 2006 Germany joins GILB, EGILB and EILB
Apr 2005 Japan joins GILB
Jan 2004 Addition of investability rule to bond criteria.
Amendment of the market aggregate face value and bond minimum issue size
requirements in relation to the application of FX rates
Aug 2003 Italy joins GILB, EGILB and EILB
May 2003 Euro and euro government inflation-linked indices launched to accommodate
the growing Euro Inflation-Linked market
Jan 2001 Sterling All Bond, Gilt and Non-Gilt indices launched.
Oct 1997 World Government Inflation-Linked Index launched with a base date of 31
December 1996. The founding markets included the UK, US, France, Sweden,
Australia and Canada
Source: Barclays Capital
Figure 131: The Barclays Capital Inflation-Linked Government Bond Index Family
Euro UK
UK Euro AU CA JP SE US Latam EMEA Asia Greece
Non- Non-
Govt Govt Govt Govt Govt Govt Govt Govt Govt Govt Govt
Govt Govt
Taken together, inflation-linked issues from the EMU form a major part of the World
Government Inflation-linked index. The Barclays Capital Euro Government Inflation-linked
Bond Index covers euro-denominated debt from euro area member governments linked to
domestic or EMU HICP inflation indices. Currently, the index includes government bonds
from France (linked to FRCPI and EMU HICP Ex-tobacco), Italy and Germany. The Euro
Government Index market capitalisation is now over €270bn ($395bn equivalent)
comprising 20 bonds with maturities out to 2041. Sub-indices are also available that cover
securities linked to EMU HICP and FRCPI inflation indices.
The index presently has a market capitalisation of over US$1.7trn and includes inflation-
linked government bonds from 19 countries; in order of size – the US, UK, France, Brazil,
Italy, Japan, Canada, Sweden, Germany, Argentina, Mexico, Greece, South Africa, Australia,
Turkey, Colombia, Chile, Poland and South Korea.
Overview
Barclays Capital launched its first inflation-linked bond index, the Barclays Capital World
Government Inflation-Linked Bond Index (WGILB), in October 1997. This index measures
the performance of the major government inflation-linked bond markets and is designed to
include only those markets in which a global government linker fund is likely and able to
invest. Investability is therefore a key criterion for inclusion of markets in this index. Markets
currently included in the index in the order of their inception are the UK (1981), Australia
(1985), Canada (1991), Sweden (1994), the US (1997), France (1998), Italy (2003), Japan
(2005) and most recently Germany (2006). Greek Government Inflation-Linked bonds
became eligible for both the Euro and World Government Inflation-Linked Indices on 30
June 2006, but due to a ratings downgrade in December 2009 became ineligible (Greece
remains as part of the Universal Government Inflation-Linked Bond Index – see Figure 131).
The total market capitalisation of the index has grown more than ten-fold since the launch of
the Index in December 1996. This is partly due to inflation accretion but by far the most
important factor is the emergence of new inflation-linked bond markets.
At inception, the index was dominated by issuance from the UK, which at that time accounted
for approximately 80% of the index by market value. Heavy issuance elsewhere has now
reduced the UK share to around 23% and we now see the US as the dominant market with
around 39% of the index. The share of the euro area has grown rapidly, from 2% in
September 1998 when France first issued, with the euro government share of the World Index
reaching around 27% by December 2009.
Figure 132: Index market value by country Figure 133: Index weights by country
Sweden UK 60%
1000 50%
US Greece
40%
800
30%
600 20%
10%
400 0%
200 Dec-96 Dec-99 Dec-02 Dec-05 Dec-08
Figure 134: World Government Inflation-Linked Bond Index inclusion criteria and review policy 12
Criteria for inclusion in the index
Inclusion of The WGILB Index is built bottom up from a selection of country/currency indices based on rating and size.
markets in the Rating: Long-term local currency ratings from S&P and Moody’s are considered, and the lower of the two is used for
World Govt Index the index. Minimum rating for inclusion is A3/A- for G7 and Euro-zone markets and Aa3/AA- otherwise.
Market size: Minimum aggregate issuance (non inflation-adjusted) for inclusion of bonds from a currency zone is
$4bn. Limits are set in local currency terms using WM closing spot rates from the last business day of the year.
These are in force for a full year, effective at the end of the first quarter of the next year, and will be reviewed
annually.
Min issue size Minimum issue sizes are set in local currency terms and may be reviewed annually by Barclays Capital taking into
account local market conditions, issuance trends and movements in exchange rates. The following minimum issue
sizes presently apply; AUD700mn, CAD600mn, EUR500mn, JPY50bn, SEK4bn, GBP300mn, and USD500mn.
Underlying Government domestic inflation-linked debt issued in domestic currency of that country only. Bonds must be capital-
Market/bond type indexed and linked to a commonly used domestic inflation index. In the euro area, domestic inflation indices and the
EMU HICP are eligible. The notional coupon of a bond must be fixed or zero. Issues that are not available in whole or
part to international investors are not eligible for the index.
Maturity and issue Bonds must settle on or before the rebalancing date and have a minimum remaining life of at least one year.
Index calculation and monthly review
Pricing The index uses mid-market prices from local market close.
methodology
All spot and forward foreign exchange rates used are official WM Company mid rates from the London market at
4pm.
Settlement The index uses standard settlement conventions for all calculations. Market calendars most appropriate for
conventions international investors are used.
Index frequency The index is calculated daily and has a value for each calendar day.
and reinvestment
Income received during the month is put on deposit until the month-end rebalancing, at which point it is reinvested
back into the index.
Reference deposit rates are set on the last business day of the month for the next month. Reference rates are 1M
LIBOR – 15bp (if LIBOR is not available in the market currency a local equivalent may be used – see Figure 24).
Review procedure The index is reviewed and rebalanced monthly on the last calendar day of each month. New bonds and
taps/increases entering the index must have settled on or before the rebalancing date.
Maturity bands Maturity bands are inclusive at the lower bound and exclude the upper bound. Bonds are allocated to maturity
bands based on their remaining time to maturity.
Index holdings The index holding of each bond for the next month is set to the amount outstanding on the review date. The face
values are used rather than an inflation-adjusted value. The indices are weighted using the market capitalisation as
standard.
12
Index rules are periodically reviewed by Barclays Capital Index research through formal index governance
procedures. The most up to date set of rules and related notifications can be accessed at
www.live.barcap.com (keyword: index).
Index structure
World Govt
Euro Govt
EMU HICP Linked
Source: Barclays Capital
Composite indices of countries such as World Ex-UK, World Ex-US, UK+US+France are also
available as standard or are customisable on request.
Maturity buckets
The standard maturity breakdowns consists of All Maturities, 1-3y, 1-5y, 1-10y, 3-5y, 5-7y, 5-
10y, 5-15y, >5yr, 7-10y, 10-15y, >10y and >15y. As of 31 December 2009, except for bonds with
less than a year to maturity, the index covers 100% of each of the eligible markets.
Figure 136: World Government Inflation-Linked Index characteristics (as of 28 February 2010)
Weight
Market cap Market % of No of Index
Base date Price source (local bn) cap ($ bn) overall issues rating
Introduction
The Euro-Inflation-Linked Index covers euro-denominated inflation-linked debt issued by both
government and quasi-government agencies, which presently includes issuers such as CADES,
CNA, ISPA and RESFER and EMU member states. At inception (31 December 1999), the Euro
Inflation-Linked Index had a market value of €11.1bn and included just French domestic
issuance. Since its launch, the index has grown in terms of market value and the number of
markets covered by the index, reaching €290.6bn at year-end 2009, with market coverage
extending to include issuance from Italy and Germany, who joined the Euro Inflation-Linked
Index in September 2003 and March 2006, respectively. Greek Government Inflation-Linked
bonds became eligible for both the Euro and Euro Government Inflation-Linked Bond Indices on
30 June 2006, but due to a ratings downgrade in December 2009 became ineligible (Greece
remains as part of the Universal Government Inflation-Linked Bond Index).
The Euro Inflation-Linked Index family includes a broad Index and sub-indices by
government and non-government sectors and by linking inflation index. Government debt
is further divided by country of issue.
Figure 137: Euro Government Index breakdown by issue type Figure 138: Euro Government Index breakdown by issue type
60%
200 Germany Govt
France Govt FRCPI-Linked 40%
150
20%
100
0%
Sep-98 Sep-00 Sep-02 Sep-04 Sep-06 Sep-08
50
France Govt FRCPI-Linked Germany Govt
0 Italy Govt France EMU HICP-Linked
Sep-98 Sep-00 Sep-02 Sep-04 Sep-06 Sep-08 Greece Govt
Figure 139: Euro Inflation-Linked Index inclusion criteria and review policy
Criteria for inclusion in the index
Issuer type Issuer must be an EMU member state or a quasi-government body of an EMU member state
Bonds must be capital-indexed and linked to an eligible inflation index. They should be denominated in Euros and
Bond type
pay coupon and principal in Euros.
Linking Index The linking Index can either be a domestic inflation index or the harmonised EMU HICP.
The index uses the lower of the Standard & Poor’s and Moody’s rating (long-term local currency rating for
Rating Government issues) to determine the index rating for each bond. All Government and non-government issues
must have a rating of A3/A- or better.
Coupon type The notional coupon of a bond must be fixed or zero.
Maturity Bonds must have a minimum remaining life of at least one year on the rebalancing date.
Issue date Bonds must settle on or before the rebalancing date to be eligible for the index
Min issue size The issue size must be equal to or in excess of €500mn.
Index calculation and monthly review
Pricing The index uses mid-market closes from Barclays Capital Market Makers.
methodology All spot and forward foreign exchange rates used are official WM Company mid rates from the London market at
4pm.
Settlement The index uses standard settlement conventions for all calculations. Market calendars most appropriate for
conventions international investors are used
Index frequency and The index is calculated daily and has a value for each calendar day.
reinvestment Income received during the month is put on deposit until the month-end rebalancing, at which point it is
reinvested back into the index.
Reference deposit rates are set on the last business day of the month for the next month. The reference rate for the
Euro and Euro Government Inflation-Linked Index is EUR 1M LIBOR – 15bp
Review procedure The index is reviewed and rebalanced monthly on the last calendar day of each month. New bonds and
taps/increases entering the index must have settled on or before the rebalancing date.
Maturity bands Maturity bands are inclusive at the lower bound and exclude the upper bound. Bonds are allocated to maturity
bands based on their remaining time to maturity.
Index holdings The index holding of each bond for the next month is set to the amount outstanding on the review date. The face values
are used rather than an inflation-adjusted value. The indices are weighted using the market capitalisation as standard
Index structure
The Euro Inflation-linked Index comprises an overall index, and is split into government and
non-government sectors and also by Inflation Index. The government sector is further
divided by country of issue or by the Linking Inflation Index. The non-government sector
has only quasi-government issues, which are broken down by the Linking Inflation Index.
Figure 140: The Barclays Capital Euro Inflation-Linked Bond Index structure
Government Non-Government
Euro Govt
EMU HICP Linked
Source: Barclays Capital
The Euro Inflation-Linked Bond Index has a base date of 31 December 1999, with individual
country and sector indices available since the inception of each market as in the table below.
As well as an overall index, sub-indices are available for the standard maturity breakdown, i.e.,
1-3y, 1-5y, 1-10y, 3-5y, 5-7y, 5-10y, 5-15y, >5y, 7-10y, 10-15y, >10y and >15y. Foreign and
hedged indices are available in AUD, CAD, CHF, GBP, JPY, SEK, USD and ZAR.
Figure 142: EM Govt Inflation-Linked Index inclusion criteria and monthly review
Criteria for inclusion in the index
Markets included Markets are included based on quantitative and qualitative criteria including, size, and depth of market, general
investability and availability of pricing information.
Countries currently included in the index are:
LatAm: Argentina, Brazil, Chile, Colombia, Mexico.
EEMEA: South Africa, Poland, Turkey
Asia: South Korea
Country rating If an index-eligible country defaults on all of its inflation-linked debt then the market will be removed from the index
at the earliest opportunity.
If individual bonds within a market default then these bonds will be removed from the index at the earliest
opportunity, and the eligibility of the overall market will be reviewed.
Bond inclusion rules For details of bond inclusion rules for each market please refer to Figure 143
Index calculation
Index pricing The index uses daily mid market closes from Barclays Capital market makers taken at local market close for all
methodology markets with the exception of South Africa, where ABSA Capital market maker mid closes are used.
Hedged and un-hedged currency indices are calculated using 4pm London rates from WM Company.
Treatment of Income from coupon payments is held in cash until month-end when it is re-invested into the index. For South
income Africa, the cash is re-invested intra-month into 1mth JIBAR – 15bp. In other markets the cash is held in the index
without accruing any interest intra-month.
Source: Barclays Capital
Index structure
As well as an overall index, sub-indices are available for the standard maturity breakdown,
i.e. 1-3y, 1-5y, 1-10y, 3-5y, 5-7y, 5-10y, 5-15y, >5y, 7-10y, 10-15y, >10y and >15y. The
overall index is also broken down by regional indices for LatAm, EEMEA and Asia, which are
broken down further by maturity.
Argentina Brazil Chile Colombia Mexico Poland South Africa Turkey South Korea
Source: Barclays Capital
Indices are available in local currency and hedged and un-hedged into the following base
currencies; USD, EUR, GBP, JPY and SGD. A constrained index with a cap on maximum
country weight of 25% is also available. Please refer to our website www.live.barcap.com
(keyword: index) for more details on this index.
At inception, the Sterling Non-Gilt Inflation-Linked Bond Index consisted of 12 issues with
an aggregate face value of £740m. Since inception, the Non-Gilt portion of the All-Bond
Index has grown substantially and at 31 December 2009 consisted of 50 bonds with an
Aggregate face value of £17.4bn ($21.4bn equivalent). Since 2005, much of the new Non-
Gilt issuance has been taken up by asset swap investors and thus has not been eligible for
index inclusion, a trend brought about by the limited number of issuers in the Non-Gilt
Inflation-Linked market and relatively poor liquidity. Towards the end of 2009, index activity
has been relatively vibrant with a number of issuers becoming ineligible due to both rating
downgrades and rating withdrawals – this trend particularly affected the monoline
insurance wrapped issuers. Supply has continued to be steady in the second half of 2009,
with UKRAIL regularly tapping and the occasional new issues, typically from utility issuers.
Figure 146: Sterling Non-Gilt Index market value by sector Figure 147: Sterling Non-Gilt weights by sector
25 100%
Insurance Wrapped 90%
Corporate 80%
20
Agency / Supra 70%
Market Value (£bn)
60%
Utilities 50%
15
40%
30%
10 20%
10%
5 0%
Dec-99 Dec-01 Dec-03 Dec-05 Dec-07 Dec-09
Utilities Agency / Supra
0
Dec-99 Dec-01 Dec-03 Dec-05 Dec-07 Dec-09 Corporate Insurance Wrapped
Figure 148: Sterling Inflation-Linked Index Inclusion criteria and monthly review
Criteria for inclusion in the index
All issues must be denominated in sterling and pay interest and principal in sterling. All (real) cash flows must be fixed
Bond type at issue. For amortising structures, the amortisation schedule must be fixed at issue and must be in the public domain
(ie. available in full via market data services such as Bloomberg).
Linking index The linking index can be UK RPI or LPI.
The index uses the lower of the Standard & Poor’s and Moody’s rating to determine the index rating for each bond. If
Rating
neither of these rates the issue, then Fitch ratings are used. All issues must have a rating of Baa3/BBB- or above
Coupon type The notional coupon of a bond must be fixed or zero.
Maturity Bonds must have a minimum remaining life of at least one year on the rebalancing date.
Issue date Bonds must settle on or before the rebalancing date to be eligible for the index.
Bonds must have an issue size of £100mn or above. Amortising inflation-linked bonds must have a remaining capital
Min issue size
amount greater than or equal to £100mn, where capital amount = face value x amortisation factor.
To be eligible for inclusion in the index a new issue has to be priced by at least two dealers. The test is implemented by
looking for pricing coverage in the last calendar week of each month for all potential index entrants. Any new issue that
has not sufficient price coverage to enter the index at the first opportunity will be checked again the following month, but
Market pricing
after that becomes ineligible even if it is subsequently priced by more than one dealer. For the avoidance of doubt, this
procedure will apply to new issues only – thus at the time that the market pricing rule was introduced (1 July 2006), then
prevailing index constituents were not subject to the pricing check.
Index structure
The sterling inflation-linked index family comprises an overall index as well as two index
sub-groups: Gilts and Non-Gilts. The Non-Gilt index is further divided into rating and sector
indices. The sector indices have been designed to reflect the types of non-government
issuance seen to date.
UK Govt Non-Gilts
Agency/ Insurance
Corporate Utility
Supranational Wrapped
The Non-Gilt, overall and sector indices have a base date of 31 December 1999. The Gilt
index has a base date of 31 December 1996; history is available for the index-linked Gilt
market back to 1981. As well as an overall index, sub-indices are available for the standard
maturity breakdown, ie. 1-3y, 1-5y, 1-10y, 3-5y, 5-7y, 5-10y, 5-15y, >5y, 7-10y, 10-15y,
>10y and >15y.
The two main benchmarks in the series-L family are the Global Inflation-Linked (Series-L)
and US TIPS (Series-L) indices. Differences between these two benchmarks and their
WGILB counterparts are as follows:
− Both the WGILB and Global Inflation-Linked (Series-L) indices include the following
countries: US, UK, Canada, Sweden, Italy, France, Germany and Japan.
− The WGILB Index includes bonds from Australia, which is not an eligible country for
the Global Inflation-Linked (Series-L) Index.
− This difference has a small impact on the overall indices, as Greece only has two index
eligible bonds (1.3% of the Global IL Index by market value) and Australia has only three
index eligible bonds (0.7% of the WGILB Index by market value).
− US TIPS in the Global Inflation Linked (Series-L) and the stand-alone US TIPS
(Series-L) Indices both exclude government/Federal Reserve purchases of TIPS
from the face amount outstanding of each bond in the index, consistent with float
adjustments made for nominal US Treasuries in the US Aggregate Index.
− The US TIPS component of the WGILB Index uses the full issue size to determine
index weights and does not adjust for Federal Reserve or government holdings.
Therefore, the amount outstanding of a TIP bond in the Global IL Index (Series-L)
may be less than the amount outstanding of the same bond in the WGILB Index if
any part of the issue was purchased by the US Government or Federal Reserve.
The index has a market capitalisation of US$1.4trn and includes 92 issues from 10
governments (as of 31 December 2009).
Figure 151: Inflation-Linked (Series-L) Index inclusion criteria and monthly review
Criteria for inclusion in the index
Inclusion of Countries currently covered include the US, UK, Canada, Sweden, Italy, Greece, France, Germany and Japan (as of 1
markets in the January 2009). A full set of inflation-linked analytics is required for a bond/country to be added to the index. The index
Global Inflation has a market capitalisation of US$1.4trn and includes 92 issues from 10 governments (as of 31 December 2009).
Linked Index Rating: Countries eligible for inclusion must have an investment grade sovereign rating (Baa3/BBB-/BBB- or above)
using the middle rating of Moody’s, S&P and Fitch, respectively.
Underlying Government domestic inflation-linked debt issued in the domestic currency of that country only. Bonds must be
market/bond capital-indexed and linked to a commonly used domestic inflation index. The notional coupon of a bond must be fixed.
type
Min issue size The following minimum issue sizes are as follows: CAD 200mn, EUR100mn, JPY10bn, SEK100mn, GBP100mn and
USD250mn.
Maturity and Bonds must have a minimum remaining life of atleast one year.
issue
Index calculation and monthly review
Pricing The index uses mid-market prices from local market close and is priced daily. Euro government, US TIPS, UK linkers and
methodology Japanese linker prices come from Barclays Capital market makers. Canadian linker prices are taken from RBC Dominion
and Swedish linker prices are taken from PMI Exchange. Prior to May 2009, US TIPS were priced with bid prices.
All spot and forward foreign exchange rates used are official WM Company mid rates from the London market at 4pm.
Settlement The index uses a T+1 settlement convention for all calculations. At month-end, settlement is the first calendar day of
conventions the next month.
Index frequency The index is calculated daily and has a value for each business day. If the last business day of the month is a public
holiday in one of the major regional markets, prices from the previous business day are used for that particular market.
Review The index is reviewed and rebalanced once a month, on the last calendar day of the month. New bonds and
procedure taps/increases entering the index must have been auctioned on or before the rebalancing date.
Maturity bands Maturity bands are inclusive at the lower bound and exclude the upper bound. Bonds are allocated to maturity bands
based on their remaining time to maturity.
Index holdings The index holding of each bond for the next month is set to the amount outstanding on the review date. The face values
are used rather than an inflation-adjusted value. The indices are weighted using the market capitalisation as standard.
Source: Barclays Capital
Methodology
The Comparator Bond Index methodology is borrowed from the excess returns analysis of
credit markets where excess returns are calculated by making a straight comparison
between returns of a credit bond versus its benchmark. However, this is not so
straightforward when comparing two distinct asset classes such as inflation-linked and
nominal government bonds. Therefore the Comparator Bond Indices, while borrowing the
principle of excess return analysis, are modified to account for the distinct nature of
inflation-linked and nominal securities. Generally comparing broad market, market-value
weighted nominal bond indices in a breakeven inflation (BEI) context with the equivalent
inflation-linked bond index would be sub-optimal since nominal indices tend to be much
broader and have different maturity profiles than their inflation-linked counterparts. To
make for a more effective comparison, the Comparator Bond indices select the comparator
nominal bond utilised for the purposes of BEI. Generally these are of equal credit quality and
similar in maturity to the equivalent inflation-linked bond. In addition, on the last calendar
day of each month, each of the nominal comparator bonds is weighted such that the
market value of each nominal comparator bond matches the market value (inflation-
adjusted) of the associated inflation-linked bond. Thus by selecting the comparator nominal
bonds and custom-weighting each nominal bond equal to that of the equivalent inflation-
linked bond it is possible to make a more effective comparison by avoiding the complexities
of broad market, market-value weighted indices.
The comparator nominal bonds are determined by market convention. Some of the key
considerations while establishing the applicable nominal comparator for each of the
inflation-linked bonds include, but not limited to, closest time-to-maturity as the inflation-
linked bond, issued with the same (or closest) original term as the inflation-linked bond and
liquidity. The nominal comparator bonds generally switch more frequently than the
corresponding inflation-linked bonds given the depth of the nominal market relative to the
inflation-linked.
Considerations
In a pure credit context one would also try to match the duration profile of the target
portfolio; however, because we are dealing with two separate asset classes, this is
problematic. How can we compare linker portfolio duration with that of a nominal bond
portfolio? We could use historical yield beta analysis of each linker verses its benchmark as
an approach to solving this problem but this has some drawbacks:
Betas would have to be re-estimated every month based on available information. This
is not a wholly transparent process; the results would very much depend on the period
chosen for analysis, which cannot be standardised.
Comparator bond indices are available for all of the inflation-linked markets covered within
the World Government, Euro Government and Euro Inflation-Linked Bond Index families.
Popular Composite Comparator Bond Indices are also produced replicating the associated
Inflation-Linked Composite Index, eg, World Government ex Japan Comparator Bond Index.
The Composite Comparator Bond indices also include Greece and associated Composites,
eg, World Government + Greece Comparator Bond Index.
Only those Emerging Market (EM) countries where the nominal government market has
sufficient coverage have comparator bond indices, hence South Africa, South Korea, Poland
and Turkey. The remaining EM inflation-linked markets have less developed nominal bond
markets than inflation. For this reason we have no aggregate or composite Comparator
Bond indices in the EM space.
Comparator bond indices are available in local and foreign currency (hedged and un-hedged).
The following foreign currencies are available for World and sub-indices thereof: AUD, CAD,
CHF, EUR, GBP, JPY, SEK, SGD, USD and ZAR. For South Africa, South Korea, Poland and
Turkey, the following foreign currencies are available: EUR, GBP, JPY, SGD and USD.
Pricing methodology
The Barclays Capital Inflation-Linked family of indices use mid quotes updated daily at local
market close. Please refer to Figure 134, Figure 139, Figure 142 and Figure 148 for the price
sources for each inflation market. All spot and forward foreign exchange rates used are
official WM Company mid rates from the London market close at 4pm London time.
Settlement conventions
The index uses standard settlement and ex-dividend conventions for all calculations. Market
calendars most appropriate for international investors are used.
On non-business days the security price, accrued and analytical values are carried over
unchanged from the previous day. This ensures that the index has no local currency
performance on days when the local market is closed.
Until 30 June 2006, income from coupon was reinvested in the index as soon as it was
received. From 1 July 2006 onwards, income from coupon (and principal payments in the
case of amortising bonds) is held as cash and earns the monthly re-investment rate until the
next rebalancing date. On the next rebalancing date, the coupon payments and interest
earned on it are re-invested back into the index. For all markets the monthly reinvestment is
applied but only markets where there is a suitable deposit rate available does the income
that is received from the coupon accrue. Thus, for all of the EM countries with the exception
of South Africa, coupon payments do not accrue at local money market rates. In the case of
South Africa coupon payments are reinvested using the method described above, less an
accrual rate. Figure 153 lists those countries where monthly reinvestment of coupons
accrue and the associated deposit rates.
13
The calculation and methodologies outlined in this section apply to the WGILB, EGILB, EMGILB and Sterling Inflation-
Linked Indices. Series-L Inflation-Linked Indices follow separate conventions.
Monthly rebalancing
Review procedure
Once a month, on the last calendar day, the indices are reviewed and rebalanced based on
inclusion criteria described earlier in the respective index sections of this guide.
Bonds entering the index for the first time must have settled on or before the review date.
Increases or buybacks to existing bonds are applied according to the same rule.
Bond selection rules are applied and bonds are allocated to the appropriate sector, rating
and/or maturity indices.
Maturity bands
A variety of different maturity bands are utilised. Some markets have a full set of maturities
while others are sparsely distributed across different maturities.
Details of the available maturity bands for each index are given in the respective index
structure and rules section.
Maturity bands are inclusive at the lower bound and exclude the upper bound. Bonds are
allocated to maturity bands based on their remaining time to maturity (for bullet bonds) or
average life (for amortising bonds).
Index holdings
The index holding of each bond for the next month is set to the amount outstanding on the
review date. The face value is used rather than an inflation-adjusted value.
The indices are weighted using market capitalisation as standard. As detailed in the index
formulae section, the weights are adjusted daily to account for price changes, accrued
interest and indexation. The holdings remain fixed for the whole of each calendar month.
Notation
The Barclays return indices are calculated daily and are chain weighted for each day t,
starting with a base value of 100 on the base date.
Ri ,b – The monthly re-investment rate for bond i on business day b, as given in Figure 154.
N i ,b – Face value amount outstanding of bond i on day t. Note this is not inflation-
adjusted but the face value.
Ni, t-1 – Face value amount outstanding of bond i on day t-1, post any capitalisation change.
The notation ∑x
i
i ,t is intended to show a summation occurring over bonds i which are
day b.
Market capitalisation
The Total Market Capitalisation M t of all the constituent bonds on day t is given by:
M t = ∑ mi ,t ,
i
∑ (P i ,t * N i ,b )
CPI t = CPI b i
∑ (P * N i ,b )
*
i ,b
i
The CPI is calculated daily as shown above for all eligible bonds.
∑ (( P i ,t + Ai ,t + X i ,t ) * N i ,b )
GPI t = GPIb i
∑ (( P + Ai ,b + X i ,b ) * N i ,b )
*
i ,b
i
∑ ((P + A + X )* N ) + CH
i ,t i ,t i ,t i ,b t
TRI t = TRI b i
∑ (( P + A + X ) * N )
*
i ,b i ,b i ,b i ,b
i
Income Index
The Income Index ( I t ) is a cumulative figure of coupon income received in the year to date.
It is reset to zero at the beginning of each calendar year. Coupons are added to the income
index for each stock paying a coupon today that was in the index yesterday.
CH t
I t = I b + GPI b *
∑ ( Pi ,b + Ai ,b + X i ,b ) * N i ,b
i
Defining the Income Index in this way ensures that the Total Return Index can also be
calculated by the equation:
(GPI t + ΔI t )
TRI t = TRI b *
GPI b
It can be shown that this calculation of the TRI is perfectly consistent with the previous
equation stated above.
∑ (Y * m * D i ,t i ,t i ,t )
Yt = i
∑ (m * D )
i
i ,t i ,t
where:
Y i,t is the real gross redemption yield (either on an annual or semi-annual basis) of bond i at
the close of day t,
This measure (which includes weighting by duration) is a better approximation of the true
gross redemption yield of all the cash flows in the index, than weighting individual
redemption yields by market capitalisation alone.
∑D i ,t ∗ mi ,t
Dt = i
Mt
where D i,t is the duration of bond i at the close of day t.
∑W i ,t ∗ mi ,t
Wt = i
Mt
where W i,t is the modified duration of bond i at the close of day t. Modified duration quoted
is the sensitivity of the quoted price with respect to changes in real yields.
Average Life
To calculate the Average Life (L t), the life in years of each bond is weighted by its face
value.
∑L *F i ,t i ,t
Lt = i
∑F i
i ,t
and F i,t is the face value of the bond i at the close of day t.
Face Value
To calculate the Face Value (F t) of the index, the face value of each bond is summed across
all bonds in the index.
Ft = ∑ Fi ,t
i
CVt = ∑ Fi ,t ∗ Pi ,t
i
⎛ TRI L,t ⎞
∑ ⎜⎜ TRI × M L ,b * S LM ,b ⎟⎟
∗ ⎝ ⎠
L
TRI M ,t = TRI M ,b
L ,b
∑ (M × S L ,b LM ,b )
L
Where:
S LM ,b is the spot exchange rate between the local currency and the currency of the Multi-
Currency Index on day b,
M L ,b is the local total market value at time b, and summation is over all countries in the
Multi-Currency Index.
The formula below is used to calculate the Foreign Total Return Index. The same technique
is used to calculate the associated clean and gross price indices.
S LF ,t
TRI F ,t = TRI L ,t *
S LF ,c
Where:
S LF ,c = Spot exchange rate between local and foreign currency at commencement date of
the index.
This has the advantage of being the simplest method of the three to apply to the index;
however, it has the major disadvantage that it would be impossible to replicate in practice.
As this method would violate the general principle that the index be replicable, the perfect
foresight method is not used for calculating the hedged returns.
This method is better in theory than the current value method, as it provides a better match
to the true currency exposure. This increase in accuracy involves significantly more complex
calculations to estimate the end-of-month currency exposure, but only provides a marginal
reduction in the currency mismatch.
The current value method is relatively straightforward to apply and is easily replicable, and
this is the method used for the Index.
Local return
this reinvestment strategy, we cannot observe the local currency return in the same way as
described above. Instead we use a simpler breakdown:
Profit or less on the hedge itself (or in this case on a series of currency hedges).
Note that, as stated above, this is a daily estimation of the Monthly-Hedged Index and not a
true Daily-Hedged Index. This is to provide continuity between the monthly and daily
hedged total return series.
The easiest way to explain this is to look at an example. Suppose we are 10 days into the month
and that the last business day of this month is the 28th. Here we need to offset the starting one-
month forward with an 18-day forward (ie, 28 –10 = 18 days). In theory we could obtain an 18-
day rate directly for the forward market, but in practice only certain periods (tenors) are quoted,
and we need to use interpolation to arrive at a rate for the desired period.
For the sake of simplicity we use a linear interpolation based on the current one-month forward
rate and spot rate. In our example we would calculate the 18-day forward rate as the current
spot rate plus the premium or discount between spot and 1-mth forward pro-rated for 18 days.
S LF ,e
Currency Return = CR = −1
S LF , s
FLF , s ,1M
Forward Return = FR = −1
S LF , s
Hedge Return = FR – CR
Hedged Index Value = Start Hedged Index Value * (1+Hedged Index Return)
Where:
s – Start date
e – End date
TRIL – Local Currency Total Return Index
S LF – Spot foreign exchange rate between local currency and the hedge currency
F LF, 1M – One-month forward foreign exchange rate between local currency and the hedge currency.
TRI L ,e
MTD Local Return = MTD LR = −1
TRI L , s
S LF ,e
MTD Currency Return = MTD CR = −1
S LF , s
FLF , s ,1M
Forward Return = FR = −1
S LF , s
S LF ,i
Hedge Reversal Return = −1
FLF ,i , R
MTD Hedged Index Return = MTD LR + (1+MTD LR) * MTD CR + MTD Hedge Return
Hedged Index Value = Start of Month Hedged Index Value * (1 + MTD Hedged Index Return)
Where:
i – Intra-month date;
F LF, i, R – Forward FX rate local currency into hedge currency on day i for forward
period R. This is calculated by linear interpolation between the spot rate S LF, s and the one-
month forward rate FLF, s, 1M where 1 < R < 1M.
TRI M ,e
Un-hedged Index Return = −1
TRI M ,b
⎡ FLM,b, 1M - SLF, e ⎤
Hedge Return = ∑ ⎢W L ,b ×
S LF ,b
⎥
L ⎣ ⎦
Hedged Index Return = Un-hedged Index Return + Hedge Return
TRI M ,e ⎡ F -S ⎤
= − 1 + ∑ ⎢WL ,b × LM,b, 1M, LF, e ⎥
TRI M ,b L ⎣ S LF ,b ⎦
15 March 2010 188
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Hedged Index Value = Start Hedged Index Value * (1 + Hedged Index Return)
Where:
b – start date
e – end date
F LF, b, 1M – One-month forward FX rate local currency into hedge currency on day b
M L ,b × S LF ,b
WL ,b =
∑ (M L,b × S LF ,b )
L
⎡ ⎛S ⎞⎤
MTD Currency Return = ∑ ⎢W L ,b × ⎜⎜ LF, i − 1⎟⎟⎥
L ⎢⎣ ⎝ S LF ,b ⎠⎥⎦
⎡ ⎛ FLF,b, 1M ⎞⎤
Forward Return = ∑L L,b ⎜⎜ S − 1⎟⎟⎥
⎢ W ×
⎣⎢ ⎝ LF ,b ⎠⎥⎦
⎡ ⎛ S ⎞⎤
Hedge Reversal Return = ∑ ⎢W L ,b × ⎜⎜ LF ,i − 1⎟⎟⎥
L ⎢⎣ ⎝ FLF, i, R, ⎠⎥⎦
MTD Hedge Return = Forward Return + Hedge Reversal Return – MTD Currency Return
MTD Hedged Index Return = MTD Un-hedged Index Return + MTD Hedge Return
Hedged Index Value = Start of Month Hedged Index Value x (1 + MTD Hedged Index Return)
Where:
F LF, i, R, – R period Forward FX rate local currency into hedge currency on day i
M L ,b × S LF ,b
WL ,b =
∑ (M L,b × S LF ,b )
L
Overview
In addition to the flagship benchmark inflation-linked indices profiled earlier, Barclays
Capital calculates and publishes a number of tradable inflation-linked index products
designed to offer efficient market access and/or rules-based alpha generation in index
form. These indices include a family of inflation swaps indices, breakeven inflation index
products, and other access products designed to offer inflation exposure to less liquid
markets. This section will profile some of our more notable inflation-linked index products.
Inflation Swaps – Constant Maturity swaps at different tenors for six markets. Variants
include zero coupon and real return indices and are available as both unfunded (excess)
and funded (total)returns.
− US
− UK
− Euro
− France
− Japan
INSTEP
Real Income
AIMS
On the last business day of each month a swap is initiated at the prevailing market rate for a
specific tenor. The position is marked to market every day as it decays over the next month.
At the next rebalance, the 1m decayed position is unwound and a new swap is initiated at
that day’s prevailing market rate for the same original tenor as before.
Indices are calculated for various tenors across the term structure and for the following
inflation markets:
Excess return indices are generated from pure swap performance. Total return indices are
calculated as well and assume an additional cash funding component.
These swap indices can be used as customized liability management tools against which to
benchmark inflation sensitive portfolios. They can also be used as components in strategy
index products.
14
Calculations and Methodology apply only to Series-B Inflation Swap Indices
Figure 155: Inflation swap indices Figure 156: Real rate swap indices
115 125
110 120
105 115
100 110
95 105
90 100
85 95
90
80
Oct 06 Apr 07 Oct 07 Apr 08 Oct 08 Apr 09 Oct 09 85
Oct 06 Apr 07 Oct 07 Apr 08 Oct 08 Apr 09 Oct 09
US 10yr inflation swap index ER
US 10yr real rate swap index ER
UK 10yr inflation swap index ER rebased UK 10yr real rate swap index ER rebased
EUR 10yr inflation swap index ER rebased EUR 10yr real rate swap index ER rebased
Source: Barclays Capital Source: Barclays Capital
Swap structures
The inflation payer agrees to pay the cumulative percentage change in a referenced and
lagged price index over the life of the swap. The inflation receiver in turn pays a fixed
compounded rate. This is the breakeven inflation rate (BE) calculated by setting the initial
net present value of the swap to zero. The fixed rate is the market breakeven of future
inflation over the horizon of the trade.
The swap NPV depends on the CPI ratio, the fixed breakeven rate (BE), the notional and a
nominal discount factor (DF). These swaps are exposed to realized and expected inflation.
⎡ C P I ma t
N × D Fm at × ⎢
⎤
[
− 1 ⎥ − N × D F m a t × ( 1 + B E )T e no r − 1 ]
⎣ CP I ba se ⎦
1
⎡ C P I m a t ⎤ T e no r
⎢ ⎥ − 1 = BE
⎣ CP I ba se ⎦
Over the life of the swap, the inflation payer agrees to pay the cumulative percentage
change in a referenced and lagged price index multiplied by a fixed real rate (RR). The
inflation receiver in turn pays a compounded Libor-based rate (Lib). The real rate is
calculated on the trade date by setting the initial net present value of the swap to zero. DCF
is the day-count fraction.
Increasingly, pension fund liabilities are becoming linked to real yields and not just inflation.
These swaps synthetically replicate TIPS style exposure since the NPV is driven by real yields
and inflation.
⎡ CPImat ⎤ ⎡Tenor ⎤
N × DFmat × ⎢
⎣ CPIbase ⎦
∏
×( 1 + RR )Tenor − 1⎥ − N × DFmat × ⎢ ( 1 + Lib× DCF) − 1⎥
⎢⎣ s ⎥⎦
Figure 157: An inflation swap transaction Figure 158: A real rate inflation swap transaction
Fixed Breakeven Rate Libor compounded
Actual
Inflation
Barclays
The index level on any index business day is derived from the index level at last rebalance
and the NPV of the underlying swap.
The NPV is calculated using Barclays Capital analytics and closing curves. The official
closing inflation curves utilise proprietary seasonality models.
All of the indices assume one is receiving inflation and paying a fixed breakeven rate on the
underlying swap.
The cash return applied for each market comes from a money market deposit with a tenor
commensurate with the floating frequency of a standard interest rate swap in that market.
The cash deposit is marked to market each day using interpolated fixes. It rebalances in the
same way and at the same time as the underlying swaps.
For example, the EUR swaps assume a cash return based on a 6m EURIBOR deposit held for
one month.
Roll convention MF MF MF MF MF
Index calendar London & TARGET London & New York London Sydney, London & Tokyo London & Tokyo
Figure 160: Total return and excess return inflation swap indices Figure 161: Japan 10y inflation swap index versus breakevens
EUR 10yr inflation swap index ER Japan 10yr swap breakeven yields (RHS)
Source: Barclays Capital Source: Barclays Capital
Additional information
Swap present value are marked to market using standard market conventions.
The indices are priced using Barclays Capital official closing curves.
The indices are published to four decimal places. Levels and supporting analytics are
available on the index website, https://live.barcap.com – keyword index.
EUR, UK and US swap indices, including nominal swap indices, are also available on
Bloomberg page BSWM. BSWI is the sub page for excess return inflation swaps, and
FSWI is the sub page for the total return indices.
Please contact the Index, Portfolio & Risk Solutions team for more information.
The indices aforementioned refer to the Barclays Capital Series-B inflation swap indices.
This document is an update to “The Barclays Capital Inflation swap Indices Family”
index document published on 9 November 2007. Please also see the funding
methodology amendment document published by the Index Products Group Research
on 6 March 2009.
Since the index is constructed using inflation swaps the index performance is exposed to
both realized and expected inflation.
Barclays Capital Zero Coupon US Inflation Swap indices, referencing US CPI, are weighted
each month to give a composite return that drives the performance of TWIST. The
weighting is derived from the amount of outstanding bonds, split by certain maturity
bucket, in the Barclays Capital US Government Inflation-Linked Bond Index (the US TIPS
index). The weighting scheme reflects broad exposure across the inflation curve and also
means the index is sensitive to changes in linker issuance patterns.
The TWIST Index is an excess return index and is tradable. The index delivers liquid
breakeven inflation exposure in a cost-efficient manner as compared to other approaches.
The TWIST Index can be used as a swap breakeven inflation benchmark or as an overlay
tool for investors wishing to access inflation as a hedge for their portfolios.
Figure 162: The TWIST Index and real swap rates Figure 163: Monthly returns correlations against TWIST index
inflation swap is an instrument that exchanges a fixed breakeven rate for actual future
inflation.
Each inflation swap index replicates the performance of continuously investing in zero-
coupon breakeven inflation swaps.
The holder of the swap receives inflation and pays the fixed breakeven rate.
The index includes TIPS with one or more years remaining until maturity.
TIPS in the index have a total outstanding issue size of $500mn or more.
Figure 165: Barclays Capital US inflation swap indices Figure 166: US Government Inflation-Linked index
110 245
105 225
100 205
95 185
90 165
85 145
80 125
75 105
Oct-06 Apr-07 Oct-07 Apr-08 Oct-08 Apr-09 Oct-09
85
USD 2yr inflation swap index
USD 5yr inflation swap index Feb 97 Feb 99 Feb 01 Feb 03 Feb 05 Feb 07 Feb 09
USD 10yr inflation swap index
USD 20yr inflation swap index US TIPS all maturities TR index
Index return is a composite return measure from the last rebalance date, as per the
weighting scheme.
The face value weighting is re-evaluated on the last business day of each month when
the index rebalances.
Note that the 3-5y and 5-7y buckets of the TIPS index are both matched off against the
5y Inflation Swap index.
US Government Inflation-Linked 1-3y bucket USD Zero Coupon Inflation Swap 2y ER index
US Government Inflation-Linked 3-5y and 5-7y bucket USD Zero Coupon Inflation Swap 5y ER index
US Government Inflation-Linked 7-10y bucket USD Zero Coupon Inflation Swap 10y ER index
US Government Inflation-Linked >10y bucket USD Zero Coupon Inflation Swap 20y ER index
Source: Barclays Capital
Figure 168: Historical swap index weighting for TWIST Index Figure 169: Weighting as of end-February 2010
0.40
USD 2yr
0.35
17%
0.30 USD 20yr
0.25 29%
0.20
0.15
0.10
0.05
0.00
USD 5yr
Oct-06 Apr-07 Oct-07 Apr-08 Oct-08 Apr-09 Oct-09
34%
USD 2yr USD 5yr USD 10yr
USD 10yr USD 20yr 20%
Figure 170: TWIST versus swap breakeven rates Figure 171: TWIST versus bond breakeven rates
Additional information
The index inception date is 31 October 2006.
The index is published on New York and London business days. The index rebalances at
month-end.
Index levels and analytics are available on the Barclays Capital Index Website,
https://live.barcap.com, keyword:index
Please contact the Index, Portfolio & Risk Solutions for more information.
Anand Venkataraman The Barclays Capital US breakeven inflation benchmark indices are designed to
+44 (0) 20 7773 0852 represent a breakeven inflation position by taking positions across the term structure of
anand.venkataraman the on-the-run (OTR) TIPS market, while minimising exposure to real yields
@barcap.com
Index rationale
Scott Harman
Investors seeking a hedge for inflation have several options at their disposal, including
+44 (0) 20 7773 1775
investments in commodities, long positions in inflation-protected securities, or the use of
scott.harman@barcap.com
inflation swaps. Another investment that has shown an even higher correlation with
inflation is an investment in breakeven inflation, which in principle is the rate of inflation
that will equate the returns on an inflation-linked bond and a “comparator” nominal bond
issue of the same term. 15 Such an investment in breakeven inflation can be obtained by
taking a long position in inflation-linked securities and a short position in a suitable nominal
comparator bond. This methodology forms the basis for construction of the Barclays Capital
US Breakeven Inflation Benchmark Indices.
Index description
The Barclays Capital US breakeven inflation benchmark indices are designed to provide
access to breakeven inflation by capturing the returns of a simultaneous long position in
inflation-linked securities (represented by OTR US TIPS) and a short position (facilitated by
a repo agreement) in suitable nominal comparator US Treasury bonds. 16 The indices also
aim to minimise exposure to real yields by scaling the total returns of the nominal
comparator Treasury bonds by a ratio equal to the duration of the OTR TIPS to its
corresponding nominal comparator Treasury bond.
15
How to Hedge Inflation...let us count the ways”, Global Rates Weekly, Barclays Capital, 23 July 2009.
16 On-the-run represents the most recently issued TIPS of a given tenor.
Treasury Bond
bond Cash
Cash
Breakeven
Inflation
Position
TIPS
Treasury Bond
bond Cash
Repo Seller
Indices represented by long positions in OTR TIPS and short positions in nominal
comparator Treasury bonds
5yr TIPS OTR 5yr Nom 10yr TIPS 10yr Nom 20yr TIPS 20yr Nom New Tenor New
Index Index OTR Index Index OTR Index Index (TIPS) Tenor(Nom)
Source: Barclays Capital
This index, at any given tenor point (for Instance, 5y), is calculated using the following equation:
Where
TIPS i ,t is the total return between t and b of a single bond index consisting of the OTR TIPS
Nomt ,b
is the total return between t and b of a single bond index consisting of the
nominal comparator Treasury bond
Rt ,b
is the repo return component between t and b
SF b
is the scaling factor and is calculated as:
⎛ LD ⎞
SF b = ⎜⎜ b
⎟⎟
⎝ ND b ⎠
Where
LDb is the real semi-annual modified duration for OTR TIPS at the tenor
NDb is the semi-annual modified duration of the corresponding nominal comparator bond
The repo return component of the index is designed to represent the return on cash “lent”
as part of the reverse repo agreement. A new reverse repo agreement is entered into at each
rebalance date maturing at the next index rebalance date (usually one month) since the
indices rebalance monthly, as described below.
⎛ repo _ period ⎞
1+ ⎜ rb ,1m × ⎟
R t ,b = ⎝ 360 ⎠ −1
⎛ days _ till _ repo _ period _ end ⎞
1+ ⎜ rt × ⎟
⎝ 360 ⎠
Where
Rt ,b
is the repo return component between t and b
rb,1m
is the one-month general collateral repo rate
rt
is the interpolated repo rate on day t. The interpolated repo rate on any day is
calculated based on days remaining until end of prevailing repo period by linear
interpolation of prevailing day’s overnight, one-week, two-week, three-week and
one-month general collateral repo rates as applicable.
repo _ period is the actual length in days of the repo period at the previous rebalancing date
(ie, days between previous rebalancing date and the following rebalancing date)
days _ till _ repo _ period _ end is the number of days remaining as on t till the end of
the prevailing repo period
⎧n ⎫
ILC t = ILC b × ⎨∑ [(TIPSi ,t − SFi × (Nomi ,t − Rt ,b )]× w i ,b + 1⎬
⎩ i =1 ⎭
Where:
ILC b
is the Aggregate Index level at previous rebalance day b
TIPS i ,t is the total return of the single bond TIPS total return index at a given tenor on day
t since last rebalance day b
Nomi ,t is the total return of the single nominal comparator bond index for the
corresponding tenor on day t since last rebalance day b
w i ,b
is the weight of an individual tenor within the Aggregate Index. Note that each
tenor point is equally weighted
Rt , b
is the repo return component as on day t
SFi is the scaling factor for a given tenor calculated as described earlier
n is the number of available tenor points on the OTR TIPS market term structure
At inception, the Barclays Capital US Breakeven Inflation Aggregate Index had equal-
weighted exposure to the three prevailing OTR tenor points: the 5y, 10y and 20y points.
Index rebalancing
Presently, the Barclays Capital US Breakeven Inflation Benchmark Index family comprises 10
separate indices (Figure 174), all of which rebalance on the last calendar day of each month
in line with conventional bond Indices.
At each monthly rebalancing, the Breakeven Inflation Index at each individual tenor point
would switch the underlying inflation-linked security only in case a new TIPS is issued at the
given tenor point, thereby becoming the prevailing OTR TIPS. In the event that the Treasury
establishes a new tenor point at which TIPS will be issued (for instance, the new 30y in
February 2010), the new tenor would be included in the Aggregate Index at the subsequent
monthly rebalancing. Conversely, a tenor point would be removed from the Aggregate
Index if there had been no new TIPS issuance for a 2y period at the tenor point. Upon the
two-year anniversary since the last occasion a TIPS was issued at a given tenor point, that
tenor point would be removed from the Barclays Capital US Breakeven Inflation Aggregate
Index at monthly rebalancing, and all remaining tenors will be equally weighted.
Notwithstanding, the Aggregate Index would weight all the tenor points equally at all times.
To reflect the mechanics of a breakeven inflation position consistently, the applicable nominal
comparator Treasury bond is determined at each monthly rebalancing based on market
conventions. Some key considerations while establishing the applicable nominal comparator
Treasury bond for an OTR TIPS are bond(s) having the closest time-to-maturity as the OTR TIPS,
bond(s) issued with the same (or closest) original term as the OTR TIPS, and liquidity among
other determinants. The nominal comparator bonds generally switch more frequently than the
corresponding OTR TIPS, due to the depth of the Treasury market relative to the TIPS market.17
17
Please refer to the Inflation-Linked Daily for contemporary comparator bond listings.
The index provides beta-style access to breakeven inflation in the euro government
bond market.
The linker and nominal bonds exposure is rates-duration-hedged to minimise real yield risk.
The short position is based on borrowing nominal bonds in the market and having the
cash lent earn a repo rate.
The future path of the INSTEP Index is sensitive to changes in future inflation
expectations and to the actual path of inflation indexation.
The index is a bond space alternative to breakevens and is similar to the Barclays Capital
Inflation Swap Indices (series-B), which replicate the performance of investing in zero-
coupon inflation swaps in which one party pays a fixed rate and receives the cumulative
percentage change in a price index over the life of the swap.
Figure 175: NSTEP Index historical performance Figure 176: Chart Title
140 114
135 112
130 110
125 108
120
106
115
104
110
102
105
100 100
95 98
90 Sep 06 M ar 07 Sep 07 M ar 08 Sep 08 M ar 09 Sep 09
Jan 00 Jul 01 Jan 03 Jul 04 Jan 06 Jul 07 Jan 09
EUR 10yr ZC Inflati on Sw ap index (TR)
INSTEP index INSTEP Index rebased
Euro gov inflation-linked index average real yield -0.236 Feb 00 - Nov 09
A nominal bond, given the above, is sensitive to real yields and forward-looking inflation
expectations.
On each rebalance date, the INSTEP index goes long the linkers index and short, by a duration-
adjusted amount, the nominal breakeven comparator index. The real yield components offset
each other, and the net position is insensitive to changes in real yields.
This applies on an average annual-modified-duration basis since the index durations we are
using to hedge are average durations for all maturities indices.
Also, due to the repo component, the net index position has a small sensitivity to 1m rates.
Additional information
The INSTEP Index inception date is 31 January 2000.
European Banking Federation repo fixings (Bloomberg page EBF) are used for the index
calculation. However, for the back-test prior to March 2002, BBA fixes were used.
The index is published with a one day lag on the Barclays Capital Index Website at
https://.live.barcap.com (keyword: index). The index is published to four decimal places.
The original version of the INSTEP Index, based on a prior methodology, was launched
in August 2008.
Please contact the Index, Portfolio & Risk Solutions team for more information.
CAD INSPIRE
Michalis Christodoulou The Barclays Capital CAD INSPIRE Index family is designed to provide synthetic inflation
+44 (0)20 3134 2751 protection for Canada using an optimised weighted combination of actively traded and
michalis.christodoulou liquid inflation swap indices from the US, UK and euro area.
@barcap.com
The Canadian inflation market is relatively small and illiquid, despite significant demand
for inflation protection from both domestic pension funds and international investors.
Yuan Tian
+44 (0)20 7773 1426 The Canadian government started issuing inflation-linked bonds in 1991, with currently
yuan.tian@barcap.com five issues outstanding and a face value of CAD28bn,
Marcela Barreto Non-government inflation-linked bond issuance has been slow to develop in Canada, with
+44 (0)20 3134 0750 the amount of corporate issuance being rather negligible as of the end of 2009.
marcela.barreto@barcap.com
Figure 178: Canada inflation is highly correlated with Figure 179: Canada inflation-linked bond market breakdown
developed inflation markets
0.25 3.56%
CAD yoy inflation
0.10
0.05
0.00
-0.05 77.66%
Jan-76 Jan-81 Jan-86 Jan-91 Jan-96 Jan-01 Jan-06
The lack of an inflation swap market in Canada makes it challenging for investors to
create economically efficient portfolios and hedge against inflationary trends. Demand
for a Canadian inflation derivatives market is so strong at the moment, that some
Canadian investors have even considered going to the US inflation swap market as a
proxy for Canadian inflation.
Barclays Capital offers an innovative solution to tackle such a challenge: The CAD
INSPIRE Index Family takes advantage of Canadian inflation being correlated with
developed inflation markets.
Canadian Inflation exhibits significant time-varying correlation features with US, UK and
EUR inflation, justifying the use of a rolling optimisation model.
Inflation being a global phenomenon implies that inflationary trends are shared across different
regions which will in turn feed into market expectations on future inflation and thus breakevens.
The spirit of the INSPIRE framework is to replicate inflationary trends in illiquid countries,
using liquid markets’ inflation instruments. Taking Canada as our illiquid inflation market,
the INSPIRE framework follows a two-step process:
Minimising transaction costs embedded in the index while achieving the best possible
correlation features.
The illiquid nature of the Canadian inflation market calls for an alternative inflation
measure to breakevens.
Therefore, an inflation measure has been constructed that reflects similar statistical
properties to the breakevens.
Forward interest rates contain information about average expected inflation and
expected real rate for a future period.
The 20y Canadian inflation measure shown is in line with inflationary developments
worldwide.
Figure 180: CAD 20y inflation measure Figure 181: CAD INSPIRE fit versus inflation measure
4.5% 4.5%
CAD 20yr inflation measure
4.0% CAD 20yr inflation measure 4.0%
3.5% 3.5% CAD INSPIRE replica
3.0% 3.0%
2.5% 2.5%
2.0% 2.0%
1.5% 1.5%
1.0% 1.0%
0.5% 0.5%
0.0% 0.0%
May-00 May-02 May-04 May-06 May-08 Apr-01 Apr-03 Apr-05 Apr-07 Apr-09
The correlation achieved between our Canadian inflation measure and its corresponding
US, UK and EUR breakeven replica is around 80-95%, indicating a very good fit
throughout the sample for various tenors.
The resulting monthly weights are then applied to the returns of the zero-coupon
inflation swap indices of US, UK and EUR to produce the CAD INSPIRE Index return.
Additional information
Barclays Capital CAD INSPIRE indices start on 30 January 2004.
The CAD INSPIRE Index is offered in 5y, 10y and 20y maturities and returns are
expressed in AUD. It can be offered at any additional tenor on request and can be used
flexibly according to investors’ needs.
Clients, once given permission, can access the index data on Bloomberg.
Type BXIICI10 Index <GO> for the CAD INSPIRE 10y Index
Type BXIICI20 Index <GO> for the CAD INSPIRE 20y Index
AUD INSPIRE
Michalis Christodoulou The Barclays Capital AUD INSPIRE Index family is designed to provide synthetic inflation
+44 (0)20 3134 2751 protection for Australia using an optimised weighted combination of actively traded and
michalis.christodoulou liquid inflation swap indices from the US, UK and euro area.
@barcap.com
The Australian inflation market is relatively under-developed and illiquid. The current
outstanding notional of AUD inflation-linked bonds issued is approximately AUD24.6bn.
Yuan Tian
+44 (0)20 7773 1426 Over the past seven years, the Australian government has issued only one inflation-
yuan.tian@barcap.com linked bond. There are four such bonds currently outstanding with a total face value of
AUD10.3bn, which accounts for 42% of the total inflation issuance.
Marcela Barreto
The Australian market has a large number of small issuers that issue inflation-linked bonds.
+44 (0)20 3134 0750
marcela.barreto@barcap.com The AUD inflation swap market started in 2006, and trading activities have been relatively
sporadic, with limited liquidity and no standard benchmark.
Figure 182: Australian inflation is highly correlated with Figure 183: Australian inflation-linked bond current market
developed inflation markets breakdown
5%
0%
-5%
Mar-76 Mar-81 Mar-86 Mar-91 Mar-96 Mar-01 Mar-06 29%
Source: Barclays Capital Source: Barclays Capital
Given the illiquidity of the Australian inflation swap market, it is challenging for investors
to create economically efficient portfolios and hedge against Inflationary trends.
Barclays Capital offers an innovative solution to tackle such a challenge: The AUD
INSPIRE Index Family.
The AUD INSPIRE Index takes advantage of Australian inflation being correlated with
developed inflation markets.
Australian Inflation exhibits significant time-varying correlation features with US, UK and
EUR inflation, justifying the use of a rolling optimisation model.
Inflation being a global phenomenon implies that inflationary trends are shared across
different regions, which will in turn feed into market expectations on future inflation and
thus breakevens.
The spirit of the INSPIRE framework is to replicate inflationary trends in illiquid countries,
using liquid markets’ inflation instruments. Taking Australia as our illiquid inflation market,
the INSPIRE framework follows a two-step process:
model 95
Core
Countries BE
US, UK, EUR 90
85
Jan 04 Nov 04 Sep 05 Jul 06 May 07 Mar 08 Jan 09 Nov 09
Minimising transaction costs embedded in the index while achieving the best possible
correlation features.
The illiquid nature of the Australian inflation market calls for an alternative inflation
measure to breakevens.
Therefore, an inflation measure has been constructed that reflects similar statistical
properties to the breakevens.
Forward interest rates contain information about average expected inflation and the
expected real rate for a future period.
The 10y Australian inflation measure shown is in line with inflationary developments
worldwide.
Figure 184: AUD 10y inflation measure Figure 185: AUD INSPIRE fit versus inflation measure
3.0% 3.0%
2.0% 2.0%
1.0% 1.0%
0.0% 0.0%
Apr-01 Jun-02 Aug-03 Oct-04 Dec-05 Feb-07 Apr-08 Jun-09 Apr-01 Jun-02 Aug-03Oct-04 Dec-05Feb-07 Apr-08 Jun-09
The correlation achieved between our Australian inflation measure and its
corresponding US, UK and EUR breakeven replica is around 80-90%, indicating a very
good fit throughout the sample for various tenors.
The resulting monthly weights are then applied to the returns of the zero-coupon
inflation swap indices of US, UK and EUR to produce the AUD INSPIRE Index return.
Additional information
Barclays Capital AUD INSPIRE indices start on 30 January 2004.
The AUD INSPIRE Index is offered in 5y, 10y and 20y maturities and returns are
expressed in AUD. It can be offered at any additional tenor upon request and can be
used flexibly according to investors’ needs.
Clients, once given permission, can access the index data on Bloomberg.
Type BXIIAII10 Index <GO> for the AUD INSPIRE 10y Index
Type BXIIAII20 Index <GO> for the AUD INSPIRE 20y Index
The stream of cash flows has a specific maturity date and is constructed in a way that the
invested principal is depleted over the index tenor. The rate of the constant real cash flows is
set at a level that makes the present value of the cash flows equal to par, and it is adjusted
bi-weekly to reflect changes in real rates.
Once the size of the cash flows is determined, a portfolio of US TIPS bonds is constructed
aiming to have the cash flows of the portfolio as close as possible to the targeted monthly
cash flows. The portfolio of replicating US TIPS is determined on each index rebalance date
using an algorithm based on cash flow matching and risk minimization.
The Barclays Capital US TIPS Real Income Index family consists of the Barclays Capital US
TIPS Real Income 2019 index and the Barclays Capital US TIPS Real Income 2029 Index.
The Barclays Capital US TIPS Real Income indices are designed for investors seeking an
investment solution that provides steady inflation-protected cash flows over a specific
period of time.
Bi-weekly rebalancing.
TIPS coupons are immediately reinvested into the index upon payment.
Eligible bonds are US TIPS issued on or before the Index Rebalance Day and maturing at
least 20 calendar days after the Index Rebalance Day.
As compared with market value weighted TIPS indices, the Barclays Capital US TIPS Real
Income indices have the advantage of not being influenced by the government bond
issuance pattern. The US TIPS Real Income Index bond weights are determined with the aim
of generating a flat real cash flow profile over the index tenor. On the other hand, market
value weights are calculated using the outstanding amount and the market price of bonds,
therefore creating a cash flow pattern that may deviate significantly from the targeted
constant cash flows.
Figure 186 compares the annual real cash flows (coupon + principal) of the Barclays Capital
1-10 year US TIPS Index versus the US TIPS Real Income 2019 Index, assuming the current
index holdings (as of October 2009 ) do not change over time.
18%
12%
10%
8%
2010 2011 2012 2013 2014 2015 2016 2017
Source: Barclays Capital
Index framework
Step 1: Determination of the target real constant cash flows using information from the real
curve.
Step 2: Selection of eligible TIPS from the universe of TIPS bonds to replicate the target real
constant cash flows.
Optimization Algorithm
Cash flow Matching Risk Minimization
Additional Information
The base date of the Barclays Capital US TIPS Real Income 2019/2029 indices is 14
October 2009. The termination dates of the Barclays Capital US TIPS Real Income
2019/2029 indices are 31 October 2019/31 October 2029, respectively. The indices
follow the New York business day calendar.
Bloomberg Tickers: Barclays Capital US TIPS Real Income 2019 Index (BXIIRI19 Index),
Barclays Capital US TIPS Real Income 2029 Index (BXIIRI29 Index).
AIMS Index
Yuan Tian The Barclays Capital AIMS (Algorithmic Inflation Momentum Switching) Index seeks to
+44 (0) 20 7773 1436 enhance returns generated by a long-only portfolio of inflation-linked securities.
yuan.tian@barcap.com
Jose Mazoy
Index rationale
+44 (0) 20 3134 0998 Investors who are concerned that realized inflation risk would erode the value of their
jose.mazoy@barcap.com nominal fixed income investment tend to focus on inflation-linked assets, which allow them
to retain the real value of their investment. One way of tracking inflation is through Treasury
Inflation Protected Securities (TIPS), issued by the US Treasury, which pay a fixed coupon on
a notional that accretes according to the US CPI-U Inflation Index.
The nominal investment grade bond market and the TIPS market have long exhibited
imbalances in supply and demand. The nominal Treasuries market is a substantial multiple
of the size of the TIPS market. During periods of heightened inflation expectations, stronger
price pressure can be observed on TIPS. This price pressure caused by size imbalances
presents an opportunity to extract excess return provided we identify the changes in current
and future inflation expectations in a timely manner.
Historically, TIPS outperformed nominal Treasury bonds during periods of rising breakeven
inflation rates, but underperformed nominal Treasuries in 2008 due to a flight to quality and
the relative illiquidity of TIPS compared with nominal Treasuries. The Barclays Capital AIMS
Index seeks to enhance returns generated by a long-only portfolio of inflation-linked
securities by adopting a fully non-discretionary algorithm. The AIMS algorithm uses publicly
available data and exploits the link between US inflation expectations and movements in
TIPS prices. The index takes a long position in a TIPS basket as default, but switches to a
short TIPS position if the signal indicates lowered inflation expectations. The rationale of the
underlying AIMS strategy is that investors switch from nominal Treasuries (and other
bonds) to TIPS when concerned about inflation, but switch to short TIPS positions if
inflation expectations fall.
Figure 187:US Treasury versus TIPS Index market capitalization Figure 188: US Treasury Index TRI versus TIPS Index TRI
5,000 240
4,500 220
4,000
200
3,500
3,000 180
2,500 160
2,000
140
1,500
1,000 120
500 100
0 80
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Index framework
Inflation concerns are both determined by the current realization of inflation as well as
future inflation expectations. The current inflationary trend can be monitored by the
evolution of the CPI y/y print. The TIPS principal and coupon are also linked to changes in
the CPI: as CPI rises, the principal in the TIPS is adjusted upward and the interest on the
bond is then accrued on a higher principal. Therefore, any increase or decrease in the
current CPI will immediately be priced into the TIPS market value.
Future inflation expectations are not directly observable. Inflation surveys or models could be
used to try to infer its value. We use a simpler approach in the AIMS index and take the slope of
the nominal yield curve as a measure of future inflation expectations, since a steep yield curve
reflects the expectation of higher nominal rates in the future and therefore higher inflation.
Figure 189: US TIPS y/y return and CPI y/y Index Figure 190: Swap slope as an approximation for future inflation
The AIMS Index has taken a short TIPS position 20% of the time since its inception in 1997
and historically. It has exhibited a bias towards being long TIPS. The index has mostly
outperformed the benchmark long TIPS portfolio during periods in which it has taken a short
position. It is worth noting that during the sharp global recession in 2008-09, the AIMS
allocation was mostly long; this was due to the fact that the signal determination takes into
account the CPI y/y print as well as the slope of the swap curve. Despite falling CPI y/y print, a
steepening yield curve may be signalling rising long-term inflation expectations.
6% 1.2
5% 1.0
0.8
4%
0.6
3%
0.4
2% 0.2
1% 0.0
-0.2
0%
-0.4
-1%
-0.6
-2% -0.8
-3% -1.0
Mar-97 Sep-98 Mar-00 Sep-01 Mar-03 Sep-04 Mar-06 Sep-07 Mar-09
Monthly Allocation CPI YOY Slope
Source: Barclays Capital
Figure 192: AIMS Excess Return Index and Long TIPS Excess Figure 193: AIMS Excess Return Index return distribution
Return Index
200 45%
AIMS return
40% distribution
180
35%
160 30%
140 25%
20%
120
15%
100 10%
5%
80
0%
Mar-97 Mar-99 Mar-01 Mar-03 Mar-05 Mar-07 Mar-09
-6.4% -4.1% -1.8% 0.4% 2.7% 5.0% 7.2%
AIMS ER Index Long only TIPS ER Index
Index characteristics
The Barclays Capital AIMS Index is available in USD; AIMS Index Inception date is 19
March 1997.
The AIMS Index is available in total return and excess return versions.
The CPI y/y change is calculated by taking the y/y percentage change on the US CPI for all
urban consumers. This number is published at 8:30 EST at about the middle of each month.
The US swap 2s10s slope is calculated using the US ISDA fixings published at 11:30 EST.
INFLATION THEMES
f = inflationary expectations
p = risk premium
The risk premium reflects the assumption that investors want additional compensation for
accepting undesirable inflation risk when holding nominal bonds, but in practice this is
usually offset by a liquidity discount, ie, the yield premium investors demand in order to
hold an inflation-linked bond rather than a nominal because the latter is usually more liquid.
If inflation and interest rates are relatively low, the formula can be approximated with an
additive form:
n=r+f+p
However, in practice separating out the inflation expectations and risk premium component
is extremely complex. From an investor’s perspective differentiating between the elements
is relatively unimportant, which has led to the market “shortcut”:
n = r + bei
Breakeven inflation
In principle, breakeven inflation is the rate of inflation that will equate the returns on an
inflation-linked bond and a “comparator” nominal bond of the same term. Calculating it by
simply subtracting a real yield from a nominal yield is a crude form of a properly
compounded calculation. While more accurate “Fisher breakevens” have been quoted in the
past, particularly in the UK, market convention has now moved decisively towards simple
spreads other than in high-yielding markets. Historically, breakeven inflation rates in the UK
were sometimes calculated in a more complicated way for old style linkers whose real yields
require an inflation assumption (3% is the market convention). In this case a semi-annual
Fisher breakeven could be used as the start of an iterative process in which the inflation
assumption was then adjusted towards a more accurate final ‘true’ breakeven inflation rate.
The equation to calculate Fisher breakeven for a market with an annual yield convention is:
(1 + bei ) = (1 + r )
1 n
(+ )
While, for a semi-annual market, the calculation is:
2
⎛1 n ⎞
⎜ + ⎟
( + )= ⎝
2⎠
1 bei
2
⎛1 r ⎞
⎜ + ⎟
⎝ 2⎠
The approximation of using a simple spread is not extreme if yields are relatively low. For
instance, with a real yield at 2% and a nominal yield at 4%, the distortion for an annual
bond is only 4bp. For semi-annual conventions the approximation is notably closer, in this
example only 1bp. These distortions are relatively small compared to other difficulties
involved that have to be accepted – invariably there is a term mismatch between linker and
comparator, there is reinvestment risk with nominal coupons larger than those in linkers,
and there is the fact that, because of the indexation lag, the real yield is not pure. In many
markets there is no unanimity over which nominal bond to use as a comparator. Breakeven
spreads are sometimes quoted versus an interpolated nominal curve, particularly in France.
A truer measure of breakeven inflation would be achieved if we were lucky enough to have
zero-coupon linkers with no lag and a zero-coupon nominal of identical term, although the
indexation imperfection would remain.
While conceptually the idea of an inflation risk premium, ie, an amount of yield that investors
will forgo in order to obtain inflation protection is inherently appealing, there is no reason that
this factor should be at all stable, as it is likely to be driven around by relative supply and
demand factors. Practically it may be feasible to observe when there are sharp changes in this
risk premium but not to put any precise value on it. The degree of liquidity discount for
inflation-linked bonds relative to nominal government issues can also vary significantly.
Logically the inflation risk premium should tend to increase as the tenor extends, along with
the degree of inflation uncertainty. In practice, demand from investors to hedge their inflation
exposures is unlikely to smoothly increase at longer maturities. Given that most investors have
a nominal benchmark, it is quite possible for inflation risk premia to be negative even before
15 March 2010 220
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Conceptually, the inflation rate implied by a zero coupon inflation swap should be the
cleanest measure of breakeven inflation available. Other than any indexation lag the
inflation swap rate is a pure measure, with the final value of an inflation swap only
determined by the difference between realised inflation and the traded level. This is only one
reason why inflation swaps may be a better measure for observing changes in inflation
expectations than bond breakevens. Full-year inflation swap rates do not have to be
adjusted to account for seasonality as bond breakevens do and there are no convexity
distortions in spot starting zero coupon swaps, unlike at the long end of the bond curve
where there can be contradictory and potentially unstable convexity issues for real and
nominal curves. A potential distortion that is more commonly an issue shorter on the curve
is that in many countries inflation-linked bonds have deflation floors that can distort the
meaning of real yield and breakeven valuations when implied inflation is very low or
negative, whereas the default zero-coupon inflation swap quote contains no optionality.
Despite the measurement advantage of inflation swap rates compared to bond breakevens,
it is not clear that it is more feasible to disentangle risk premium elements from inflationary
expectations in swaps than in bonds. Arguably swap rates do not include the same element
of liquidity discount evident in most inflation-linked bond markets. However, it is not a safe
assumption that the difference between bond and swap valuations accurately measures the
liquidity discount of cash linkers, even if changes in relative bond market liquidity may show
up on this measure. Supply and demand factors can distort inflation swap rates away from
expectations just as easily as in bonds, indeed in markets where swaps are less liquid than
bonds, the distortions can be more extreme and less stable. During late 2008, although
liquidity in inflation swaps was particularly limited in most markets, inflation swaps were
less distorted by deleveraging pressures than bond breakevens.
Practically separating out these factors is virtually impossible even when in extreme periods
when we can be confident of the direction of moves in individual variables. By way of
example we consider the market dislocation from September 2008, when liquidity premia in
all asset class increased sharply and expected inflation, as measured by forecasts and
Figure 194: Relative asset swap valuations may highlight Figure 195: US breakevens are poorly correlated with inflation
moves in bond liquidity premia, eg, UK spiked on nominal QE expectation surveys
Source: Barclays Capital Source: University of Michigan, Federal Reserve Bank of Philadelphia, Barclays Capital
surveys, also fell. Inflation risk premium in short-dated bonds likely turned negative
temporarily due to investors who had nominal benchmarks being forced to cut off
benchmark breakeven positions. During the period from mid September to the end of
November, off-the-run TIPS breakevens in the 5y sector cheapened 3.8%, to -2.3%,
compared to a 2.1% fall for the on-the-run issue, whose floor value suddenly became much
more valuable, and a 2.2% fall in the 5y CPI swap. Over the period from September to
December, the University of Michigan consumer survey of 5-10y inflation expectations fell
only 0.4% on a median basis, with 1y expectations falling 2.6%. While market expectations
of inflation may have adjusted more dramatically than those of consumers, the Livingston
Survey of professional forecasters suggested not, with no change in 10y inflation
expectations between June and December 2009 and a 1.5% move in 1y forecasts. This
implies that the inflation expectations component was the least important factor of the 5y
breakeven collapse, liquidity could have been worth up to the 1.3% implied by the relative
asset swap move of the off the run 5y, but that the most important driver was probably the
temporary, position-driven move to a negative inflation risk premium.
Figure 196: 3 month yield betas in various countries Figure 197: Effective IL55 duration with different beta methods
0.6 30
25
0.4 20 Real duration
3m returns beta
0.2 15
3m yield beta
10
0.0 12m weekly yield beta
5 12m weekly returns beta
-0.2 0
May-06 May-07 May-08 May-09 Jan-06 Jan-07 Jan-08 Jan-09
Beta estimation provides a useful starting point for relative analysis of inflation-linked versus
nominal duration. The estimate of beta is sensitive to the methodology used, but also to the
time period assessed. Beta should never be considered a stable relationship, but the type of
beta that is most appropriate depends on its use. For a trader looking for a short-term
hedge for linker exposure with nominal futures or bonds, a short-term yield beta estimation,
eg, based on daily changes over one to three months, may make sense, although in periods
of extreme carry, a returns beta may be more advisable. For active money managers, a
three- or six-month yield daily or weekly change beta may be the most representative, and
this is also the time horizon over which yield level betas are most commonly used, as stable
yield trend regimes typically last three months to a year. For longer-term total return
investors, a two-year or longer monthly total return volatility beta is a logical starting point
for asset allocation. For those with long-term real return aims or inflation-linked liabilities,
real, not nominal, duration should arguably be the more appropriate measure for assessing
risk, anyway, with a returns beta to compare the two.
Betas will vary due to many factors, but one of the most significant is the type of investor in
the asset class. Betas are usually lower when real money, real yield investors are predominant.
Historically these have been most prominent at longer maturities, often leading to lower betas
at the long end of the curve. At the short end, inflation uncertainty becomes an increasingly
significant factor relative to the decline in nominal price volatility; indeed, for very short-dated
bonds, betas will often be significantly above 1 as a result. Until the final principal value is
fixed, the nominal volatility of a sub-1y bond is likely to be significantly higher than 1.
This formula shows that provided the covariance between the real yield and breakeven
inflation is not sharply negative, real yields will be less volatile than nominal yields. In other
words, the yield sensitivity, or ‘beta’, of an inflation-linked bond to a change in the equivalent
nominal yield will usually be less than one. However, while the covariance between real yields
and breakevens had been small over most of the past decade in the US and Europe, indeed
close to zero in the five years until mid 2008, at this stress point the covariance moved sharply
negative and real yield volatility spiked notably higher than nominal vol.
The volatility of linkers spiked above that of nominal bonds in the 2008 turmoil, and with
correlations disappearing, realised breakeven vol rose to similar levels. In US TIPS, after the
liquidation phase in Q1 09, there was an illiquid period in which breakeven volatility fell
sharply, with real yield volatility similar to that of nominals. During this time, betas briefly
rose back to previous levels, albeit with a relatively low statistical significance since the
market was still poorly defined. As liquidity improved, the lower dependence of TIPS yields
to nominals has become clearer, with 10y real yield volatility falling much more rapidly than
in nominals. We expect this tendency for lower real yield than nominal volatility to continue.
The market is not back to the pre-2002 era when real yields were extremely stable and
breakevens were almost as volatile as nominals, but we expect that breakevens are likely to
be more volatile than between 2003 and mid-2008. In our view, this reflects a combination
of the new market dynamic and increased inflation uncertainty.
Figure 198: TIPS yield betas appear to have fallen structurally Figure 199: Realised volatilities have fallen from peak but
still high
0.4 0.08
0.06
0.2
0.04
0.0
10y TIPS 3m yield change beta 0.02
The beta and volatility trends for linkers in the euro area are in many ways similar to those
in US TIPS. Even more so than in the US, real yields displayed a similar volatility to nominals
for the five-year period until mid-2008. Breakevens were in a tight range for most of this
time, with yield betas averaging close to 0.9. The dislocations in the three months from
September 2008 were less extreme than in TIPS, largely due to the lower share of the euro
linkers held by breakeven investors; however, volatility trends during this time were very
similar, with breakeven vol almost equal to that of nominals and real yield vol higher still.
3mth yield betas briefly spiked higher during this period, but this corresponded with the
least statistical significance of the relationship.
As the euro linker market once again became fully functional, it settled into a beta regime
that while notably less stable and well defined than previously, was still relatively high
compared with the US – indeed, akin to that in TIPS in the middle part of the decade. This is
despite euro bond breakeven trading and investment remaining relatively limited as in the
US. The major market dynamic behind this is likely to be that among the largest buyers of
euro linkers in 2009 were asset swap investors. As relative asset swap levels returned to
normal ranges, bond and swap breakevens started to trade together again, meaning pure
inflation activity once again often being the marginal driver of €i prices. This is addition to
the structural factors that make euro breakevens more stable – ie, having a central bank
with an explicit inflation target, and with euro governments notably more sensitive than the
US Treasury to breakeven levels with regards to how heavily they issue euro linkers.
In the UK, there does not appear to have been a fundamental change after the late 2008
dislocations, which caused far less breakeven position cutting than elsewhere as less of the
market was held by breakeven investors to start with. 10y real yield vol has settled back at a
similar level to nominals, with breakeven vol an order of magnitude lower. The IL17 3
month yield change beta stayed at 0.8-1.0 even through 2008, and the statistical
significance of this beta returned above 80% from mid-2009. Betas at longer maturities
have tended to be lower than at 10y, but the difference became more extreme in 2009 than
in previous years.
Figure 200: €i yield betas have been less volatile than TIPS Figure 201: Euro real yield vol remains relatively close to
nominals
1.2 OAT€i20 3m real yield beta 0.12 10y OAT€i 3m real yield realised daily bp vol
R-squared of beta relationship 10y OAT 3m nominal realised daily bp vol
1.0 0.10 10y OAT€i 3m breakeven realised daily bp vol
0.8 0.08
0.6 0.06
0.4 0.04
0.2 0.02
0.0 0.00
Apr-04 Apr-05 Apr-06 Apr-07 Apr-08 Apr-09 Jan-02 Jan-04 Jan-06 Jan-08
Real return assets are those whose real returns are not typically eroded by inflation, which in
this chapter we consider as consumer inflation. We examine a range of assets, including
commodities, equities, dividend swaps, timberland, commercial property and volatility to
determine their inflation hedging properties. However, as many investors seek to match
future real liabilities that are based on real interest rates rather than just inflation, we also
seek to determine whether these assets hold a strong correlation to the level of real interest
rates. Here, we define the real interest rate on an ex post basis – for instance in the case of
the US, the 1-year realised Fed Funds rate minus the annualised rate of US headline CPI.
History has shown that during periods of high inflation it is essential to focus allocations in
the narrow band of assets which offer effective inflation protection, as simply relying on
portfolio diversification is insufficient to ensure a positive real return. Each real return asset
offers a different real risk/return profile across the business cycle. Commodities and equities
still offer the least attractive risk/reward profile, with their volatility increasing significantly
as a result of the deleveraging moves that occurred in H2 08. Meanwhile, UK linkers, global
linkers and real estate provide the most attractive returns relative to risk, despite also being
affected by the deleveraging. Indeed, the volatility of inflation-linked bonds increased the
least of all the asset classes surveyed in this chapter since the previous Global Inflation-
Linked Products: A User’s Guide was published in February 2008. However, the real estate
and timber real total return data used here are based on appraisal, rather than transaction
data, which are likely to bias risk lower.
Figure 202: Annual real return and risk of real return assets since 1987
30%
Real Return
Real Risk
25%
20%
15%
10%
5%
0%
Commodities Timberland UK linkers* Global Real estate Equities Treasuries
linkers*
Note: *Hedged to USD, UK linkers shown due to long sample; Global linkers from 1997. Source: GSCI, NCREIF, Barclays Capital
The extent to which the assets in this study are capable of hedging real rates or inflation depends
crucially on the holding period as many of these assets must be held for much longer than five
years to generate consistently positive inflation-adjusted returns. Equities failed to provide
sustained short or long-term inflation protection, although resource-based equity sectors may
offer protection from commodity-related inflation spikes. Dividends offer an attractive alternative
to pure equity exposure, as profit growth tracks inflation, and show a solid short-term positive
correlation with inflation. However, there are significant divergences in the payout ratios across
equity sectors and dividends prove an imperfect tool for accessing earnings streams and are
less efficient as an inflation hedge than they first appear. Commodities, timberland and short-
dated inflation-linked bonds hold the most positive correlations with inflation over periods of
three years and less. However, the volatility of commodity returns means they do not provide a
reliable inflation hedge for investors with either short or long real liabilities. Commercial property
and timberland display more traditional cyclical movements across the business cycle,
providing the strongest performances when growth is strongest. However, despite relatively
low volatility, commercial property provides little exposure to inflation over shorter periods.
Timberland offers broadly robust inflation and real rate protection, although this is over a
relatively short data sample. Overall, inflation-linked bonds remain the most attractive because
they can provide a complete inflation hedge if held to maturity or exposure to expectations of
future real rates if held over a shorter period.
Figure 203: Correlations between asset total returns and headline US CPI (UK RPI for UK
bonds) since 1987, under different holding periods
Monthly Quarterly Yearly 3 Year
In some cases, the returns on each asset class can also be split into a price return and an
income return. Depending on the asset, either component may be exposed to inflation, but
not necessarily both. For example, the coupon and principal of inflation-linked bonds are
directly exposed to inflation, but the traded price of the linker fluctuates based on
expectations of real interest rate levels. In this study we examine each asset class with
reference to the extent to which each element is exposed to inflation.
Equities
Equities have been considered a real asset because the investor receives dividend payments
plus any capital appreciation of the stock. However, equities themselves do not provide a
good long-run hedge against inflation, with the rolling 20y correlation between the total
return on US or UK equities and inflation negative for most of the past century. Short-term
equity total returns are negatively correlated with inflation, and the volatility of equity
capital returns makes equities highly risky for investors with short duration real liabilities.
However, historically, the strongest real returns on US equities and nominal bonds are
during periods that begin with very high inflation, as the rapid decline in inflation following
the initial spike leads to a sharp pick-up in the value of these assets.
Figure 205: Long-term real total returns of US and UK equities Figure 206: Equities are de-rated by inflation in the short term
45% US equity 10y annualised real total returns 16% S&P 500 Earnings yield
UK equity 10y annualised real total returns US CPI y/y (RHS)
14%
35% US equity 20y annualised real total returns
UK equity 20y annualised real total returns 12%
10%
25%
8%
15% 6%
4%
5% 2%
0%
-5%
-2%
-15% -4%
1909 1919 1929 1939 1949 1959 1969 1979 1989 1999 200 1955 1961 1967 1973 1979 1985 1991 1997 2003 2009
Although equities underperform gilts substantially in periods of high inflation, the falls in
equity values during these periods tend to be temporary, while falls in bond prices persisted
until inflation subsided. Bonds underperform equities during broader inflationary periods
because a rise in profits can offset some downside to equities. Hence, investors can protect
themselves in periods of sustained inflation by selling nominal bonds versus equities.
Not all equity sectors underperform in periods of high inflation. Equity returns in the US and
UK were narrowly distributed in the 1970s and real returns were negative across most sectors
and positive equity total returns were concentrated only in sectors which directly benefitted
from inflation, such as the oil and gas and industrials sectors. The basic materials sector
performed poorly, as oil prices outperformed industrial metals and other commodity prices.
While utilities provide the highest fraction of return from dividends among all equity sectors,
they also tend to be correlated with bond yields, leading to notable underperformance,
particularly during periods of high inflation. Therefore, a means to protect against inflation
appears to be to buy equities in the oil and gas and industrial goods sectors versus equities
in the consumer goods services, healthcare, utilities and technology sectors.
Figure 207: Equities do not provide an effective real rate hedge Figure 208: Total return for US equity sectors (1969-79)
100% S&P 500 annualised total returns 14% Oil & Gas
12% Industrials
80% US real interest rate (1y realised US Fed Funds
rate minus y/y US CPI) 10% Market
60% Utilities
8%
40% 6% Telecoms
20% 4% Consumer goods
Basic materials
0% 2%
Financials
0%
-20% Technology
-2%
-40% Healthcare
-4%
Consumer services
-60% -6%
1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 -10% -5% 0% 5% 10%
However, while equity returns do not typically offer protection in periods of high inflation,
this is usually caused by de-rating rather than weak earnings. Therefore, while equity capital
returns are volatile, the dividend income component may be used as an alternative
investment, leading us to examine dividend swaps as an inflation hedge.
Dividend swaps
Until recent years it was only possible to gain exposure to corporate profits via a
traditional equity investment. However, it is now possible to access corporate profits
almost directly by investing in dividend swaps. Dividend swaps are over-the-counter
derivatives used to trade dividends on a forward basis. They enable investors to receive
exposure purely to dividends, rather than dividend plus book value plus market
capitalisation of the long-run flow of dividends. Since a key component of corporate
profits is output pricing, it is logical that profit growth should track inflation. Over the past
century, dividends have a modest positive correlation with inflation, rising as the holding
period is extended to 5-10y and dividend growth generally became elevated when
inflation spiked above trend levels. While equities failed to provide positive real returns in
the stagflationary 1970s, dividends provided strong inflation protection, with the
correlation between S&P 500 and US inflation dividend growth almost 60% over the
decade. Dividend distributions have mostly proven resilient to modest market downturns
as corporates have generally been reluctant to cut dividends except as a last resort. The
stable long-run performance of dividends also offered an effective real rate hedge.
Figure 209: Dividends modestly correlated to inflation Figure 210: Dividends mostly offer protection against real rates
10% 4% 10%
5% 2% 5%
0% 0% 0%
Source: Robert Shiller, Barclays Capital Source: Robert Shiller, Barclays Capital
Dividends from commodity-related sectors, particularly energy and materials, have provided
the best overall performance and remained mostly stable. However, as Tim Bond highlights
in Nine steps to profit from inflation, 26 June 2008, the historical performance of dividends is
deceptively mixed, with significant divergences in the payout ratios across equity sectors.
Dividend payout ratios in the US and UK fell over the 1970s, particularly in the oil & gas and
industrials sectors. In other words, sectors whose earnings would have provided the most
effective inflation hedge during the 1970s reduced their payout ratios the most. Thus,
dividends were an imperfect tool for accessing earnings streams.
Figure 211: Change in dividend payout ratios, UK market, end 1969-end 1980
Market Oil & Basic Industrials Consumer Health Consumer Financials
Gas materials services care goods
There are also risks to dividend swaps that are not reflected in earnings streams, namely
liquidity and the credit risk of the counterparty. Therefore, while dividend exposure offers an
attractive alternative to pure equity exposure from the point of view of hedging inflation and
real rates in a general sense, they may provide imperfect access to earnings streams,
implying that their specific inclusion in a portfolio as a hedge against inflation or real rates
should remain modest.
Commodities
During inflationary periods it is useful to focus on the resources that are causing inflation.
While commodities do not provide direct income return, food and energy are two important
components in most headline CPI baskets; hence, the price of energy- and food-based
commodities logically hold some correlation with inflation. In US headline CPI, for instance,
food and energy collectively represent 23% of the CPI basket. Meanwhile, food- and energy-
based commodities make up 82% of the S&P GSCI Index, 70% of which is due to energy.
While 16% of the US headline CPI basket is linked to energy, energy has accounted for the
majaority of total inflation volatility over the past ten years.
Figure 212: Long-term commodities real total returns Figure 213: Real geometric annual returns by asset class,1970s
30% GSCI 5y annualised real total returns Real returns, geometric annual, 1970-80 US UK
GSCI 10y annualised real total returns
25%
GSCI 15y annualised real total returns Oil 11.5% 16.9%
20%
Industrial metals (fibre INDEX) -4.0% 1.4%
15%
Foodstuffs (CRB) -4.7% 0.8%
10%
Source: GSCI, Barclays Capital Source: Haver Analytics, Ecowin, Barclays Capital
Examining optimal holding periods reveals that investors have needed to hold commodities
for at least 15 years to ensure they achieve positive real returns. However, the main benefit
of investing in commodities lies not in the long-run returns, but in the ability to protect
investors from unexpected spikes in inflation. Commodities also have the advantage of
being negatively correlated with most other assets in this study and thus act a strong
diversifier within a real asset portfolio. As highlighted by their performance in the 1970s,
commodities provided significantly positive real returns on the decade and dramatically
outperformed other asset classes in this study. However, positive real returns were focussed
on a narrow band of commodities, namely the energy sector and specifically oil, but not
industrial metals or foodstuffs. The main disadvantage of commodities is their volatility, as
witnessed by the 15-year holding period that has historically ensured a positive real return.
Figure 214: Correlation of commodity real total returns and RPI Figure 215: Commodities are not an effective real rate hedge
Source: GSCI, Datastream, Barclays Capital Source: GSCI, Datastream, Barclays Capital
Therefore, while commodities benefit from unexpected spikes in inflation, their high short-
term volatility makes them an impractical inflation hedge for investors with short duration
real liabilities, unless held on a purely tactical basis. While, from a diversification perspective,
commodities may have a role in inflation hedging portfolios over a long-term horizon, their
extreme volatility suggests that even this should remain modest. There is virtually no
correlation between the level of commodity price and real interest rates; hence, they do not
offer protection from a real rate hedging perspective.
Timberland
Another natural resource that has been considered correlated with inflation is timberland.
Although timber can be put to many uses, the price of timber is the single most important
return driver and we examine the inflation and real rate hedging characteristics of the asset
class on this basis. The advantage of timberland is that it shows a relatively low dispersion
across the business cycle. US timberland has provided sustained positive real returns over a
relatively short holding period of five years and broadly positive returns over the past twenty
years. The real risk/return on timberland appears far more attractive than on other
commodities. However, this analysis uses the NCREIF US timberland data, which are based on
appraisal rather than transaction values and hence may understate the risks of investment.
Poor liquidity and maintenance costs are not captured in these data. There are also notable
divergences in timber prices between regions. Furthermore, with such a short sample – from
1987 in the case of the NCREIF US Timberland Index – we do not have sufficient data to
conclusively determine the long-term inflation hedging capabilities of the asset class.
Figure 216: US timberland hedges inflation over 5y period Figure 217: US timberland provides robust real rate hedge
An alternative source of timberland returns is timber equities, such as the S&P Global Timber &
Forestry Index and its constituents, but these are subject to the same problems as a typical equity
investment and hence do not offer a significant advantage over a timberland investment.
Commercial property
Annual returns for real estate, particularly commercial property, illustrate how closely its
capital growth and therefore total returns are related to the general business cycle. Rising
commercial property prices tend to benefit the corporate sector and affect economic
activity through business investment. Meanwhile, residential housing is usually the largest
component of household wealth; thus, house prices tend to lead consumer spending. In
addition, bank’s balance sheets are exposed to housing through loans and mortgages. As
has been particularly notable since the start of the credit crisis, real estate price movements
have a major effect on economic conditions and hence inflation. For inflation, this is especially
so if there is a direct house price component in the inflation basket, as is the case for UK RPI.
Figure 218: Real total returns on US commercial property Figure 219: UK real property returns annual % change
2% -20%
0%
-30%
-2%
-4% -40%
1983 1986 1989 1992 1995 1998 2001 2004 2007 1971 1975 1979 1983 1987 1991 1995 1999 2003 2007
Note: US Commercial Property is an aggregate of the NCREIF Industrial, Office, Source: IPD, Barclays Capital
Retail and Hotel Property Indices. Source: NCREIF, Barclays Capital
A large proportion of the return on commercial property comes from rent, which is mostly
very stable over time. Indeed, the appeal of commercial property as a real return asset is in
its stable real performance, with only a few periods of outright negative real returns.
Investors could have taken exposure to US commercial property for the past thirty years and
gained positive real returns over any 9-year period. Income returns on commercial property
have generally been relatively stable and far less volatile than equity income returns. Indeed,
US commercial property appears to offer a more attractive inflation-adjusted risk level than
other real asset classes. However, as the NCREIF US commercial property returns data are
also based on appraisal rather than transaction values, in reality, commercial property
returns are likely to be more volatile than these data would suggest. Investment in
commercial property is also subject to other constraints that are not captured in these data,
such as very poor liquidity, tenant default risk and maintenance costs. When these effects
are adjusted for, they are likely to bring the Sharpe ratio on commercial property closer in
line with riskier assets in this study. Despite relatively low volatility, US commercial property
provide little exposure to inflation over shorter periods and returns are only capable of
occasionally outpacing US real interest rates despite a modest long-run correlation.
8% US real interest rate (1y realised US Fed Funds rate minus y/y US CPI, RHS)
6%
4%
2%
0%
-2%
-4%
-6%
-8%
-10%
1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008
Note: US Commercial Property is an aggregate of the NCREIF Industrial, Office, Retail and Hotel Property Indices.
Source: NCREIF, Barclays Capital
An alternative way to source property market returns is via property derivatives, whose
returns are linked to the performance of an index of property prices such as the UK IPD All-
Property Total Return Index or the US NCREIF index. However, volumes and liquidity in
property derivatives are low compared with inflation derivatives. There are also considerable
lags between the valuation and publication dates for the property indices, while the recent
dramatic falls in residential and commercial property prices may have dampened investor
demand for the asset class despite its scope to develop.
Infrastructure assets such as roads, airports, water management and power networks have
also been noted as real return assets, given that their values exhibit positive correlations with
inflation, economic and demographic growth. Infrastructure assets are excluded from our
analysis given their short data sample, with data on the UBS US Infrastructure index, for
instance, only available since 1995. However, it should be noted that utilities equities provided
the highest fraction of return from dividends among all US equity sectors since 1989.
Volatility
An asset class that did not exist during the inflationary 1970s is volatility. Investors can now
take positions in actual and implied asset volatility as a separate asset class, such as the VIX
and MOVE indices. Equity and interest rate volatility typically rise with inflation, and while
inflation is certainly not the sole cause of volatility, it is often a principle contributing factor.
Equity earnings yields and the volatility of profits are very well correlated and, in turn, the
volatility of profits and economic growth are also tightly correlated with inflation. Thus, the
economic and profit climate becomes far more volatile and uncertain during inflationary
periods. Indeed, the broad rise in earnings yields that accompanies periods of high inflation
is due, in part, to demand for higher risk premia to compensate for the increased volatility of
profits. Meanwhile, sharp rises or falls in the rate of inflation tend, by their very nature, to be
unexpected and so also produce unexpected changes in interest rates and as a result
interest rate volatility also tends to be correlated with inflation. Therefore, a means to hedge
against inflation is to be long nominal bond and equity volatility.
Figure 221: US 20y equity volatility versus inflation, 1962-80 Figure 222: US 20y rate volatility versus inflation, 1962-80
20% S&P 500 1y standard deviation 14% 12% US 20y nominal swap, 1y standard 14%
18% of quarterly returns deviations of quarterly return
US CPI, % y/y (RHS) 12% 10% 12%
16% US CPI, % y/y (RHS)
14% 10% 10%
8%
12% 8% 8%
10% 6%
8% 6% 6%
4%
6% 4% 4%
4%
2% 2% 2%
2%
0% 0% 0% 0%
19611963 19651967 1969 19711973 1975 19771979 1981 1962 1965 1968 1971 1974 1977 1980
In order to strip out real rate exposure, investors must look at breakeven inflation positions
or TIPS returns relative to nominals. We note that even if investors buy TIPS on an outright
basis, if the decision is whether to purchase TIPS or another fixed income product, then this
is essentially a breakeven position. Therefore, an outright position in TIPS still offers a good
inflation hedge relative to other fixed income products. The diversification hypotheses
outlined above and potential inflation hedge are good reasons to add TIPS to a diversified
portfolio. However, before getting to an asset allocation decision, we first show how TIPS
total returns have held up compared with other assets.
Figure 223: Y/y return correlations show TIPS have a low correlation with other assets, suggesting diversification benefits
Short Tsy
BEI Index
AUDCAD
NZDNOK
Inflation
1-3y BEI
CPINSA
Global
GSCI
Gold
TIPS
CRB
S&P
Oil
Asset
TIPS 100%
Global Inflation 74% 100%
1-3yBEI 72% 83% 100%
BEI Index 31% 46% 96% 100%
Gold 13% 36% 56% 27% 100%
Oil 28% 38% 81% 68% 25% 100%
CRB 57% 64% 83% 64% 47% 80% 100%
GSCI 50% 43% 72% 59% 20% 86% 90% 100%
S&P -9% 34% 73% 58% 6% 37% 32% 27% 100%
Short Tsy -54% -14% 43% 61% 16% 36% 13% 11% 60% 100%
AUDCADNZDNOK 53% 86% 81% 57% 52% 41% 62% 38% 39% 13% 100%
CPINSA 31% 26% 60% 39% 25% 62% 74% 75% 25% 8% 20% 100%
Source: Barclays Capital, March 1997 to December 2009, Using Barclays TIPS Index
Figure 224: Total returns of all assets in a diversified portfolio since inception of TIPS
240%
210%
180%
150%
120%
90%
60%
30%
0%
-30%
-60%
Mar 97 Sep 98 Mar 00 Sep 01 Mar 03 Sep 04 Mar 06 Sep 07 Mar 09
Figure 224 shows that since their inception (March 1997), TIPS have held up fairly well in
the volatile trading environment. In fact they have returned the highest among a portfolio of
Treasuries, High Yield and Investment Grade corporate debt, MBS, the S&P 500 and the
Goldman Sachs commodities indices on a total return basis over this period. Only the
unhedged Global Inflation-linked Bond Index had a higher return among the assets shown
in Figure 225. TIPS initially had relatively low volatility, then benefited as energy prices
rallied significantly. TIPS underperformed nominals along with other riskier assets during
the financial crisis as there was a significant flight to liquidity to nominal Treasuries. We
note that the volatility of TIPS returns during this time was lower than other assets which
are seen as an inflation hedge, such as commodities.
Figure 225: TIPS have delivered high returns with less volatility
S&P
High 500 US Global
TIPS Tsy Yield Corp MBS TR GSCI Agg Inf.
Annualized 5y 4.6% 4.8% 6.5% 4.6% 5.8% 0.4% -3.0% 5.0% 4.9%
Returns 10y 7.7% 6.2% 6.7% 6.6% 6.5% -0.9% 5.1% 6.3% 7.8%
Since inception 6.7% 6.1% 6.3% 6.4% 6.4% 4.5% 2.6% 6.2% 7.7%
1y 7.9% 5.5% 13.0% 6.8% 2.5% 22.3% 24.5% 3.3% 10.5%
3y 8.7% 5.7% 17.2% 9.0% 3.3% 19.9% 31.1% 4.2% 11.8%
Annualized
5y 7.2% 4.8% 13.5% 7.4% 3.0% 16.0% 28.2% 3.7% 9.8%
Vol of monthly
returns 10y 6.7% 5.1% 11.5% 6.3% 2.9% 16.1% 25.6% 3.8% 8.6%
Since inception 6.0% 4.8% 10.4% 5.9% 2.8% 16.6% 24.6% 3.7% 8.0%
Return/Risk 10y 1.15 1.22 0.59 1.04 2.23 -0.06 0.20 1.65 0.91
Since inception 1.10 1.26 0.60 1.07 2.29 0.27 0.11 1.67 0.96
Note: March 1997 to December 2009, Using Barclays TIPS Index. Source: Bloomberg, Barclays Capital
We expect TIPS to continue to become more liquid as the Treasury is taking steps to
improve the programme. Most recently the Treasury solicited market suggestions and has
not only acted to increase the size of TIPS auctions and reintroduce 30y TIPS issuance, but
has also indicated consideration for increasing the frequency of TIPS auctions. We see these
actions as a positive for TIPS in a market with improving economic fundamentals. Lastly, a
fundamental feature of TIPS is that they provide increased certainty of real returns to
maturity, protecting purchasing power of investors and are therefore unlike any other
securities. Thus, we view them as a legitimate asset class that should be included in a
diversified portfolio. In our view, most investors agree with this assessment but wonder how
much they should allocate to TIPS. Figure 226and Figure 227 show that using historical
data, TIPS continue to improve the efficient frontier. According to this analysis, a diverse
portfolio should include more than a 10% weight in TIPS.
However, we know that historical returns are no indication of future returns, although
historical covariance can be a guide to future covariance. We use the Black-Litterman model
to determine, and to some degree confirm, optimal weight allocation to TIPS for a diverse
portfolio consisting of assets shown in Figure 2 and 3m T-bills. We incorporate our view of
1y ahead total returns in this model to arrive at optimal allocation weights.
Alternatively, these same market equilibrium returns can be retrieved using the following
method (this is the solution to above optimization problem).
Figure 226: TIPS push out the efficient frontier when using Figure 227: Optimal portfolio includes TIPS across risk
historical returns as expected returns appetites, using historical returns
Source: Barclays Capital, Bloomberg, Constraint: Weights < 0.25, Bills < 0.1 Note Analysis done in March 1997 to December 2009. Constraint: Weights <
0.25, Bills<0.1 Source: Barclays Capital, Bloomberg
∏ = λ ∑ω
To estimate the risk aversion coefficient, we used historical returns and expected 1y future
risk-free rate (50bp). The risk aversion coefficient we arrived at is roughly 9.1. This rather high
coefficient will not determine relative weights of each asset in an optimal portfolio; it will tune
the magnitude of returns. Figure 228summarizes our market portfolio. Excess equilibrium
returns for equities are quite high relative to others as this return is needed to justify higher
weight held for this asset in an optimum market portfolio. As a side note, the optimal weights
(maximum Sharpe ratio), derived using historical covariance matrix and noted excess
equilibrium returns below, are the same as relative market weights of all assets. We have
forced this condition by defining the relationship in the equation above. Hence under market
equilibrium assumptions, TIPS should be allocated 3% (market cap weight) if your asset
allocation choices are the same as in this table. In line with this, investors should allocate TIPS
according to the relative market caps of their investable asset universe.
An efficient frontier of these equilibrium returns and covariance matrix has one optimal
portfolio which assigns as much as 7% weight to TIPS. Letting weights vary freely, although
mathematically sound, is not very practical. Below we have shown the equilibrium market
return frontier with a constraint of each weight being below 25% of total allocation (Bills are
limited to a 10% allocation). TIPS push the efficient frontier of the constrained market
portfolio to the left as they offer higher returns compared with nominals and bills while
maintaining only slightly higher volatility. Under this constraint, TIPS are allocated as much
as 12% at the least risk optimal portfolio.
Figure 228: Market equilibrium portfolio weights Figure 229: Equilibrium Market Portfolio Efficient Frontier
the market portfolio. Barclays Capital holds the following six views about one year ahead
excess returns of the assets in consideration (non-rates views are taken from the respective
Barclays Capital strategy group in each asset class).
Using the Black-Litterman model (See paper for implementation details, Black, Fischer, and
Robert Litterman, “Global Portfolio Optimization,” Financial Analysts Journal,
September/October 1992: 28–43.), we included these views to form our expected excess
returns vector (Figure 228, last column) and derive view based efficient portfolios (Figure
230). Again, for a relative view (such as view 1, 2), derived absolute return levels are not a
concern because we have only identified views of TIPS versus Treasuries and Treasuries
versus MBS. We notice that our first view regarding TIPS is not far away from what is
implied by relative equilibrium returns. Figure 6 shows that TIPS outperform nominals by
0.9% in a market equilibrium portfolio, thus view 1 only adds 0.40% of return change to our
equilibrium view. In other words, we do not expect significant change in our efficient
frontier owing to view 1. The most significant change comes from view 3, for commodities.
The commodity view return expectations are well above what is priced in by equilibrium
returns. Figures 8 and 9 confirm that the allocation of TIPS at more than 10% lowers risks.
This exercise corroborates the view that TIPS should be included in a diversified portfolio.
We recommend that investors start with the market portfolio (relative market caps) to
determine specific weight allocated to TIPS and adjust allocation based on their views.
Figure 230: TIPS push out the efficient frontier left, reducing Figure 231: Optimal Portfolio Includes TIPS at lower risks,
risk, using our 2010 asset views (Black-Litterman Model) using our 2010 asset views (Black-Litterman Model)
4.0%
60%
3.5%
40%
3.0%
20%
2.5%
0%
2.0%
3.48% 3.50% 3.80% 4.10% 4.40% 4.70%
3.45% 3.65% 3.85% 4.05% 4.25%
Risk
RiskDiversified Portfolio + TIPS TIPS Treasury HY Corp
Diversified Portfolio SP GSCI MBS 3m T-Bill
Note: Constraint: Weights < 0.25, Bills < 0.10.Source: Bloomberg, Barclays Capital Note Analysis done in March 1997 to December 2009. Constraint: Weights <
0.25, Bills < 0.10.Source: Bloomberg, Barclays Capital
However, the correlation between the total return on TIPS breakevens and the Broad Trade
Weighted USD is statistically still very weak, with an r-squared of only 18% from 2000-
2009. This was notably higher from late 2004 to mid 2008, at 78%, as a greater number of
central banks began buying TIPS. However, the correlation fell sharply from Q3 08 as the fall
in TIPS breakevens, from deleveraging and the economic slowdown, and their subsequent
re-widening were much greater in magnitude than the movement in USD. A broad
directional relationship was still clearly intact, but the weak correlation between TIPS
breakevens and the broad Trade Weighted USD suggests that a portfolio of domestic
inflation-linked bonds offers very limited value as a currency hedge, especially in periods of
economic volatility.
Figure 232: US CPI not correlated with Trade Weighted USD Figure 233: TIPS breakevens not an effective currency hedge
The benefits of including global inflation-linked bonds within a portfolio are evidenced by the
large number of institutional and retail investors that have made explicit allocations to the
asset class within their portfolios. In our view, the best way to diversify a portfolio by including
linkers is to invest in global inflation-linked bonds hedged to the local currency. In isolation, a
currency-hedged global linker portfolio maintains the most attractive features of a domestic
inflation-linked bond portfolio, specifically; better diversification, enhanced returns and low
risk. However, global inflation-linked bonds add a further diversification benefit to a portfolio,
even if this already includes domestic inflation-linked bonds, leading us to see a global linker
portfolio as the best way to capture the strategic benefits of owning linkers.
When examining the properties of the Barclays Capital World Government Inflation-Linked
Bond Index versus Treasuries, US high yield and investment grade corporate debt, mortgage-
backed securities, the S&P 500 and the GSCI, we find, unsurprisingly, that the relationship
global linkers hold with these assets is very similar to the relationship of TIPS to these assets.
For instance, like TIPS, the performance of our global inflation-linked bond index holds a
high correlation to the total return on US investment grade corporate debt. The total return
on our global inflation-linked bond index also holds a low correlation to riskier assets such
as commodities. However, we also find that the total return on our global inflation-linked
bond index holds a stronger correlation than TIPS to the total return on the GSCI and S&P
500, but a weaker correlation to the performance of US fixed income assets.
The explanation for this is intuitive. A global portfolio of currency-hedged linkers maintains
almost the same correlation to US inflation as a pure TIPS portfolio and currency hedging a
global inflation-linked bond portfolio generally compensates for any differences in local
inflation rates. There is also a broad similarity between the expected paths of developed
country interest rates, hence the correlation our global inflation-linked bond index holds to
US fixed income assets. The slight divergence is explained by the fact that returns on TIPS
are purely sensitive to changes in expectations for the future path of US real rates while the
global inflation-linked bond index provides exposure to future real rate expectations over a
variety of developed countries.
Figure 234: Global linkers greatly enhance the efficient Figure 235: Global linkers in an optimized portfolio of US
frontier, using Historical mean and covariance assets, using historical mean and covariance
Source: Bloomberg, Thomson DataStream, Barclays Capital Source: Bloomberg, Thomson DataStream, Barclays Capital
The return volatility on our index of global inflation-linked bonds has been considerably
lower than on a pure exposure to TIPS, suggesting that, from a mean-variance perspective
at least, history indicates that a portfolio of global linkers is the fixed-income asset of choice.
The difference in volatility between the returns on a currency hedged index of global linkers
and on an index of domestic inflation-linked bonds is that, for a portfolio of global linkers,
the impact of macroeconomic shocks and financial market volatility should be more evenly
spread than for domestic linkers. For example, the total return on the Barclays Capital TIPS
1y+ index fell by 10.7% peak-to-trough through the deleveraging of financial assets from
October to December 2008, while the total return on the currency hedged Barclays Capital
World Government Inflation-Linked bond index fell by 7.6% over this period.
As a result, the inclusion of global inflation-linked bonds within our mixed portfolio of US
assets lowered risk and sharply increased returns, thereby pushing out the efficient frontier.
The example we present here is when exposure to the total return on the Barclays Capital
World Government Inflation-Linked Bond Index, hedged to USD, is added to our portfolio of
US Treasuries, MBS, investment grade corporate bonds, the S&P 500 and GSCI. Naturally,
we do not include separate exposure to TIPS in addition to our global inflation-linked bond
index as TIPS already make up the largest domestic allocation in our global inflation-linked
bond index; 39% at the end of 2009. As the Sharpe ratios on global inflation-linked bonds
are much higher than on the other asset classes in this study, we limited the contribution of
global linkers to a maximum of 25%. Global linkers were selected at a negligible weighting
when the risk on the portfolio was below 1% and occupied less than 10% of the portfolio
until risk was increased to just above 1.2%, at which point this rose rapidly to the maximum
allocation of 25%. The inclusion of global linkers was to the detriment of equities and US
fixed income assets, in particular investment grade corporate debt.
The volatility advantage which global linkers hold over the other asset classes in our study is
even greater when considering real, rather than nominal, risk/return. Therefore, the benefits
of exposure to inflation-linked bonds from a variety of developed markets holds positive
implications for a portfolio of assets, even for investors who do not have explicit inflation-
linked liabilities.
Modelling the dynamics of real yields and breakevens has long been both an academic and
Henry Skeoch
a market goal. However, as we have previously pointed out, including in earlier editions of
+44 (0) 20 7773 7917
this guide, the topic is something of a notorious analytical quagmire. In this section, we put
henry.skeoch@barcap.com
forward linear regression fair value models for the US, euro area and UK. There are
significant shortcomings to this approach, in particular the instability of regression
components over time which leaves this type of model inappropriate to form the basis of a
trading strategy. Nonetheless, we have not found that more complex time series methods,
such as ARIMA and GARCH-based modelling, provide sufficient extra statistical rigour to
compensate for their lower transparency. Even so, there is a significant danger of spurious
regression relationships where coincidentally similar trends create a strong statistical link.
This is a particular danger for real yields, where statistically the series in all three markets
appear to be non-stationary, with a trend towards falling real yields over the course of the
last decade. Hence, we aim to include variables that are both consistently statistically
significant through the observation period and economically logical.
We attempt to model 10y real yields and breakevens as a sector that is likely to be relatively
liquid but also more fundamentally driven than elsewhere on the curve. The front end is
liable to be particularly sensitive to short-term carry and seasonality factors. The long end is
more likely to be significantly distorted by supply/demand considerations. Many of the
variables within the models for different countries are similar and, in some cases, the factors
are almost unchanged from those that were shown in previous editions of this publication.
10y real yields are understandably highly correlated with policy rates in the US, euro area
and UK. For breakevens, it is typically the slope of front end of the cash curve that is a more
significant variable, with breakevens higher in environments when the front end is pricing in
significant rate hikes.
Figure 236: 10y TIPS real yield versus fair value model Figure 237: 10y TIPS real yield model coefficients and statistics
0.5
Jan 98 Jan 00 Jan 02 Jan 04 Jan 06 Jan 08 Jan 10
Source: Haver Analytics, Bloomberg, Barclays Capital Source: Haver Analytics, Bloomberg, Barclays Capital
The uncertainty over model estimation has obviously increased during the period of
elevated market and economic volatility in 2008 and 2009. The standard deviation of the
model over this time was 43bp, double that in the previous ten years and with the largest
deviation between actual and model yields both to the high side and low side occurring
during this period. However, despite the volatility of real yields and many of the inputs, the
model itself has proved surprisingly resilient. Testing out of sample after August 2007, when
the first strong indication of the end of the credit boom emerged with the dislocation of
cash and Libor rates, the real yield model is always less than 10bp away from that
estimating over the whole sample. The most dramatic overshoots versus the model are also
entirely understandable in hindsight. Real yields were exceptionally cheap versus the model
between October and Thanksgiving 2008 when the TIPS market dislocation due to
breakeven deleveraging was at its extreme. Real yields by contrast moved very rich versus
the model in March 2009 following initial the FOMC announcement of direct purchases.
Note that we estimate our 10y TIPS models from the start of 1998 rather than the first
TIPS issuance in February 1997. This is for two main reasons. The first of these is the
uncertainty as to the status of CPI measurement during 1997 following the publication of
the Boskin Commission report in December 1996. Coupled with a relative illiquid and
poorly defined market for the new asset class, we see 1997 as a poor starting point for
estimating TIPS fair value.
Over most of the sample period the most important factor driving the breakeven model is
global industrial confidence. We use a proprietary series that Barclays Capital economists
have developed based on combining major national business confidence surveys
weighted by purchasing power parity adjusted industrial production. Notably this series
tends to dominate currency and broad commodity variables. It also tends to lead
breakeven movements modestly; hence, it is in the model with a one month lag. However,
while this series has been consistently statistically significant its importance has varied
substantially over time. Similarly, while retail gasoline prices have always been a major
driver of breakeven valuations, the sensitivity to energy movements has varied
considerably over time. The final variable in the model is the slope of the Fed Funds
futures strip, whose importance has also been unstable and was not significant during
2008 and 2009. Nonetheless, we see this as an appropriate measure to capture the
degree of risk premium within the market and expect that it will quickly become an
important factor again as rate hike expectations rise, which we would expect to coincide
with a rise in inflation expectations.
Figure 238: 10y TIPS breakeven versus fair value model Figure 239: TIPS breakeven model coefficients and statistics
0.5
0.0
Jan 98 Jan 00 Jan 02 Jan 04 Jan 06 Jan 08 Jan 10
Note that we estimate our euro real yield model since the start of 2002, rather than using
OATi data to extend the time period back to 1998. Liquidity prior to the launch of the
OAT€i12 was limited and whilst the trend for French inflation over this time was relatively
similar to that of euro HICPx, it is hard to know if inflation expectations were consistent. The
ECB policy rate and energy were highly significant for French real yields during this period
though. M3 is a significantly less reliable variable through the initial introduction of the euro
and was again distorted into the start of 2001 with the introduction of the physical currency
alongside the entry of Greece into the monetary union.
Figure 240: 10y Euro real yields versus fair value model Figure 241: Euro real yield model coefficients and statistics
1.5
1.0
0.5
0.0
Jan 02 Jan 04 Jan 06 Jan 08 Jan 10
The level of the FTSE 100 share index is an important determinant of gilt linker valuations,
with real yields tending to fall as equity prices rise. While the direction of this relationship is
seemingly illogical, the long-term sensitivity of 10y nominal yields to share prices is low and
the level of the equity market is a significant factor in determining asset allocation decisions
from pension funds into gilt linkers. When share prices are high it accelerates the ongoing
trend since the late 1990s for schemes to switch out of equity and into fixed income, with
real yields being particularly sensitive to this flow. Unsurprisingly, in aggregate pension
funds tend to reduce their asset versus liability mismatch when they can afford to do so.
We model UK real yields since the start of 1998. This is in order to avoid misspecification
stemming from the structural break following the decision in May 1997 to give the Bank of
England monetary policy independence. The dynamics for linkers were also significantly
changed around this time with the implementation of the minimum funding requirement
for pension funds. In common with real yield models in other countries, the spike yields in
Q4 2008 is not picked up, with the model dealing poorly with the period prior to this, from
mid 2007, when the linker market started to become significantly dislocated and
dysfunctional. However, the dynamics of yields in 2009 are tracked closely even estimating
out of sample from the period prior to mid-2007.
Figure 242: 10y UK real yields versus fair value model Figure 243: UK real yield model coefficients and statistics
0.5
0.0
Jan 98 Jan 00 Jan 02 Jan 04 Jan 06 Jan 08 Jan 10
10y UK breakeven
10y UK breakevens, like real yields, are heavily influenced by monetary policy, with a strong
positive correlation with both the level of the Bank Rate and the slope of the Short Sterling
strip. The importance of these factors is not just due to real yields being naturally stickier
than nominals. Movements in front end yields also feed into RPI inflation via mortgage
interest costs. Unlike in either the US or euro area, UK breakevens are not highly responsive
to oil prices. This is due to the combination of a notably lower pass through of energy price
movements into RPI inflation than elsewhere and with most of the drivers of the 10y sector
coming from the supply/demand dynamic at longer maturities. Broader cyclical and
commodity price pressures remain significant though, hence our model includes variables
of global industrial confidence and the level of CRB raw industrials prices. On their own,
either of these two would show up as the most statistically significant factor within the
model. The correlation between these two cyclical variables masks the statistical
importance in the model of each of the variables and leaves the value of each relatively
unstable, but not including both notably reduces the effectiveness of the overall model.
Without these two variables, the level of sterling would also show up as an important
component of breakevens, but it is dominated by the two factors that we use.
Testing the model out of sample using an estimation to mid-2007 highlights that the
sensitivity to various components has been unstable over time, despite all components
remaining highly significant, with the out of sample model implying breakevens 30-50bp
lower than the whole period estimation. The importance of the interest rate variables is
notably lower over the whole sample period than the pre July 2007 period, while the relative
significance of the industrial confidence and industrial materials prices coefficients is
reversed. The former may be more concerning than the latter, given the difficulty of defining
the significance of the cyclical coefficients stemming from their statistical overlap.
Figure 244: 10y UK breakeven versus fair value model Figure 245: UK breakeven model and coefficients
1.5
1.0
0.5
Jan 98 Jan 00 Jan 02 Jan 04 Jan 06 Jan 08 Jan 10
To the extent that seasonal trends are predictable an efficient market ought to price these
in. However, it is not always straightforward for market participants to observe seasonality.
Even in countries where statistics agencies publish full seasonality estimates for market
participants to observe, there is uncertainty as to how this seasonality will develop over
time. Typically both bond and swap markets price in significantly less seasonality for future
years than statisticians would estimate given current data. However, there is an argument
that certainty over the seasonal vector declines over time so any forecasted seasonal should
contain a decay component. A counter-argument to this is that there is no reason to
dampen an unbiased estimation, while from a practical perspective there is evidence in
many countries that there has been a tendency for the magnitude of seasonality to increase
in recent years. We see it as appropriate to assume a static seasonal vector and adjust this
as more information becomes available.
Four methods of seasonal adjustment are common: ratio versus moving average (or sum);
regression using dummy variables; variants of US Census Bureau’s X-12-ARIMA; and the
Bank of Spain’s TRAMO/SEATS procedure. X-12 is the most widely used by statistics
agencies, having existed in some form since the late 1960s (although many agencies use
the less complex X-11 variant), while TRAMO/SEATS is widely used in Europe. The main
difference between TRAMO/SEATS and X-12 is that TRAMO/SEATS estimates and fits an
ARIMA model to the data in order to derive an estimate of seasonality, whereas the core
seasonal adjustment method underlying X-12 uses a series of non-parametric moving
average filters, with fitting an ARIMA model an optional extra (full details are available on
the US Bureau of the Census Website). This gives X-12-ARIMA the upper hand when
constructing a seasonal vector including individual components that do not exhibit highly
statistically significant seasonality, but still show discernible patterns, which the vector
needs to account for.
Generally we prefer to use seasonality estimates from statistics agencies if they are
published in sufficient detail, but where this does not exist we suggest using an X-12
estimation at as low a level of aggregation as feasible. Estimating on an aggregate series on
average tends to produce a slightly higher seasonality than considering sub-components,
whether or not the sub-component approach estimates all variables or only those with
significant seasonal components. Problems can arise in adjusting an inflation index
constructed from unchained series using a set of weights, then chained at the index level.
Successive unchaining and chaining of series is not only laborious, but can bias the end
result (due to the Cauchy-Schwarz inequality). This can, however, be circumvented by
estimating a seasonal vector for each component of the index, then aggregating these
individual vectors by the index weights to give an overall seasonal vector for the index.
United States
Consumer inflation, as measured by the Consumer Price Index, follows a regular seasonal
pattern. One way to visualise the typical seasonal pattern is shown in Figure 246, which
plots the average difference between the Not Seasonally Adjusted (NSA) and Seasonally
Adjusted (SA) monthly percentage change in the CPI. While seasonals vary over time there
has been a consistent tendency for prices to rise more quickly in the early part of each year,
with seasonality negative from July onwards. The NSA CPI tends to rise most notably
between February and May, but decline significantly in November and December.
Figure 246: CPI seasonals, 2001-08 (difference between m/m NSA and SA% chg)
0.2%
0.1%
0.0%
-0.1%
-0.2%
-0.3%
-0.4%
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Source: Bureau of Labor Statistics, Haver Analytics, Barclays Capital
The Bureau of Labor Statistics (BLS) estimates the seasonal factors by applying the X-12-
ARIMA seasonal adjustment model developed by the US Bureau of the Census, which
attempts to adjust for monthly distortions, at an individual component level. Where
appropriate an intervention analysis is used, which adjusts for sharp and permanent shifts in
the underlying trend. These have the potential to distort the results of the seasonal
adjustment, and are accounted for via regression-ARIMA models. Seasonal factors are re-
estimated and published at the start of each year, coinciding with the January data release
and causing a change in the historical seasonally-adjusted series. The adjustment is carried
out on subcomponents of the index and then aggregated, which tends to produce a slightly
more conservative estimate of seasonal factors than top-down estimation, even before the
dampening impact of the intervention analysis is considered.
Figure 247 shows a breakdown of the estimated seasonal contributions from the main CPI
components. They are calculated as the implied difference between the NSA and SA
percentage changes for these components, weighted by their relative importance in the
headline CPI. Seasonal fluctuations in the headline index are driven mainly by movements in
core and energy prices during the year; food prices, in contrast, are relatively stable
throughout the year. Energy prices tend to rise most significantly in April-June, as the
anticipation and onset of the summer driving season tends to put upward pressure on retail
gasoline prices in those months. In contrast, energy prices tend to fall off in the fourth
quarter. The seasonal pattern of home heating oil is generally the reverse of gasoline but
has a much smaller weight, so gasoline is the dominant factor. However, while the
seasonality of energy is larger than other components it is also relatively less stable.
0.40% Core
0.30% Energy
0.20% Food
0.10%
0.00%
-0.10%
-0.20%
-0.30%
Jan Feb M ar Apr May Jun Jul Aug Sep Oct Nov Dec
As can be seen in Figure 247, NSA core inflation tends to rise significantly in February and
March, fall off in May-July, rise in October, and fall sharply in November and December.
Figure 248 shows the most important factors behind this. The main contributors are shelter
and apparel costs. Seasonality in shelter is mainly driven by out-of-town lodging, which
tends to soften in September-December, as demand falls off when the summer travel
season ends. December marks the low point of the year for out-of-town lodging costs,
presumably because travellers tend to stay with family rather than in hotels during the
holiday season. Shelter costs begin to rise early in the year as travel patterns start to
normalise. Another important contributor to core seasonality is apparel. These prices fall off
in November-January as holiday discounting dominates, then rise notably in February and
March. They fall off again in June-July as summer merchandise is cleared ahead of the back-
to-school season, when they pick up again.
Figure 248: Contributions of core components to headline CPI seasonal pattern, average
2001-08 (difference between m/m NSA and SA % change of CPI components)
0.20% Apparel
0.15% Shelter
0.10%
0.05%
0.00%
-0.05%
-0.10%
-0.15%
-0.20%
Jan Feb M ar Apr May Jun Jul Aug Sep Oct Nov Dec
One of the most noticeable effects that inflation seasonality can have on the tradable
inflation market is the seasonal pattern of breakeven returns. While divergences certainly
occur, Figure 249 shows that breakeven returns have traditionally risen in the first half of
the year and fallen in the latter. We calculate these averages by looking at duration
weighted breakeven return averages by month, going back to March 1997. This pattern in
the US is due primarily to the CPI seasonality discussed above. It is notable that this
tendency towards seasonal performance is much stronger in TIPS than in other inflation
markets, despite seasonality itself not being significantly larger.
Figure 249: Average monthly breakeven returns are higher in the first half of the year
0.80% 0.50%
0.60% 0.40%
0.40% 0.30%
0.20% 0.20%
0.00% 0.10%
-0.20% 0.00%
-0.40% -0.10%
-0.60% -0.20%
-0.80% -0.30%
-1.00% -0.40%
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Breakeven Returns (Monthly, LHS) Average CPI Seasonals: 2001-2008
While the carry-based seasonality buying effect may retreat in future as efficiency increases
in the TIPS markets, we also find that the market misprices issue-specific seasonal
differences, depending on your seasonal assumptions. For example, all else being equal,
April maturity issues should trade cheap to those that mature in January because the April
issues picks up one extra December print, when seasonality is consistently and significantly
negative due to holiday discounting and other factors. Similarly, July issues should trade rich
to January and April TIPS because they are expected to benefit from the typically good
spring carry period. All else is not equal though, and investors need to account for curve,
floor value and other factors that can drive relative value to isolate seasonal effects. Below
we discuss a methodology by which it is possible to use a seasonal assumption to detect
value among TIPS issues that are less than one to two years apart.
Methodology
The general approach to calculate appropriate issue-specific breakeven levels based on the
current market-implied forward breakeven trend and seasonal assumptions uses the trend
and adding cumulative seasonality to project maturity NSA CPIs for specific issues. These
projected NSA CPIs can then be used to calculate appropriate breakeven levels (see ‘Market-
implied breakeven curve,’ Global Inflation-Linked Monthly, 15 October 2009, for related notes).
All breakevens should be adjusted for respective floor values. A quick way to estimate floor
value is by taking the difference between Traded ASW versus calculated ASWs levels.
Example:
We pick two issues that are exactly one or two years apart in maturity and calculate their
implied forward breakeven. The forward breakeven is assumed to be the general trend of an
NSA CPI path between them. As an example, for the Jul12s, we first calculate the forward
breakeven trend using Jan12 and Jan14 bonds. The Jan12 and Jan14 maturity NSA CPI, as
implied by the markets, are 220.97 and 229.14 (as of 18 November, 2009), respectively,
giving a forward annualised breakeven trend of 1.83%. Then, we proceed to calculate the
cumulative seasonality between the Jan12 and Jul12. The cumulative seasonality between
the two issues is 71bp. Remember the seasonality should be calculated between the lagged
maturity dates; here it is positive, as the seasonal effect tends to be this way during the first
half of a year. We use seasonality assumptions similar to that of 2008 (CPINSA-CPISA).
Seasonality = Cumulative Seasonality between Jul12 (April15) and Jan12 (Oct15) = 71bp
For this example, Jul12 market implied (with our seasonal assumption) Maturity NSA CPI
should be 225.49 (implied breakeven: 1.50%), while the market is implying it should be
222.77 (breakeven: 1.18%). In other words, Jul12 breakevens should be 32bp higher if one
assumes seasonality similar to 2008. Based on this approach and using 2008 seasonal
assumptions, the market consistently underpriced breakeven levels for July issues relative to
nearby January issues throughout 2009, a continuation of an ongoing theme.
Euro area
Eurostat does not publish official seasonally adjusted HICP series, but the ECB does publish
seasonally adjusted series for both headline HICP and HICPx, using the latter to seasonally
adjust bond breakevens in its macro analysis. The ECB series indicate that there is now more
seasonality in the euro area than in the US. These series use an X-12-ARIMA model similar
to that in the US, but conducted at a higher level of aggregation (only five sub-sectors) and
excluding energy, as the ECB argues that there is no evidence of seasonality in this series
while estimation is volatile. Relative to the BLS estimation at the lowest possible level of
aggregation and with more intervention, the ECB methodology may produce a slightly
higher estimate. However, the exclusion of energy components is somewhat controversial
and tends to reduce aggregate seasonality. An X-12 analysis of the energy series suggests
that any seasonality is highly unstable and unlike the US there are few months in which the
sign of the month-on-month seasonal has stayed the same over the long term. On the other
hand, for example, there has been a strong tendency for petrol prices to rise in the three
months to June for the past 10 years, with petrol prices increasing an average 5%, offset by
a similar fall in the three months to January.
Seasonality in euro area HICPx increased in the early part of the decade, mainly due to
changes in measurements and the deregulation of retail prices in several countries. Eurostat
has encouraged a standardisation of processes across Europe, leading to more seasonality
in the aggregate series due to the timing of distortions being more consistently measured.
This has lessened the tendency for the impact to the aggregate index to be smoothed out.
The clearest example of this is that until 2001, Italy and Spain did not include apparel sales
prices in their price data, but there has been a tendency towards increasing volatility of retail
prices in other countries, too.
We are relatively comfortable to use the ECB data for HICPx seasonality for considering
valuations within the euro inflation market, but given the exclusion of energy components
we would caution against comparison with seasonals in other countries that include this
sub-sector. The Insee statistics institute estimates French inflation seasonality at a headline
level rather than within components or sectors so is not directly comparable to the
equivalent ECB series despite using a similar X-11 approach. However, in practice there is
not a large difference between the relative seasonality using the reported series and
adjusting each on a consistent basis. Figure 250 and Figure 251 show monthly seasonality
Figure 250: Euro HICPx m/m seasonality Figure 251: French CPIx m/m seasonality
-1.0% -0.5%
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
p
Source: ECB, Eurostat, Barclays Capital Source: INSEE, Barclays Capital
estimates for Euro HICPx and French CPIx, respectively, for the past three full years of data.
The trends in Euro and French seasonality have become relatively similar in recent years.
The differences stem mainly from varying sales periods and a greater upward bias for euro
(particularly German) prices for Christmas and Easter that are subsequently unwound.
Euro seasonality is generally priced more efficiently than in other markets. This may be due
in part to more active trading in swaps than elsewhere, as it is very hard to trade swaps
without using a seasonal vector to price non full-year positions. In the cash linker market
there is very little evidence of seasonality biasing performance across the year. Historically
there was a correlation between the performance of OATis versus €is and the deviation of
French and euro seasonality, but this is no longer significant. In terms of micro valuations it
is notable that OAT€is maturing in July mature at a favourable time to benefit from the
upswing in prices during the spring, whereas German issues redeem close to the least
favourable time in April. On average, since the DBR€i16 was first issued in 2006 this bond
has priced around two thirds of the negative seasonality versus the OAT€i breakeven curve.
However, the spread of more recent German issues to the OAT€i breakeven curve have on
average priced seasonality in line with the ECB estimate of seasonality – 0.9% based on
2008 full-year seasonals.
United Kingdom
For the UK, seasonal adjustment of the overall RPI Index derives statistically significant
seasonality, but the Office for National Statistics (ONS) does not produce a seasonally
adjusted RPI series. To gauge the stability and drivers of seasonality, we created a seasonal
vector by using X-12 to adjust the key components of the RPI, then aggregating these
vectors using the official index weights. Statistically significant seasonality was detected in
Seasonal Food, Alcohol and Tobacco, Other Goods and Services ex-Utilities. However,
Council Tax is only raised in April, so is clearly seasonal, and there is a borderline statistical
rejection of Petrol and Oil. While it is true that the seasonality in oil is somewhat unstable,
there is still a clearer pattern than in the euro area, partly due to seasonality of petrol tax
increases. We found that a more consistent seasonal vector was generated when seasonally
adjusting all components other than mortgage interest payments. We avoided deploying an
ARIMA model as part of the X-12 process, as for some components there was no evidence
of an underlying ARIMA process, which would mean an inconsistent seasonality estimate.
Unusually, the seasonal estimate we obtained from this disaggregated level of RPI produced
slightly more seasonality that running X-12 on the aggregate RPIX series.
Figure 252 shows the estimated seasonality for the past three years, which indicates that the
seasonal vector is reasonably stable, although an estimation without stripping out mortgage
interest payments would be notably less stable. The relatively long sample period from
1987-2009 allows for a more accurate estimation of the underlying trend and seasonal
factors than in the euro area. Even so, seasonality appears to be also more stable than the
US and most large euro countries. Figure 253 shows the contribution of the various key
components of RPI to the month-on-month seasonality. As can be seen, there is a strong
upwards April effect due to council tax, alcohol and tobacco. This is understandable as
council tax has always been increased at the start of each fiscal year since it was introduced,
while alcohol and tobacco taxes are consistently raised immediately following the budget,
so almost always showing up in April RPI. Goods show a strong downwards January effect
due to seasonal sales; this is also seen in July. Petrol and oil has consistently shown a
downward contribution in November, while the strong trend towards April price rises from
taxation has eased, even though the aggregate trend towards higher prices in Q2 is
undimmed. The ONS produces a seasonally adjusted series for RPIY, which excludes indirect
taxation as well as mortgage interest payments, but with a significant element of UK
seasonality driven by tax factors we prefer to use our own series.
The degree of seasonality priced into the UK inflation market is harder to observe than in the
US or euro area due to irregularly dated old inflation-linked gilts and the inconsistency of
these with new style issues. Relatively poor inflation swap liquidity at shorter maturities
does not help, making it harder to accurately measure monthly swap rolls that could
otherwise be used to back out implied seasonality. However, even as market efficiency was
curtailed in early 2009, the instability of any seasonality estimates based on headline RPI
and hence distorted by the impact of aggressive rate cuts on mortgage interest costs has
led to an increased need to accurately estimate RPI seasonality. Hence, while historically
seasonally adjusting breakevens has done little to smooth either the curve or time series of
breakevens, it is more likely to be an important factor. In recent years there has been little to
choose between the seasonality of gilt linkers, but the introduction of 22 March maturing
new-style issues, which capture most of the January negative seasonal, creates a relatively
stark differential versus the most common 22 November redeeming bonds, -0.75% on our
seasonality estimate.
Figure 252: Barclays Capital estimated UK RPI m/m seasonality Figure 253: Key contributions to 2009 RPI seasonality vector
Japan
1) Seasonal features of the Japanese core CPI
As is the case with consumer prices elsewhere, the Japanese core CPI (CPI excluding fresh
food) exhibits fairly regular seasonal patterns that reflect corporate sales practices, as well
as consumer habits and preferences. Historically, seasonality has been “most negative” in
the first quarter of the year, ie, Japanese consumer prices in Q1 tend to be lower than prices
in other months regardless of whether the year as a whole exhibits positive, flat or negative
inflation rates. The primary reason behind this is that in the January-March period, prices of
Japanese goods (especially clothing) tend to fall sharply due to the impact of New Year and
year-end inventory clearance sales. Prices of services also tend to decline overall in Q1, as
fees for entertainment facilities, where usage declines in winter, decrease.
In contrast with the first quarter of the year, seasonality tends to improve from the second
quarter of the year and actually becomes “most positive” (or strongest) during the August-
December period. This reflects price hikes in clothing from April, as well as the fact that
during the summer vacation season of July-August, leisure-related service prices, such as
travel, tend to increase significantly. On the other hand, prices of goods tend to decline due
to the impact of sales following summer bonus payments and due to clearing sales on
summer clothing. Clothing prices subsequently rebound with winter lines coming in and
prices of goods such as food and rice tend to rise as preparations for the New Year holidays
get underway, a development that pushes consumer prices significantly above the year’s
average in the last quarter of the year.
Over the past two decades the major items accounting for the index’ seasonality have been:
1) clothing/apparel; 2) recreation and entertainment; and 3) transportation and
communications. Clothing is by far the most volatile item in terms of seasonality,
determining the bulk of the core CPI’s seasonal patterns. In Q1 for example, this item
accounts on average for nearly 60% of the CPI’s negative seasonality (based on monthly
data starting from 1993), while in the September-December period clothing explains nearly
80% of that period’s positive seasonality. Similarly, on average, recreation/entertainment
accounts for roughly 20% of the negative seasonality observed each year in Q1 and it
explains about 40% of the positive seasonality around the summer period (when it is offset
by negative clothing seasonality).
Historically Japanese seasonal patterns have been generally stable, especially around Q1 and
Q4 and over shorter sample periods (eg, five years or shorter), but some changes did occur
over time with regards to the strength of seasonality in Q2 and around the summer period –
in the early 1990s the month of May for example featured stronger positive seasonality than
it currently does (the 5y average ratio of non-seasonally adjusted core CPI to the seasonally
adjusted core CPI was 1.0036 versus 1.0010 under the 2005-base core CPI), while the
month of August exhibited negative seasonality as opposed to it currently being seasonally
positive (ie, a ratio of 0.9974 compared to 1.0016 under the 2005-base core CPI; for further
details please refer to Figure 256). The seasonal patterns followed by Japanese consumer
prices are fairly different from those observed in the US (a m/m correlation of just 24%
since 2000) but similar to those seen in the euro area (an 82% correlation).
2) Seasonal factors
The seasonal patterns of the Japanese core CPI described above can be visualised using
official seasonally-adjusted core CPI numbers published by the Ministry of Internal Affairs
and Communications Statistics Office. These are disseminated on a monthly basis, at the
same time as the release of headline and core CPI figures and past data is available to
January 2000. The seasonality adjustment applied by the Ministry to original core CPI time
series is based on the X-11 ARIMA method originated by the US Census Bureau 18.
Figure 254 illustrates the time series of the ratio between the non-seasonally adjusted core CPI
and the seasonally adjusted core CPI, over the period between 2000 and 2009. The ratio
highlights stable seasonal patterns, with February being the month with most negative
cumulative seasonality and October/November featuring the most positive seasonality.
Furthermore, the graph also reveals no major change in seasonality due to the rebasing of the
core CPI from the year 2000 to the year 2005. This suggests that the seasonal adjustment of
JGBi real yields and breakevens can employ longer-run averages, such as 5y or 3y averages,
without the risk of overstating or understating the impact from core CPI seasonality.
Figure 254: Time series of ratio between NSA core CPI & SA Figure 255: Average m/m change in monthly seasonal
core CPI factors under 2000-base and 2005-base core CPI
0.34%
2000-Base CPI 2005-Base CPI
0.30%
1.006
0.28%
0.24%
0.20%
0.06%
Octo ber is usually the mo nth with mo st po sitive seaso nality . . . 0.4%
0.16%
0.12%
1.004
0.10%
0.10%
0.06%
0.06%
0.04%
0.00%
0.2%
1.002
0.0%
0.00%
1.000
-0.08%
-0.10%
-0.10%
-0.14%
-0.14%
-0.2%
-0.18%
0.998
-0.22%
-0.4%
0.996
-0.47%
February is usually the mo nth with mo st negative seaso nality . . . 2005-Base core CPI
0.992 -0.8%
May
Dec
Nov
Jan
Jun
Sep
Feb
Aug
Mar
Apr
Oct
Jul
2000 2002 2004 2006 2008 2010
Source: Ministry of Internal Affairs and Communications Statistics Office, Note: 2000-base CPI data covers period from Jan 2000 until Dec 2004, and 2005-
Barclays Capital base CPI data is for Jan 2005 – Nov 2009. Source: MIAC, Barclays Capital
2000-Base 0.9962 0.9947 0.9972 1.0006 1.0018 1.0010 0.9996 1.0002 1.0022 1.0028 1.0018 1.0024
2005-Base 0.9964 0.9942 0.9970 1.0000 1.0010 1.0010 1.0000 1.0016 1.0026 1.0030 1.0012 1.0012
2000-2009 Avg. 0.9963 0.9945 0.9971 1.0003 1.0014 1.0010 0.9998 1.0009 1.0024 1.0029 1.0015 1.0019
*1990-1995 0.9964 0.9946 0.9970 1.0021 1.0036 1.0030 0.9995 0.9974 1.0010 1.0020 1.0021 1.0014
Note: 2000-base and 2005-base core CPI data sample from Jan 2000 to Nov 2009. Source: MIAC, Barclays Capital.
18
Although the Japanese core CPI aggregates prices of more than 550 items, the MIAC publishes time series adjusted
for seasonality only for the headline and core CPI along with six other broader sub-indices: CPI excluding imputed rent,
CPI excluding imputed rent and fresh food, CPI excluding food and energy (known as the “core core” CPI), the goods
index, the semi-durable goods index and the index of goods excluding fresh food.
0.47% m/m). In contrast, the largest positive inflation carry due to seasonality occurs
during the 10 July-10 August period, as the m/m change in seasonal factors is most positive
in April (+0.30% on average under the 2005-base). The basis point impact on 1mth carry
will depend on the residual duration of JGBis, becoming larger the shorter the duration of
the bond. For a 5y duration bond, a -0.47% seasonal implies negative carry of over 9bp and
a positive seasonal of 0.30% implies a 6bp positive carry.
Apart from the short-term inflation carry, seasonality also has an impact on the relative pricing
of individual JGBis. In general, inflation-linked JGBs whose base CPI is taken from a month with
poor seasonality tend to trade at a discount relative to neighbouring bonds. June bonds (issues
JGBi2, JGBi4, JGBi8, JGBi12 and JGBi16), whose base CPI is taken from March, in accordance
with the three-month indexation lag, have the poorest seasonality among JGBis’ base CPI
months. In contrast, linkers issued in December (and thus featuring base CPI from September)
enjoy the strongest seasonal factors. The difference in the seasonality of each JGBis’ base CPI
is generally consistently discounted by the market, whether in terms of inflation-adjusted
prices/ yields or breakevens. For instance, issues JGBi2, 4, 8 and 12 have been trading on
average between 2bp and 4bp cheaper versus our parametric real yield curve model, while
other issues have generally been trading fair to the curve. The reason behind this discount is
naturally the fact that the cash flows of weak-seasonality linkers, including the principal at
redemption, are adjusted for inflation based on core CPI index levels taken from March and
September. While half of each June bond’s coupons are paid with a relatively favourable
seasonal, the other half and much more importantly, the final principal repayment are
adjusted for inflation based on March CPI numbers (an average seasonal factor of 0.9970).
In general seasonality is less important in emerging markets due to more unstable inflation
and less regular distortions. In particular in Brazil we see less statistically significant
seasonality in IPCA inflation than in any other major economy, with an X-12 estimation of
seasonality across the year varying by less than 0.4%. Despite measurement changes in
Argentinean CPI, this also demonstrates very limited seasonality, while the seasonality in
Mexico and Colombia is of a similar order of magnitude to that in the US despite having
inflation that is notably more volatile. The Latin American country with the most extreme
seasonality is also the one with the lowest and most stable inflation trends, Chile. Here the
seasonal trend is for prices to rise from February through September before seasonals turn
negative for each of the following five months, such that since 2003 prices have on average
fallen in this negative seasonal period even with inflation averaging almost 4%. Hence a UF
bond redeeming in November would have around 1.9% better seasonal than a bond
maturing in April.
In the more recently launched inflation-linked markets, seasonality has been a factor
encouraging investment during positive carry periods but has not been that important
relative to the underlying volatility of the inflation indices. For instance, in South Africa the
dispersion of seasonality is the same order of magnitude as in the US despite inflation being
around four times as volatile. There is also a higher degree of uncertainty of seasonality due
to the significant change in inflation measurement that occurred from the start of 2008,
which we attempt to minimise by considering seasonality in CPI excluding housing.
Nonetheless, December maturing bonds still have a better seasonality than those redeeming
at the end of March by more than 1%. Turkish seasonality is relatively strong, though not
especially stable. With Turkish linkers being short dated the sensitivity to this seasonality is
relatively high. Prices tend to fall in summer as elsewhere in Europe such as Poland, but
more dramatically due to the particularly heavy weight of food in the Turkish CPI basket and
with a sharp rebound in October. August maturing bonds ought to have the most
favourable seasonality, although February is almost as positive, whereas if a bond were
issued with a maturity at the start of December its seasonality would be up to 2% worse.
Figure 257: Estimated Latin American m/m seasonality Figure 258: Estimated m/m seasonality selected EMEA
0.2% 0.5%
0.0%
-0.2% 0.0%
-0.4%
-0.5%
-0.6%
-0.8% -1.0%
Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov
We first look at the construction of a forward breakeven trade in bond space and use TIPS in
our illustration. For the sake of clarity, we first define what a breakeven is and how a spot
breakeven trade is constructed. A breakeven rate is the difference between the real yield of
an inflation-linked bond and the nominal yield of its nominal comparator.
Essentially, this represents the market’s expectation of inflation over a given period, plus an
inflation risk premium, less a liquidity premium differential (for a detailed discussion of
these concepts, please see Breakeven inflation and the relationship between real and
nominal yields earlier in this publication). A pure breakeven trade is a DV01-weighted one.
The ratio of notionals of the linker and nominal legs will be the inverse of their DV01 ratio, in
the same way that a DV01-weighted curve trade in nominal bond space would be
constructed. We highlight, however, that the DV01 of a linker is a “real” DV01, ie, expresses
the change in the linker’s full price following a 1bp move in its real yield. Figure 261shows
the cash flow and weighting details of a long breakeven trade on the TIIJul14.
Figure 259: Our conceptual measure tracks spline-based 5y5y Figure 260: Forward TIPS breakeven curve
3.5%
3.5%
2.85% 3.0%
3.0% 2.5%
2.35%
2.0%
2.5%
1.5%
1.85%
2.0% 1.0%
0.5%
1.5% 1.35%
Jul-03 Jul-04 Jul-05 Jul-06 Jul-07 Jul-08 Jul-09 0.0%
5y5y Spline Zero-Breakeven (Nominal-Real, LHS) Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16
ModDur Weighted 5y5y BEI (RHS) Forward 1yr breakevens
TIPS (Buy) TII 2Jul 14 122.72 0.43% 5.31 1.14721 100.0 (122.7)
NOMINAL (Sell) T 2.625Jul 14 102.59 2.30% 4.33 122.7 125.9
Breakeven 1.87% Cash residual 3.1
Note: As of 14 January 2010. Source: Barclays Capital
There will be some residual starting cash flows (due to dirty price mismatch), as well as
coupon accrual differential depending on the path of inflation and reinvestments. The
return attribution to this residual when compared with realised – expected breakeven
returns will be small. A breakeven position can really be thought of as a synthetic par
breakeven bond. The risk measure of such a bond would be:
Residuals (bp) is the offsetting residual costs of two simultaneous breakeven positions (two
needed to construct a forward breakeven rate) described above. These residuals exist
because we are approximating forwards using cash coupon bonds, whereas bootstrapping
spot rates to create forwards is only accurate using zeros. An approximate ForwardBEIRate
can be derived in the same way as forward nominal rates using zeros. Here, we assume zero
breakeven rates (BEI) to be spot breakeven rates of equivalent maturity bonds. We further
assume constant rates and continuous compounding to arrive at a ForwardBEIRate.
EXP( BEI 2 * T2 )
EXP (ForwardBEIRate * (T2 – T1) )= = EXP ( T2 * BEI 2 − T1 * BEI1 )
EXP( BEI1 * T1 )
For zero-coupon bonds, the modified duration (D) is roughly equal to its time to maturity.
So the time component in the above formula can be replaced with modified duration
(breakeven modified duration). We use TIPS modified duration for approximation purposes.
( D2 * BEI 2 − D1 * BEI1 ) ( D1 ) , W = ( D2 )
ForwardBEIRate = , W1 = 2
( D2 − D1 ) ( D2 − D1 ) ( D2 − D1 )
1.1472
TIPS TII 2 Jul 14 122.72 0.43% 5.31 1 100.0 122.7 4.32, (4.96)
Nominal T 2.625 Jul 14 102.59 2.30% 4.33 122.7 125.9 4.22
Breakeven 1.87% Residual 3.1
1.0127
TIPS TII 1.875 Jul 19 106.84 1.34% 9.24 8 113.3 121.0 8.71, (8.82)
Nominal T 3.625 Aug 19 100.28 3.74% 7.95 131.6 132.0 7.90
Breakeven 2.40% Residual (11.0)
Net
Fwd BEI 2.92% (Residual) (7.9)
Source: Barclays Capital
For a 5y5y fwd breakeven rate, with respective modified durations for 5y and 10y of 4.33
and 8.73, the above weights (W1 and W2,) would have roughly a 2-to-1 ratio, meaning the
Breakeven_DV01 of the longer security has to be twice that of the shorter security. This
forces the inflation-adjusted notionals of the two TIPS securities involved in forward
breakevens to be the same. This is an important step for a forward position because equal
TIPS notionals force the accrual inflation gains/losses prior to the forward start date to be
zero. For a forward breakeven position, we want inflation exposure to be defined from the
forward start date and not prior.
= 113.3*1.01278 – 100*1.14721 = 0,
In other words, till the maturity of TIIJul14, the position will not incur any P&L due to net
notional inflation accretion. In the example below, we show how this translates to notionals
on a 5y5y fwd breakeven trade.
In Figure 263, we look at how this position performs in various breakeven curve shifts and
changes in forward rate. Given the index ratio-weighted positions ($100mn to $113.3mn) on
each breakeven leg with modified durations of 4.32 and 8.71, if breakevens across the curve
rise in parallel by 10bp (fwd breakeven will rise by 10bp), the long position in 10y will rise in
value by $1mn, and the short position in 5y will lose $0.5mn, netting $0.5mn of positive P&L. If
the 5y breakeven falls by 10bp while the 10y breakeven is unchanged, the forward breakeven
will also rise by 10bp (breakeven curve steepens), and the 5y leg incurs positive P&L of
$0.50mn. This P&L matches the net P&L of a 10bp parallel shift in forward breakeven position.
We also show the declining forwards (-10bp) scenario through parallel curve shifts, as well as
curve flattening. In both 10bp fwd breakeven narrowing scenarios, the trade loses $0.50mn.
Therefore, we feel confident that the above-described weighting method approximates
forward rate exposure, even though we are using cash bonds rather than zeros.
Figure 263: Index ratio-weighted breakeven position gives inflation exposure only
$100mn short $113.3mn short TII Fwd
TII Jul14 position P/L Jul19 position P/L BEI Overall
(BEI_DV01: 4.96) ($mn) (BEI_DV01: 8.82) ($mn) rate P/L ($mn)
− Jul19Notional = $112.58mn
Swaps are a more natural instrument to look at forward inflation pricing than bonds. This is
because the commonly traded structure in inflation swaps is a zero coupon one, such that a
‘clean’ forward rate can be bootstrapped from the curve without any distortion from
coupon payments. The generic formula to calculate an inflation swap rate with a tenor S, at
a forward date F is:
(1 + Y )F +S −1
(1 + X )F
S
Where: Y is the zero-coupon rate for a spot starting swap of tenor F+S
Alternatively, expressed in terms of CPI Index values, the above formula can be rewritten as:
E (CPI F + S )
S −1
E (CPI F )
Where: E(CPIF+S) is the expected value of the CPI Index in F+S years
However, the above formula is not equivalent to the strict theoretical computation for a
forward rate. If expressed in CPI values, the forward rate is theoretically expressed as:
⎛ CPI F + S ⎞
S E ⎜⎜ ⎟⎟ − 1
⎝ CPI F ⎠
In other words, the theoretical computation of a forward inflation swap rate is related to the
expectation of the ratio of future CPI values. The generic formula, on the other hand,
computes a ratio of expectations. Mathematically, they are not equivalent given that the
future CPI values are not independent variables. A convexity adjustment therefore needs to
be applied to the ‘naïve’ forwards calculated from the generic formula. The need for this
adjustment is evident when pricing y/y structures that are effectively portfolios of
instruments priced from forward starting swaps. Caps and floors on inflation are such
instruments. Inflation market models such as those by Belgrade, Benhamou and Koehler
provide a convexity adjustment formula when computing the forward value of the CPI ratio.
For an intuitive grasp of the notion of convexity in forward inflation swaps, we look at the
dynamic hedging of a forward swap position. We assume a 5y5y forward swap trade in
which an end user goes long inflation, ie, pays the compounded quoted fixed rate and
receives accrued inflation at maturity. The other counterparty (typically a bank) will hedge
the 5y5y forward position through a combination of 5y and 10y zero coupon inflation
swaps at the inception of the trade; the hedge will be a long position in 10y and a short
position in 5y swaps. By definition, a forward starting swap should have no accretion from
realised inflation before the forward date. Therefore, the notional on the 5y and 10y swaps
should be equal to immunise the inflation accretion before the forward date. While this may
seem straightforward, the delicate element is setting the initial notional on the swaps.
Assuming a €100mn notional on the 5y5y trade, and assuming positive inflation over the
five years preceding the forward start date of the swap, then setting the notionals on the 5y
and 10y at €100mn is likely to result in an overexposure to the 5y5y forward rate. After five
years, the initial 5y swap would have expired but the initial 10y one will effectively be a swap
with a remaining term of five years and its notional will have accrued with actual inflation
and risen above the €100mn that needs to be hedged. Hence, if positive inflation is
expected over the first five years, the notional on the swaps in the hedge portfolio should be
set at less than €100mn. The €100mn notional of the forward swap therefore needs to be
discounted with inflation that is expected over the initial 5y forward horizon. In practice, the
5y zero coupon inflation rate at inception serves as the best guess for this expected
inflation. Yet, unless actual inflation proves to be equal to the expected inflation, the
notional will still need to be rebalanced if inflation overshoots or undershoots initial
expectations. This rebalancing mechanism illustrates the convexity effect at work.
Let us assume that after inception of the hedging strategy, inflation is significantly higher
than initially discounted. This would point to a potential over-hedge at the end of the
forward horizon. The hedger will therefore need to reduce the notional on the 5y and 10y
swaps by effectively unwinding part of the hedge. The likelihood is that the unwind will be
done at a profit, excluding transaction costs, given that breakevens tend to rise with higher-
than-expected inflation. Assuming a higher breakeven curve, a profit will be made on the
partial unwind of the 10y swap, but a loss will be made on the 5y leg. However, given the
higher inflation PV01 of the 10y leg, the net effect should be a profit, unless the curve
flattens significantly and/or the discount factor on the longer swap is significantly lower. On
the other hand, if actual inflation undershoots initial expectations, then the notional on the
hedge will need to be increased. In this case of lower-than-expected inflation, the breakeven
curve will tend to move lower, which implies that the increase in the hedge notional is likely
to be done at cheaper levels than initially. In this case too, this is positive for the hedger.
Hence, there is a positive convexity effect in hedging a short position in forward inflation
swaps. This effect comes from the likely positive correlation between actual inflation and
expected inflation, ie, breakevens. If this convexity is passed on to the buyer of forward
inflation, then the quoted forward rate should be lower than what is implied by the ‘naïve’
forward. Obviously, this convexity effect will work the other way round for a seller of
forward inflation.
It is, however, difficult to re-adjust a hedge portfolio frequently to take into account the
convexity effect, given that the magnitude of the latter tends to be relatively small
compared to bid/offer spreads on swaps. Typically, because of transaction costs,
rebalancings due to convexity are rarely done more than once a year.
Although the hedging of a forward swap position may not be straightforward conceptually,
taking positions is relatively easy in the developed swaps markets. Forward positions are
directional ones, with an added exposure to the curve, as in nominal rates space. This is
obvious from the generic forward formula above. Other things being constant, an inflation
swap with a tenor of S years, starting in F years has a positive sensitivity to the spot swap
rate with a S+F year tenor. This is the directional element. Although the sensitivity to the
curve is less straightforward formulaically, a steepening/flattening curve move will tend to
increase/decrease the forward rate. Hence, the best configuration to enter a long/short
forward trade is when bullish/bearish expectations are blended with a
steepening/flattening view. However, given that the slope of the curve tends to be inversely
related to the level of breakevens, the directional element has historically been the strongest
driver of forward swaps. Forward strategies therefore seem better suited to fade unjustified
relative value distortions on the curve, rather than targeting curve slope moves.
400
350
300
250
200
150
100
euro HICPx US CPI UK RPI
50
0
0 5 10 15 20 25 30
Source: Barclays Capital
The analysis of inflation swaps can be broken down in 1y forwards, an insight that is not
possible in bond markets given the limited number of maturity points. Micro distortions on
the swap curve and cheap segments are therefore more easily identified: for example,
Figure 264 shows that 1y forwards between the 10y and 15y on the US CPI swaps curve
tend to be low. While such granularity can be useful to fine-tune relative value strategies,
some distortions may sometimes be justified. For instance, the relative cheapness of the
segment between the 10y and 15y points on the US CPI swaps curve is largely explained by
illiquidity in that part of the curve, which itself stems from the lack of corresponding TIPS in
this gap. In the euro and UK markets, where asset swapping is an important feature, heavy
asset swapping on a specific issue could explain the cheapness of the corresponding
maturity on the swaps curve.
The most obvious use of inflation swap (and bond) forwards is within fundamentally-driven
strategies. Inflation is a tangible economic phenomenon, on which most central banks
around the world have more or less official target levels or ranges. Within developed
economies, the UK and the euro area have explicit targets/ranges on the desired inflation
levels, which imply that expectations of inflation beyond the short term should remain
anchored unless the central bank’s policy is not credible. The deflationary period amid the
crisis has some interesting elements with regards to this; 5y5y forwards have certainly been
volatile during this period but significantly less than spot 10y rates. Inflation forwards can
indeed be related to a policy-related “fair value”, which reduces the probability for extreme
valuations. Strategies to fade sharp divergences from the perceived fair economic value of
forward inflation swaps are relatively common. Towards the end of 2009/start of 2010,
5y5y forwards stood at relatively elevated levels, probably pricing in the probability that
beyond the short term, extreme simulative economic policies implemented during the crisis
will generate inflation above desired levels. This can be seen as a justified risk premium, in
which case fading elevated levels may not have a good rationale.
Figure 265: Forward inflation rates less volatile than spot rates
3.0
10y euro HICPx swaps
2.7
5y5y euro HICP swaps
2.4
2.1
1.8
1.5
Jun 05 Mar 06 Dec 06 Sep 07 Jun 08 Mar 09 Dec 09
Source: Barclays Capital
It is true that the inflation and deflation scare of 2008-09 failed to kick-start a self-
sustaining market in inflation options. The UK remains the only market where non-linear
activity (ie, in LPI swaps) finds support from the regulatory framework for the pension fund
industry and even here activity has been discouraged by the realised volatility. However, we
highlight that despite limited, and sometimes very poor, liquidity in the US and euro inflation
volatility markets, it is now much easier to obtain implied volatility data on which pricings
can be done. The general approach before consisted mainly in looking at historical
volatilities as a starting point to estimate implied volatilities. The obvious drawback is that
historical volatilities are just one component of implied volatilities; a risk premia based on
flow/hedging consideration further needs to be added.
More readily observable implied volatilities in the caps and floors market are valuable inputs
in other inflation volatility products. For instance, the pricing of TIPStions and options on
breakevens can then be based partially on a market-related volatility input, rather than
simply on a calculation of realised real yield or breakeven volatility which would entail much
more subjective elements. This should tend to encourage more consistent pricing which, in
the long term, can help create a self-sustaining market.
The model consists of a three-factor framework in which the three components are the
nominal, real, and inflation rates. The nominal and real interest rates are assumed to follow
an HJM diffusion process. The CPI is driven by an instantaneous inflation rate defined as the
difference between the nominal and the real interest rates. The Jarrow-Yildrim model
provides arbitrage-free conditions between the three components. In their paper, the
authors obtain the nominal and real rate term structures by applying standard stripping
techniques to nominal US Treasuries and TIPS. Volatility parameters for the nominal and
real forward rates are computed from historical data on TIPS and nominal US Treasuries,
while the volatility of the inflation rate is derived from the CPI time series. Finally, estimates
of the correlation parameters between the three components are calculated through sample
moments using historical inflation, real, and nominal interest rate data.
Advantages of this model are its simplicity and intuitiveness and the fact that its framework
is easy to implement. Also, in the particular case where the CPI process is linked to an
instantaneous inflation rate, it provides closed form solutions for inflation swaps and Black-
Scholes formulas to evaluate inflation options, whether they have a zero coupon or a y/y
format. However, its main drawback is that it is particularly suited to markets in which
calibration needs to be done from data on the bond market. This is especially problematic
for the euro area inflation options market, where the more natural curve for data calibration
would be the inflation swap curve. To this end, other models which fall under the ‘market
models’ category have emerged as more suitable candidates.
One such approach is the one developed by Belgrade, Benhamou and Koehler. They link the
zero-coupon and the y/y inflation derivatives in the European inflation swap and options
markets through a market model. Forward values of the CPI are modelled. The authors
provide a convexity adjustment formula when computing the forward value of the CPI ratio.
They show that the forward value of the ratio is the respective forward CPI multiplied by an
adjustment that is an explicit function of the forward CPIs, the forward zero coupon bond
and the correlations between them. Unlike in the Jarrow-Yildrim framework, the availability
of the CPI forward is considered to be sufficient such that real rates are not used as an
input. As for the nominal curve, an HJM-type diffusion is assumed. The calibration to market
data is done using money market and swap prices for the nominal zero coupon term
structure. Traded optional instruments are used to define the nominal volatility structure.
The authors give closed formulas for the valuation of breakeven swaptions and numerical
integration for options on real yields.
Sophisticated models are needed to cater for the complexity in volatility products. For
example, a common but complex (in terms of pricing) product is the LPI swap. An LPI swap
is an interest rate product dependent on the path of inflation. Hence, to value the product a
simulation of annual inflation rates is needed up to the maturity of the swap. The simulation
process needs to be carried out within an inflation model. The simulation of nominal rates
can be implemented through a standard HJM framework.
In the euro inflation market, the most liquid non-linear inflation instruments are caps and
floors. Traders build the inflation volatility term structure (vol as a function of expiry) for
individual ATM caplets/floorlets (not directly quoted in the market) so that they can match
ATM cap/floor straddles (quoted in the market). The SABR (stochastic volatility) model
seems to be a natural choice as a calibration tool for inflation smiles. As inflation can
become deflation and drop below the 0% level (as happened in several countries in 2009),
therefore the SABR parameter beta is set close to zero (ie, assuming the normal-like
distribution for inflation rates). Inflation volatilities implied from option prices are then
calibrated in the SABR model, in which the correlation (an inflation rate versus its volatility)
and the vol-of-vol (volatility of inflation volatility) parameters determine the “skewness”
(asymmetry) and the “smileness” (curvature) of vol smiles. Typically, the correlation
parameter tends to be negative up to the 10y expiry. When inflation goes down, there is
more demand for low-strike, shorter-dated floors as a protection/hedge, hence inflation
volatility is supported and goes up. On the other hand, positive correlations make sense for
longer-dated options (10yrs and out). When inflation goes up, pension funds should be
natural buyers of high-strike, long-dated caps to hedge their liabilities.
Deflation floors
In the previous Inflation-Linked Users’ Guides that Barclays Capital has published, there has
been usually only a passing mention of the embedded deflation floor. This is primarily
because the value of the floor was seen as miniscule and only very rarely were there pactical
questions. However, during the financial crises when US breakevens out to the 9y turned
negative and investors were increasingly risk averse, the value increased significantly: for
example, at one point the real yield spread between TIIApr13s and TIIJul13s was 200bp,
with most of this difference explained by the floor. The non-linear inflation market began
quoting cumulative caps and floors by early 2009 and the value of the embedded option
could then be priced separately from TIPS. In the following, we explain the details of the
floor and discuss its effect on potential returns to maturity.
Several inflation-linked markets, including the US, euro area and newer issues in Sweden,
have an embedded par floor such that at maturity, the investor gets the greater of par or the
inflation-adjusted principal. Since the inflation-adjusted principal is the par amount times
the index ratio (which is the ratio of the reference CPI to the base CPI), this is another way
of saying that, at maturity, the index ratio is floored at 1 as it applied to the principal. The
pay-off on the principal amount at maturity can be written as:
It is important to remember that the “strike” on the floor is at par, or an index ratio of 1, not
where the index ratio is at the time of purchase. For this reason, the floor value of newer
bonds tends to be more valuable because the index ratio is typically lower than seasoned
TIPS. As an example, the TIIApr13 Index ratio, for a trade that settles on 25 January 2010, is
1.02332. This means that there has been 2.33% cumulative inflation accrued since issuance
in April 2008. The floor would kick in if the index ratio fell below 1, so the inflation accrued
since issuance would first need to be fully reversed out in a period of deflation. With about
3.2 years left, there needs to be 0.71% annualised deflation to maturity for the floor to be at
the money at maturity.
1) The inflation that has already accrued on the bond – The higher the current index ratio,
the lower is the strike on the embedded floor.
2) The remaining time to maturity of the bond – The higher the maturity, the closer the
strike is to zero. This means that for short maturity bonds, only those with little inflation
accretion (likely those that have recently been issued) will tend to have floors with
significant value.
3) The level of breakevens – This determines the extent to which the floor is considered to
be in the money. A fall in the bond’s breakeven brings the floor closer to the money and its
theoretical value can increase substantially. This element is likely the most important one.
The remaining time to maturity is a known parameter and inflation accrual changes
relatively slowly. On the other hand, breakevens can be subject to big changes over a
relatively short period. Figure 268 shows the decline in estimated floor premium levels with
the rise in breakeven levels.
Figure 266: Realised CPI swap volatily is generally much higher than realised CPI NSA Vol
30
bp/day
20
10
-
Feb 04 Feb 05 Feb 06 Feb 07 Feb 08 Feb 09
Source: Barclays Capital
In practice, gauging the fair value of floors on linkers is not trivial. An option-pricing
framework would be the natural starting point for valuation purposes, but the limited depth
of the inflation options market means that such an exercise is not straightforward. In the US
for instance, poor liquidity in inflation volatility products means that implied volatility levels
are not easily observable. The asset swap market, however, provides an indication about the
value that the market gives to embedded floors in linkers. This is because the price of a
linker, as quoted in the market, will incorporate the market’s valuation of the floor. The
market-quoted asset swap level will be based on the quoted linker price but will also take
into account the floor value that may, for example, be considered as an up-front option
premium. The effect of the floor on the quoted asset swap valuation should therefore, at
least theoretically, be neutral. On the other hand, if the asset swap is calculated in the usual
way by taking the quoted price and without adjusting for the value of the floor, then the
resulting asset swap will tend to be richer than the asset swap quoted in the market. For this
reason this relative richness is an indication of the value of the embedded floor (Figure 267).
Figure 267: Calculated and traded TIPS ASWs difference is Figure 268: Estimated floor premium declines as BEIs rise
roughly floor premium
These notes left dealers structurally short inflation floors. Although the bulk of such positions
were likely in 0% strikes, higher strikes were not uncommon. From a hedging perspective, the
issuance of such notes would have required offsetting long positions in inflation floors, as a
perfect hedge, or short positions in inflation swaps if only delta hedging is considered.
However, with breakevens relatively stable and around 2% prior to 2008, the mark-to market
value of low-strike floors was then negligible. During H1 08, breakevens surged on higher
inflation and heavy structured note issuance. Although implied inflation vols increased
substantially, most embedded floors were deeply out of the money, even those with above
zero strikes. The tide however turned as breakevens collapsed from the levels reached in
summer that year. The impact on short floor positions was not heavy until when short-dated
swaps fell below 1% in October. Given that most embedded floors carried a 0% strike, the
sharp bearish momentum in short-dated swaps increased their negative mark-to-mark
valuations as levels drifted towards zero. Dealers tried to cover their short positions by buying
mostly 5y 0% floors in the broker market, pushing floor vols sharply higher.
This trend had important implications for zero coupon inflation swaps. Indeed, vega and
delta risk on floors can only be hedged perfectly simultaneously with other floors. However,
the depth of the inflation options market relative to outstanding embedded exposures is
small. Dealers often tackle the vega exposure in inflation options using nominal vols as a
proxy hedge. On the other hand, the delta exposure can be more easily hedged using
swaps. Given the illiquidity of the inflation swaps market towards the end of 2008 and the
fact that most dealers had similarly short exposures in inflation floors from structured notes
issuance, this put further substantial downward pressure on euro HICPx swaps. This was
especially true in short maturities, given the greater concentration of structured notes there,
the steepness of the breakeven curve and the greater gamma on shorter-dated floors. The
need to delta hedge the short exposures in inflation floors created a self-reinforcing bearish
momentum in breakevens, an effect that had already been seen a few weeks earlier in the US
where the initial shock was larger despite the size of US CPI floor exposures being smaller.
Indeed, as swaps breakevens moved lower, the embedded floors also moved closer to the
strike. A dealer delta hedging a position therefore needed to increase its shorts in inflation
swaps as inflation breakevens moved lower, given that a typical floor would then be
increasingly sensitive to moves in the underlying (ie, the swap breakeven rate).
Figure 269: Impact of delta hedging on euro HICPx swaps Figure 270: Extreme richening in implied vols in 2008-09
3.0 200
5y euro HICPx swap
2.5
150
2.0
100
1.5
delta hedging of
short floor
50
1.0 positions 5y Inflation YoY euro HICPx Capvol
10y Inflation YoY euro HICPx Capvol
0.5 0
Jun 05 Mar 06 Dec 06 Sep 07 Jun 08 Mar 09 Dec 09 Jan 08 Jun 08 Nov 08 Apr 09 Sep 09 Feb 10
Price trends in the past decade have followed two discernible patterns. First, goods price
deflation was evident in the early part of the 2000s as producers in emerging economies
raised their production capacity, resulting in greater competition at a global level. Second,
and also related to the increasing importance of the emerging economies, a step-up in
demand for food and energy has meant that these factors had a more substantive bearing
on global consumer-price indices in the latter part of the decade.
2007-08 is the most obvious example of rising food prices at an international level, with the
price of cereals reaching their highest levels in almost 30 years in 2008. Thereafter,
international food prices declined as the latter part of 2008 saw a fall in oil prices and a
reduction of food demand related to the global recession. Looking ahead, food prices are
likely to remain an important consideration for inflation investors. The Organisation for
Economic Co-operation and Development (OECD) and the UN’s Food and Agriculture
Organization (FAO), for example, have argued that some of the factors behind the 2008-09
food-price spike could sustain higher prices over the next decade. These issues include
draughts in key grain-producing regions, increased biofuel feedstock demand, high oil
prices and US dollar depreciation.
For a number of countries, particularly developed ones, housing costs carry the largest
weight in the CPI. Among these countries, where attempts have been made to incorporate
Figure 271: Developed economies’ CPIs, % y/y Figure 272: Emerging economies’ CPIs, % y/y
7 US CPI 45 Brazil
E13 HICP ex. tobacco Argentina
6 40
Japan CPI ex. perishables Mexico
5 UK RPI 35 Israel
4 Sweden CPI 30 South Africa
3 25
2 20
1 15
0 10
-1 5
-2 0
-3 -5
97 99 01 03 05 07 09 00 02 04 06 08
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
owner-occupied housing costs, this has been a source of volatility in their overall CPIs.
Countries that have taken into account these costs, notably the US, UK and Sweden, see
that housing overall comprises a significant share of their price baskets and has made
significant contributions to CPI inflation over the past decade, commensurate with the
housing booms witnessed in those economies.
There are different approaches to accounting for owner-occupied housing costs, with the key
issue that a consumer-price basket would view housing as an investment and not a
consumption good. In the US CPI basket, this factor is accounted for by owners’ equivalent
rent of primary residence (OER) sub-component (which has a weight of 24.4%). Specifically,
this refers to the perceived rental value of property. In the UK, the housing component of the
RPI includes mortgage interest payments (MIPS: cf. weight: 4.1%) and “depreciation” (cf.
weight: 5%). The depreciation element represents expenditure by owner-occupiers necessary
to maintain their dwelling at a constant quality. The majority of countries have changed their
methodologies for housing costs in the last three decades, with the OER approach most
widely adopted when this has occurred, most recently by South Africa at the start of 2009. In
the euro area, HICP series do not include any component for only occupied housing despite
Eurostat studying include a factor to add them for at least seven years. In its last published
assessment of this project in March 2009 it was suggested that a net cost approach be used,
which if implemented would effectively link HICP to new home prices only.
Energy costs, whether manifest in gasoline prices or household utility costs, also have a
major weighting across countries. These factors have been a source of volatility for
consumer-price indices. In 2009, for example, gasoline price inflation was very negative
across countries, reversing the trend that prevailed through 2008. An interesting aspect of
these moves was that swings in US gasoline price inflation more closely reflected changes in
the oil price. This can be explained by the fact that the US has the lowest gasoline tax
among industrialised countries. In Europe, where gasoline duties are higher, a given
Administered prices
Another issue to consider when comparing the structure of consumer-price baskets across
economies is that of price changes that are “administered”. That is, the price of a good or
service that is determined directly by producers, which can be influenced or set by
governments and their agencies without reference to market forces.
This factor is an important consideration for students of consumer-price indices for three
reasons. First, CPI modelling techniques related to business cycle dynamics, for example,
are not particularly valuable in predicting sub-components of the basket that are not subject
to demand-supply forces. Second, the administered part of the CPI basket carries an
element of risk that is more political, than economic. In other words, governments and their
agencies have to ability to apply price changes to these goods and services in an
extemporaneous manner.
The presence, treatment and weight of administered prices differ across countries. While
this effect is largely absent in Sweden, Israel and the US, in countries where such prices
exist, there are differing views as to which sub-categories pertain to such a classification.
For the purposes of this article, administered prices can be defined as follows. Prices not
covered by these definitions would include goods and services that are subject to excise
duties and indirect taxes – alcohol, tobacco and petrol, for example.
Applying these definitions yields the weights of administered goods and services, as
outlined in Figure 274. Across economies, energy and transport emerge as the two most
heavily administered sub-components of the price basket. Energy prices include gas and
electricity prices, while transport costs include items such as costs pertaining to owning and
running automobiles, public transport and road and air travel. Meanwhile, in contrast to
most other economies, housing costs include a notable administered component in the UK.
This is in the form of council tax (4.0% of the RPI) – a system of local taxation that is based
on the value of dwellings – that is set by local authorities. In Latin America, a significant
portion of the administered basket is taken by transport costs, broadly taking the form of
gasoline and public transport fares.
In the euro area and South Africa, steps have been taken by the statistical agencies to
account for these effects. For the euro area, Eurostat estimates that 10.7% of the HICP
basket is administrated, with 14.5% of the CPI basket in South Africa subject to such
controls. Figure 275 and Figure 276 show that, for both countries, excluding this portion of
the basket yields inflation series that have broadly similar trends to the ‘headline’ series over
time. A notable exception, however, is in the euro area in 2004, when administered prices
were raised 3.6%, while the HICP as a whole increased only 2.1%. This disparity can be
largely explained by healthcare reform that led to a spike in such costs.
Euro area 2.1 2.2 1.6 1.2 0.2 0.5 3.0 10.7
Japan 3.1 0.0 0.0 3.4 0.4 0.0 6.8 13.7
UK 1.4 4.6 4.0 0.7 0.1 0.0 1.1 11.9
Brazil 4.5 4.7 0.0 10.6 4.9 0.4 4.3 29.4
Colombia 0.0 3.8 2.6 9.0 0.0 0.0 0.0 15.3
Mexico 0.9 4.1 0.0 8.5 3.5 0.0 0.1 17.2
South Africa 3.31 1.87 0.0 3.9 2.9 0.0 2.5 14.5
Source: Haver Analytics, national statistical agencies, Barclays Capital
Figure 275: Euro area HICP (%, y/y) Figure 276: South Africa CPI (%, y/y)
5 20 CPI
CPI - administered prices
4
15 CPI ex. admin. prices
3
2 10
1 5
0
HICP 0
-1
HICP - administered prices
-2 HICP ex. admin. prices -5
02 04 06 08 02 04 06 08
Source: Eurostat, Haver Analytics, Barclays Capital Source: Haver Analytics, Barclays Capital
4
3
2
1
0
-1
Other Education Medical care
-2
Transport Apparel Housing
-3
Food & beverages
-4
97 99 01 03 05 07 09
-1
-2
97 99 01 03 05 07 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
1.0 Food
0.0
-1.0
-2.0
-3.0
97 99 01 03 05 07 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
6
5
4
3
2
1
0
-1 Food & catering
Alcohol & tobacco
-2
Housing & household expenditure
-3 Personal expenditure
-4 Travel & leisure
97 99 01 03 05 07 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
4 Transport Housing
Food
3
2
1
0
-1
-2
-3
-4
97 99 01 03 05 07 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
20 Clothing
18 Communication
16 Domestic goods
Food & beverages
14
Housing
12
Transport
10
Other
8
6
4
2
0
-2
01 02 03 04 05 06 07 08 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
-2
-4
01 02 03 04 05 06 07 08 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
-2
01 02 03 04 05 06 07 08 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
16 Food
Clothing
14 Housing
12 Health
Transport & communication
10 Education
8 Other
6
4
2
0
-2
01 02 03 04 05 06 07 08 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
-5
04 05 06 07 08 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
-2
-4
01 02 03 04 05 06 07 08 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
8 Food
Clothing
6 Housing
Health
4 Transport & communication
Education
2 Other
-2
-4
01 02 03 04 05 06 07 08 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
14 Food
12 Clothing
Housing
10
Health
8
Transport
6 Education
4 Other
0
-2
-4
03 04 05 06 07 08 09
Source: Thomson Datastream, Ecowin, Haver Analytics, Barclays Capital
The conventional output gap estimates, which get revised retrospectively, are problematic
for real-time forecasting and policymaking, since the revisions can be significant and even
of sufficient magnitude to alter the sign. However, by supplementing conventional
(retrospectively revised) output gap estimates with survey-based estimates (which are
not revised retrospectively, and so have better “real time” information), we can form
stronger conclusions about the output gap, particularly at a time such as the present,
when both measures signal a very wide negative output gap.
The reduced sensitivity of the change in core inflation to the output gap seen in many
countries since the early 1980s is associated with a decline in published inflation rates
and an associated decline in the variance of inflation expectations. In the context of a
move towards more independent central banks and more visible inflation targets, this
may also reflect a stronger role for inflation expectations in determining price setting
behaviour (as can be seen by contrasting the relatively stable relationship of the output
gap to underlying inflation in Germany since the early 1970s with that of the US and rest
of euro area).
The currently large negative output gaps across all economies imply downward
pressure on core inflation rates, which means deflation risks have not disappeared. That
said, headline CPI measures are likely to continue to have a positive differential over
core inflation in future years, given the likely continuation of commodity prices and
government indirect tax increases outpacing core inflation rates.
Figure 290: The relationship between the change in G3 inflation and output gap
estimates
0.5 1
0 0
-0.5 -1
-1 -2
-1.5 -3
BarCap survey based output gap (4 qtr ma)
-2 -4
Underlying inflation change (4 qtr ma of 4 qtr change in annual % rate)
-2.5 Output gap (composite measure 4 qtr ma, RHS) -5
82 86 90 94 98 02 06 10
Source: Haver Analytics, Datastream, Barclays Capital
In this article we seek to focus on the relationship between core inflation and the output gap,
examining data from the US, euro area and Japan. In so doing, we seek to show that output gap
estimates, while vulnerable to significant retrospective revision, can still offer some insights into
the future behaviour of core inflation, and hence contribute to inflation forecasts.
Constructing demand-sensitive As part of our analysis, we have constructed our own series of core/underlying consumer
core consumer price indices price inflation. We have done so primarily to obtain a series that represents prices that are
likely to be particularly influenced by demand factors, as opposed to supply shocks
(commodities) and government interventions. The underlying inflation measures that we
have constructed and that we use in the following analysis exclude not just commodity
prices but also, in the case of the US, those price series that are estimated by statisticians
(such as imputed rent 19, certain financial services, IT and photographic products, and used
cars). The series we have constructed also excludes the effects of the major changes in
indirect taxes and administered prices (such as the sales tax increase in Germany and Japan,
as well as tobacco 20). Our series are chain-weighted and, for the US and Japan, based on
consumer expenditure weights and price indices (which helps to avoid breaks in the
headline CPI series).
It is important to appreciate from the outset that core inflation measures typically reflect
50-70% of CPI measures, by no means accounting for all inflationary pressure. As well, in a
highly globalised and integrated economy characterised by excess labour supply and
pressure on limited commodity stocks, with economic growth tending to be led by
emerging economies, it is likely that headline inflation measures will tend to outpace core
inflation measures, as has been happening. Another reason to expect headline inflation to
outpace core inflation is the record dimensions of budget deficits in all major economies: it
is very likely that governments will continue to rely heavily on indirect taxation as a source
of additional revenue generation.
As it is, if we compare the difference of “headline” annual inflation measures in the US, euro
area and Japan with our core measures, we find that in the past ten years the mean
differential has averaged 0.78pp for the US, 0.49 for the euro area and 1.00pp for Japan.
This can also be seen in the comparison of our “core” inflation rates with headline measures
for the US, euro area and Japan (Figure 291).
While taking account of these differentials, in our view, it is still instructive to focus on core
inflation and, in particular, its relationship to measures of economic slack. Even though it can
be hard to predict commodity price movements, and exogenous assumptions may need to be
made with respect to government-driven price changes, by seeking to identify the relationship
between “core” inflation and economic slack we at least can have a basic building block for
linking projections of economic activity to longer-term inflation projections.
19
While we exclude imputed rent among components for the purposes of this analysis of US inflaton,
note that other research we have published suggests that overall rents are particularly sensitive to the
output gap (see “The output gap is likely smaller than many think” (P Newland) , Global Economics
Weekly 19 February. This note also explains the calculation of the BarCap production function based US
output gap
20
Overall, our “Underlying” US PCE deflator still covers 54% of US private consumption – this ratio has
been relatively stable since 1960, when it covered 48%. For the euro area, we use the ECB’s seasonally
adjusted series for core inflation excluding food, drinks, tobacco and energy, which we have adjusted
for particularly large changes in administered prices and value-added taxes (this series is chain-
weighted by construction and tends to correspond to the US core PCE deflator; it excludes imputed
rents). This is backcast pre-1990 by using the OECD’s estimate for the euro area CPI excluding food
and energy. For Japan, we have constructed a chain-weighted index based on the private consumption
deflators for durable and semi-durable goods, plus services. This is also backcast (before 1980) using
the OECD’s series for Japan’s CPI excluding food and energy. By using chain weighting (based on
nominal consumption weights), we try to avoid discontinuities associated with CPI re-basing.
Figure 291: Comparing headline CPI indices with core inflation measures
US
15 US PCE core annual inflation % y/y
-3
63 67 71 75 79 83 87 91 95 99 03 07 11
15 Euro area
Euro area core inflation (BarCap/ECB/OECD
12 measure, % y/y)
Euro area HICP (Datastream calculated pre
9 1990, % y/y)
Euro area HICP less BarCap core rate, 5 yr MA
6
-3
71 75 79 83 87 91 95 99 03 07 11
24 Japan
21
18 Japan BC core PCE annual inflation
15 (OECD core CPI pre 1980)
Japan CPI ex fresh food % y/y
12
Moreover, recently the “real time” estimates of output gaps have continued to be revised
significantly. For example, in Figure 292 we compare the IMF’s real time estimates of the
output gap of “advanced economies” with its latest estimates (all data taken from various
issues of the World Economic Outlook).
Figure 292: IMF estimates of the output gap for “advanced Figure 293: How the OECD’s estimates for Japan’s output
economies”: “real time” vs. latest available estimates gap for the years 1994-98 were revised over time
2 % 3 %
1 2 1997 OG
1
0
0
-1 1994 OG
-1
-2 -2
-3 1995 OG
-3 latest estimate (Oct 09 WEO)
-4 1998
-4 "real time" estimate, spring WEO for that year
-5 1996 OG
-5 Dec Dec Dec Dec Dec Dec Dec Dec
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 94 96 98 00 02 04 06 08
Source: IMF World Economic Outlook issues Source: OECD Economic Outlooks
Figure 294: Comparing IMF estimates of advanced economies’ output gap in real time with the latest IMF estimate
% of potential GDP average 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
21
This issue has been researched extensively by academics. See, for example, “The reliability of output gaps in real
time” (A Orphanides & Simon van Norden, 1999).
Euro area output gap (OECD series, using BarCap GDP projection)
6 US output gap (composite of BarCap production and survey based, plus OECD
series)
4
2
0
-2
-4
-6
62 65 68 71 74 77 80 83 86 89 92 95 98 01 04 07
Source: OECD, Haver Analytics, Barclays Capital
As can be seen, the IMF’s “real time” estimates of advance economies were persistently biased
downwards in relation to the current, retrospective perspective (see also Figure 293 for more
information behind Figure 291). Hence, if used in “real time” for economic projections, such
projections would risk generating inflation projections with a downward bias.
Meanwhile, as the experience of the 1970s suggests, real time estimates of output gaps are
particularly susceptible to retrospective revision when a major productivity shock is
experienced. This was also apparent from considering “real time” versus retrospective
estimates of Japan’s output gap during the 1990s. Figure 293 shows how the OECD’s
estimates of Japan’s output gap swung markedly as time went on; for example, in
December 1996, the OECD estimated that Japan’s output gap for that year was -3.8%, but
by its December 1999 Economic Outlook this had been revised to +0.9%).
Given the propensity for significant retrospective revision of output gap estimates, there can
be a tendency to downplay the output gap as a concept of use for practical policymaking.
However, in our view this ignores the significance of both retrospectively estimated output
gap measures, as well as survey-based measures (which are not revised), with respect to
the evolution of core inflation, particularly when, as at now, both approaches signal a large
negative output gap.
Figure 296: Comparing output gap measures with the change in the underlying consumer inflation rate
US
0
-2
-4
-6
-8
62 66 70 74 78 82 86 90 94 98 02 06 10
Underlying inflation change (4 qtr ma of 4 qtr Underlying inflation change (4 qtr ma of 4 qtr
change in annual % rate) change in annual % rate)
Output gap (OECD measure, 4qtr ma)
Output gap (OECD measure, 4qtr ma, RHS) 5
4 5 BarCap survey-based OG (4 qtr ma)
4 4
3
3 3
2 2 2
1 1 1
0 0
0 -1 -1
-1 -2 -2
-3 -3
-2
-4 -4
-3 -5 -5
71 75 79 83 87 91 95 99 03 07 11 71 75 79 83 87 91 95 99 03 07 11
Japan G3
Underlying inflation change (4 qtr ma of 4 qtr Underlying inflation change (4 qtr ma of 4 qtr
change in annual % rate) change in annual % rate)
BarCap survey based OG (4 qtr ma)
Output gap (OECD measure, 4qtr ma)
1.5 5
1 Output gap (composite measure 4 qtr ma, RHS) 2
4
1 0.5 1
3
0.5 2 0 0
1 -0.5 -1
0
0
-1 -2
-0.5 -1
-1.5 -3
-2
-1 -2 -4
-3
-1.5 -4 -2.5 -5
82 86 90 94 98 02 06 10 82 86 90 94 98 02 06 10
In our analysis we have found that the best single retrospective predictor of core inflation
series for the G3 countries is the OECD’s standard production function based output gap
estimate. That said, this measure is still prone to revision – indeed, the OECD undertook a
fundamental revision of its estimation procedure for potential GDP during the past year,
which has resulted in substantial differences from its previous estimates 22.
Another approach to calculating output gaps that is not prone to such major revisions is to
base the estimate upon survey data. This has the advantage of using data that are not
revised, thereby solving the problem of retrospective revisions. The drawback is that such
an approach is constrained by data limitations, particularly the difficulty of obtaining reliable
estimates of labour and capacity shortages in the service sector – our estimates tend to
reflect developments in just the manufacturing sector 23. Overall, our analysis suggests the
sensitivity of our underlying inflation measures to the survey-based output gap series has
been somewhat less than that of the OECD’s production based series, though this is offset
to a considerable extent by the more reliable “real-time” aspect of survey data.
Some results of our analysis can be seen in Figure 296, which shows the ongoing
significance of the relationship between the output gap and the change in underlying
inflation rate. It also reveals that the sensitivity of changes in core inflation with respect to
the output gap has changed over time. Often referred to as a “flatter Phillips curve” 24, the
US and euro area charts show that this elasticity since the mid-1980s has been much
weaker than during the 1970s and early 1980s.
Comparing the 1985-2009 episodes with that of 1972-1984, we see that the slopes of
the best fit lines for the change in the underlying inflation rate for the US and euro area
flattened and converged to a similar coefficient on the output gap of around 0.25.
However, the German slope was already quite flat during the earlier episode (at 0.34),
and consequently flattened much less, to 0.22.
This raises the question of why the German “Phillips curve” slope has historically been
flatter and more stable than that of the rest of the euro area and of the US. In our opinion,
this reflects in part the stability of German inflation expectations, which are derived from the
independence and credibility of the Deutsche Bundesbank. The flattening of the Phillips
curve in the US and much of Europe during the 1970s and 1980s is associated with a
significant decline in inflation rates, as central banks and governments gave much clearer
signals as to what they considered an acceptable inflation rate, and of their determination to
enforce this. The linkage between actual inflation and inflation expectations in two years’
time is apparent from the Livingstone published by the Philadelphia Fed (Figure 298).
22
These revisions affected the OECD’s estimates of the output gap for the euro area and Japan more than those of the
US. For example, in its December 2007 Economic Outlook, the OECD estimated that the 2006 euro area output gap
was -0.9%, whereas in its December 2009 Economic Outlook it pegged the 2006 output gap at +1.0%. Similarly, for
Japan, the OECD estimate of the 2006 output gap, as published in December 2007, was -0.2%, whereas in December
2009 the estimated was +1.7%. However, the December 2007 estimate of the US 2006 output gap was +0.7%,
compared with +1.1% in the latest (December 2009) estimate.
23
For the US, we use the ISM semi-annual capacity utilisation data, rather than the FRB series, because it is not
historically revised and is constructed on a similar basis to the European Commission and Bank of Japan Tankan
measures. For US labour shortages, we use the NFIB labour shortages diffusion balance, while the EC and BoJ
measures identify labour constraints within their quarterly surveys.
24
While the Phillips curve was originally specified as the relationship of the inflation rate and the unemployment rate,
we refer to it here to describe the relationship of the change in the inflation rate to the output gap.
Figure 297: Comparing the relationship between the change in underlying inflation rate and output gap
US Q1 72 to Q4 84 US Q1 85 to Q4 09
8 8
6 y = 0.7643x + 0.3342 6
change in inflation rate
0 0
-2 5 -2
-4 -4
-6 -6
-6 -3 0 3 6 -6 -3 0 3 6
output gap (composite measure) output gap (composite measure)
5 y = 0.6693x + 0.8564 5
4 2
R = 0.4154 4
change in inflation rate
change in inflation rate
3 3
y = 0.267x - 0.0414
2 2 2
R = 0.5281
1 1
0 0
-1 -1
-2 5 -2
-3 -3
-4 -4
-5 -5
-6 -3 0 3 6 -6 -3 0 3 6
output gap (OECD estimate) output gap (OECD estimate)
Germany Q1 72 to Q4 84 Germany Q1 85 to Q4 09
5 5
4 4
y = 0.3406x - 0.4304 y = 0.2219x - 0.0368
change in inflation rate
3 2 3
R = 0.5297 2
R = 0.2076
2 2
1 1
0 0
-1 -1
-2 -2
-3 -3
-4 -4
-5 -5
-6 -3 0 3 6 -6 -3 0 3 6
output gap (OECD estimate) output gap (OECD estimate)
Note: Inflation rates are BarCap underlying series except for Germany (OECD series for core CPI ex food and energy for west Germany from 1972 to 1993, then for
pan-Germany from 1994 onwards). Source: Haver Analytics, Thomson Datastream, Barclays Capital
As mean expectations for US inflation came down during the 1980s and into the 1990s, so
did the variation of individual forecasts. Figure 299 illustrates this point by showing the
standard deviation of forecasts for the US annual CPI inflation rate in two years’ time, as
contained in successive vintages of the December Livingstone survey.
For the euro area, we are not aware of an equivalent survey with a time history as long as
the Livingstone survey (the nearest equivalent is the ECB’s Survey of Professional
Forecasters, which only dates back to 1999). However, in Figure 300 we show OECD
projections for inflation in Germany and the euro area (again, the projections published in
successive December Economic Outlooks). We can see how the OECD’s projections for
future CPI inflation within the euro area countries excluding Germany declined significantly
from 1985 to 1999, to close to German levels.
Figure 298: US inflation expectations have moderated in the Figure 299: As US inflation expectations moderated, the
past decade deviation of expected outturns has narrowed
Source: Federal Reserve Bank of Phladelphia, Haver Analytics Note: This chart shows the deviation of individual responses for the projected US
CPI annual inflation rate two years ahead. Substantial outliers have been
excluded from the calculations. Source: FRB of Philadelphia (Livingstone survey)
Figure 300: Inflation forecasts two years ahead: euro area vs. Figure 301: Bundesbank inflation projection vs. German CPI
Germany outturns and OECD forecasts
Source: OECD (Economic Outlooks), ECB Source: Bundesbank Discussion Paper 12/2009, OECD, Haver Analytics
In our view, a significant part of the explanation for the flatter Phillips curve in Germany
during the 1970s and 1980s can be ascribed to the greater prominence in price setting paid
by inflation expectations, which reflects the credibility and independence of the
Bundesbank. Figure 301 shows how the Bundesbank began to lay out a desired inflation
estimate from 1976 onwards. Along with a pragmatic approach to money targeting and the
institutional framework for German wage setting, this is often considered to have helped
explain the smaller degree of second-round price increases in Germany during the 1970s
and 1980s compared with other economies 25.
Recent developments
As illustrated in Figure 302, if we consider developments since 1990, the relationship
between the output gap and underlying inflation still appears to be at work, suggesting that
core inflation rates are likely to continue to decline in response to the large amount of
economic slack. Moreover, a cross check of production function-based estimates of
inflation with “real time” survey measures signals that both are at very negative, suggesting
that the signs on the production-function based estimates are unlikely to be revised.
These charts illustrate that it is a pressing need for economies to recover sufficiently so that
their output gaps return to positive territory in a relatively short period of time. If this does
not happen, then the deeply negative output gaps are likely to continue to exert downward
pressure on underlying inflation rates. The experience of Japan – and, more recently, of
countries in Europe, such as Ireland – shows that it can be relatively straightforward to slip
into deflation. Moreover, Japan’s experience, which saw the consumer expenditure deflator
and GDP deflator hit record lows of close to -3% y/y in Q4 09, suggests that once deflation
occurs, it can be very difficult to extricate back into positive inflation, given the importance
of expectations.
Figure 302: Comparing the recent relationship of the output gap to the change in the underlying inflation rate
US Euro area
25
See “Opting out of the great inflation: German monetary policy after the breakdown of Bretton Woods” (A Beyer, V
Gaspar, C Gerberding and O Issing, Bundesbank Discussion Paper 12/2009).
Inflation has historically been a powerful force to achieve fiscal and real adjustments.
We estimate that having 5% inflation rather than 2% can dramatically accelerate the
decline in unemployment, as unemployment can fall by 1.5pp in one year solely due to this
higher inflation. On the fiscal front, inflation can achieve what no congress can, fast
reductions in fiscal deficits. As the majority of government expenditures are not indexed
to inflation and taxes rise one-to-one with inflation, we estimate that 3pp higher inflation
can reduce fiscal deficits by 1pp of GDP, while eluding the political hurdles that typically
prevent expenses from falling. Undeniably, as most emerging markets can testify, inflation
is an undercover fiscal reform.
Combating deflation and reducing real debt burdens are not, in our view, important
reasons to increase inflation or inflation targets. Quantitative easing (QE) has
shown that the zero interest rate bound does not mean central bankers are left
without ammunition against deflation, and the redistributive loss from higher
inflation would hit the balance sheets of ailing domestic financial sectors, the main
creditors of the G4 economies.
Central bankers, especially the Fed and the BoE, are likely to tolerate higher inflation in
coming years. But contrary to popular belief, we believe this would be a positive
development for the global economy.
-4
Jan-06 Jan-07 Jan-08 Jan-09 Jan-10
Note: Breakeven inflation based on inflation indexed bonds with around 5yrs maturity. Source: BarCap Live
Rudiger Dornbusch once said that Argentina needed an Austrian central banker to impose
credibility and reduce inflation. If alive, he would probably be saying today that G4 central bankers
would benefit from having an Argentine on their board. This illustrates the main point of this piece,
that the benefits from higher inflation in the context of a developed world with high
unemployment, inflexible institutions – except in the US – and high fiscal deficits are considerable.
Central bankers know this and they are likely to tolerate higher inflation in coming years rather than
change their inflation targets. Inflation acts as an undercover real and fiscal reform as inflation can
accelerate the fall in unemployment and the consolidation of fiscal accounts. Contrary to previous
times of high unemployment (eg, 80s or early 90s), today inflation rates are substantially lower
making the adjustment process harder and slower. The main contribution of this note is to quantify
the gains from moderately higher inflation. We do not attempt, however, to quantify the cost of
having temporarily high inflation, but if these costs are not highly variable with unemployment and
debt – as they are likely not to be – the time for high inflation is now.
A February 2010 IMF report by Blanchard, Dell’Ariccia and Mauro reignited a debate about
whether inflation targets should be higher. Contrary to the view spelled out in that work, we do
not think that higher inflation targets are warranted. The new blueprint for QE that the Fed and
BoE have effectively established is more than enough ammunition for central bankers to fight
deflation when hitting the zero-interest rate bound, in our view. QE can side-step the banking
system and provide direct credit support, and effectively reduce the yield curve in ways that
lowers the effective credit costs. But there are other reasons that provide a much larger benefit of
having higher inflation. A modest increase in inflation can produce large gains in terms of falls in
unemployment. For reasons that keep behavioral economists busy, people don’t accept nominal
wage cuts but are insensitive to inflation eating into their purchasing power. For this reason, at
times where unemployment is high – that is, at times where the real wages are too high –higher
inflation helps reduce real wages and, in turn, unemployment. Especially for countries/regions
with highly inflexible labor markets, inflation can be your biggest ally.
Moreover, inflation acts as an undercover fiscal reform. Since a large portion of government
expenditures are not indexed to inflation, inflation allows for a politically costly fiscal
consolidation to occur without the need for the political support needed to cut government
expenditure. We estimate that the gains from higher inflation could be as high as 1pp of GDP
per year of reduction in fiscal deficits purely due to a change in inflation from 2% to 5%.
Figure 304: Unemployment is close to previous highs… Figure 305: … but inflation is not nearly close to its previous
highs, making today’s situation more complicated to resolve
4 2
0
2
-2
0
-4
80 85 90 95 00 05 10
80 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10
Contrary to popular belief, the redistributive power of inflation from debtors to creditors is
not a large plus to G4 economies. The domestic financial market is the largest creditor of
debt instruments in most of these economies, and so the income loss from inflation to
creditors would be an additional hurdle in the rebuilding of their balance sheets. Moreover,
Europe and Japan are net foreign creditors, which mean that the income losses are not
borne by foreigners. We estimate that the gains from reducing the real burdens of debt are
small for all G4 countries economies.
But how much can inflation really help reduce unemployment? Figure 307 helps understand
what variables matter to understand the quantitative impact of higher inflation on lower
unemployment. At today’s real wages (eg, rw0) in the figure, the gap between labor demand
(how many people firms want to employ) and labor supply (how many people want to work)
is positive, which gives rise to the currently high levels of unemployment. Real wages do not
fall fast enough to reduce unemployment because nominal wages are ‘sticky’ downwards, and
with low rates of inflation the economy is left hanging in a high unemployment range for a
long time. Unanticipated Inflation can reduce real wages, (eg, to rw1), and concomitantly
reduce the levels of unemployment. Formally, the extent of the decline in unemployment from
a fall in real wages is given by the elasticity of labor supply minus the elasticity of demand.
Conservative estimates of both these elasticities suggest that for every 1 percent fall in real
wages achieved via a higher inflation rate, unemployment can fall by around 0.5%. This
Figure 306: Nominal wage growth doesn’t turn negative… Figure 307: … even though a fall in real wages can reduce
unemployment
Real wage: w/p Labor
Wages (wkly earnings) y/y
supply
10
unemployment
9
without inflation
8 rw
unemployment
7 with inflation
rw'
6
5
4
3
Labor
2 demand
1
0
65 68 71 74 77 80 83 86 89 92 95 98 01 04 07 Quantity demand of labor: L
implies that having 5% inflation in a given year can reduce unemployment by 1.5% more than
in a world with 2% inflation. Thus in two years, unemployment can fall by 3% only as a result
of having temporarily higher inflation for two years. It is likely that after some time the benefits
of having higher inflation would dissipate, as unions internalize that they need to bargain for
larger increases in nominal wages and the benefit of falling real wages gets eroded. But
without certainty about whether inflation will be higher in the future, and given the current
high unemployment rates, it is unlikely that wage negotiations will incorporate higher inflation
so rapidly. Inflation expectations have remained steady through this period and policymakers
know the power of inflation surprises at this juncture. If unemployment doesn’t fall in coming
years, the outlook for higher inflation expectations would probably increase as the
temptations will be substantial.
However, some government expenditures are indexed to inflation. This means that while tax
revenues grow at the rate of inflation, so do all the expenditures that are indexed to inflation.
While it is hard to generalize, most social security spending in G4 economies are adjusted
according to a cost of living price index. Figure 308 shows the share of spending for different
expenditure categories. Social security benefits are typically around one-third of all
government expenditures, encompassing on average around 14% of GDP. To understand the
impact that inflation has on fiscal savings we provide a simple illustration. Assume that, just as
Figure 308: Social security benefits are mostly indexed to Figure 309: Reduction in fiscal deficits due to inflation when
inflation nominal wages hit the zero bound
% Share of government expenditures by type in 2009 % GDP Reduction in government expenditures from a 3%
60 1.2 increase in inflation
Government wages
Socical security benefits
50 Other 1.0
40 0.8
30 0.6
1.1
49 52 1.0
46 0.8
20 0.4
38 35 0.7
33 33
27 28
10 21 23 0.2
16
0 0.0
US Euro area UK Japan US Euro area UK Japan
in the previous section, inflation runs at 5% rather than 2% for one year. Assume also that
social security benefits and taxes increase at 5% and that all non-indexed expenditures grow
at only 2% (like the typical inflation target). How large would the fiscal savings be in this case?
Figure 07 provides the answer for each country. Fiscal deficits would fall by between 0.7pp
of GDP (US) and 1.1pp of GDP (UK) solely from higher inflation – that is, ignoring any
explicit effort to reduce real wages or increase tax rates. Considering the magnitude of
today’s deficit in all G4 economies, inflation could provide a much needed acceleration in
fiscal consolidation without confronting political barriers.
Once again, this is not a permanent benefit from inflation. When budgetary discussions
start to internalize the higher inflation this would prevent government expenditures from
falling in real terms, and the benefits from inflation as an implicit fiscal reform disappear.
However, for some time, maybe a couple of years, above-target inflation could imply lower
real burdens for governments.
Figure 310: Debtors and creditors in the credit market Figure 311: Net credit assets by sector
Net credit liabitlies of the nonfinancial sector
Nonfinacial domestic sector Financial Sector RoW
Households Businesses General Govt
Composition of debt % GDP
US -181 156 41
US 65.0 75.9 55.4
Euro 46.8 67.2 73.2
EA -2 -187 189 UK 100.0 84.2 55.0
Japan 43.5 78.7 96.0*
UK -239 225 14 Average maturity of debt (in years)
US 28.1 10.2 4.3
Japan* -20 -218 213 Euro … 4.9 6.8
UK …. 5.3 13.5
-300 -200 -100 0 100 200 300 Japan … 4.2 6.3
Note: Shows only net credit asset positions. The non-financial domestic sector Note: Average maturity of household debt is for mortgages. *Japan general
comprises households, businesses and general government. *Japan general government net debt as reported by the OECD, all other net credit liabilities
government net debt from OECD. Source: OECD, Flow of Funds, Barclays Capital calculated from Flow of Funds data. Source: Flow of Funds, Barclays Capital
Second, not all debt is subject to fixed interest payments. After a bout of inflation, it is likely that
short-term debt would be rolled over at higher nominal interest rates, undoing any of the
benefits of higher inflation, as nominal interest rates are likely to incorporate the higher inflation
expectations. Figure 9 shows the average maturity and overall debt composition of each
component of domestic debt in G4 economies. Household debts in the US – mostly mortgages –
stand out as having long maturities (and mostly tied to fixed interest rates). This highlights the
benefits for this sector to increase inflation. The general government debt in the UK is also of
particularly long maturity, making the UK Treasury a primary beneficiary of potentially higher
inflation rates. However, in most countries the main creditors of debt instruments are the
financial sector and this implies that while inflation is probably a benefit for most, it is a drag on
financial institutions. In the context of the need for the financial sector to start lending and
improve their balance sheets, the real losses from inflation would not help this process.
How to quantify this potential benefit of inflation? Figure 312 shows a simple back-of-the-
envelope calculation. Assume that the propensities to consume of debtors are 1 – ie, out of an
increase of 1 dollar of income, they consume 1 dollar – while that of creditors is 0.5. (The gap
is large to highlight that the small results are particularly stark). Also assume that all housing
and equity assets increase in value one-for-one with inflation (ie, they are real assets), while
credit market assets are half fixed in nominal terms, and only half effectively indexed to
inflation. While of course this assumption is important to quantify the precise benefits of
reducing the real burden of debt, it is enough to illustrate how small these benefits are relative
to those previously quantified. The figure shows that the increase in consumption due to a
3pp higher inflation rate is around 0.06 percent of GDP, a negligible amount. This is essentially
the reason why inflating debts is not a big boost to demand.
Combating deflation
Based on a simple Taylor rules nominal interest rates should be between -3 to -5 percent in
several of the G4 countries. But the zero nominal interest rate bound prevented them from
being so low. Blanchard et al (IMF, 2010) highlighted that one of the one main implications
of this was the need for more reliance on fiscal policy and for larger deficits than would
have been the case absent the binding zero interest rate constraint. As a result, the IMF
Figure 312: Back of the envelope calculation – benefit of inflating away debt
% of GDP Japan UK EA US
(1) Net debt assets of domestic economy (excludes the foreign sector)
Nonfincial sector -218.3 -239.1 -187.2 -180.7
Financial sector 213.3 225.5 189.1 156.2
(2) Value of debt after 5% inflation (assuming 1/2 of assets grow 1-1 with inflation)
Nonfincial sector -215.1 -235.7 -184.5 -178.2
Financial sector 210.3 222.2 186.4 154.0
(3) Reduction in value of real debt
Nonfincial sector 3.1 3.4 2.7 2.6
Financial sector -3.0 -3.2 -2.7 -2.2
(4) Reduction in debt service (reduction in real debt*average interest rate)
Nonfincial sector 0.04 0.18 0.11 0.12
Financial sector -0.04 -0.17 -0.11 -0.10
(5) Estimated impact on consumption from reduction in debt servicing cost*
Nonfincial sector 0.04 0.18 0.11 0.12
Financial sector -0.02 -0.09 -0.06 -0.05
Net consumption gain 0.02 0.10 0.05 0.07
Note: In (4) the interest rate is used is the average on government bonds, corporate debt and mortgages weighted by
share of all nonfinancial debt. In (5) we use simple model where marginal propensity for net debtors (non-financial
sector) to consume is 1.0 and for net creditors (financial sector) is 0.5. Source: Barclays Capital
suggests: “It appears today that the world will likely avoid major deflation and thus avoid
the deadly interaction of larger and larger deflation, higher and higher real interest rates,
and a larger and larger output gap. But it is clear that the zero nominal interest rate bound
has proven costly. Higher average inflation and thus higher nominal interest rates to start
with, would have made it possible to cut interest rates more, thereby probably reducing the
drop in output and the deterioration of fiscal positions.”
We believe this conclusion assumes that short-term nominal interest rates are the only
instrument available for central bankers. As has been particularly clear in 2008-09, central
bankers have been able to influence other interest rates beyond the overnight policy rate.
Through purchases of assets of different maturity they have been able to affect the entire
yield curve without changing the overnight rates. Figure 313 shows how much the
relevant interest rates for the private sector have fallen since short-term rates have hit
their lows. While it is true that long-term rates have moved for reasons other than policy,
it is undeniable that QE has had an impact on the relevant funding rates for the private
sector. QE is enough ammunition for central bankers to fight deflation when hitting the
zero-interest rate bound. For this reason, we do not believe that it is warranted to modify
the inflation target to combat deflation. It is hard to quantify how much the option value
is of having more room to manouvre in terms of short-term rates to prevent deflation, but
we believe that given the role played by QE, this value has fallen considerably and may
now be negligible.
However, some of the benefits from above-target inflation that we highlighted in previous
sections could also be achieved with an immediate change in inflation targets. If central
bankers announced an immediate increase in targets and inflation moved beyond previous
targets, existing labor contracts and expenditure plans are likely not to incorporate the
higher inflation for some time, typically until the contracts are renegotiated. This could
imply some short-term benefits from the nominal stickiness of many of the existing real and
fiscal conditions, as argued in our previous discussion.
Figure 313: Change in average interest rates since last central bank rate cut
-0.2
-0.4
-0.6
-0.8
-1.0
-1.2
US Euro UK Japan
Source: Haver, Barclays Capital
APPENDICES
Treasury
United Australian
Inflation OBL€i Swedish Canadian
Kingdom OATi, OAT€i, Capital
Generic name Indexed BTP€i GGB€i Government Real Return JGBi
Index-Linked BTAN€i DBR€i Indexed
Securities, Index-Linked Bonds
Gilts Bonds
TIIS, TIPS
No bonds
28 16 10 3 7 2 5 5 3 16
Outstanding*
Market value
$563.235bn £211.211bn €152.425bn €30.137bn €91.758bn €12.636bn SEK238.12bn CAD45.576bn AUD11.767bn ¥6.424tn
outstanding bn*
Market value
outstanding $563.235bn $341.074bn $218.692bn $43.239bn $131.650bn $18.130bn $33.350bn $43.474bn $10.583bn $69.012bn
$bn*
First issue date January 97 March 81 September 98 March 06 September 03 March 03 April 94 December 91 July 85 March 04
French CPI Nationwide
CPI All urban ex-tobacco Euro HICP Euro HICP Euro HICP CPI All Items CPI General
Linking Index RPI CPI nsa All groups CPI
nsa Euro HICP ex- ex-tobacco ex-tobacco ex-tobacco nsa ex-Fresh
tobacco Food
FRCPXTOB
Linking Index
CPURNSA Index, CPTFEMU CPTFEMU CPTFEMU JCPNGENF
Bloomberg UKRPI Index SWCPI Index CACPI Index AUCPI Index
Index CPTFEMU Index Index Index Index
ticker
Index
2-3 months
8 months or
Indexation lag 2-3 months 2-3 months 2-3 months 2-3 months 2-3 months 2-3 months 2-3 months 6 months to 10th of
2-3months
month
3 with par Coupon and
Floor? Par floor No floor Par floor No floor Par floor Par floor floor, No floor principal par No floor
3 without floor
Coupon Annual or
Semi-annual Semi-annual Annual Annual Semi-annual Annual Semi-annual Quarterly Semi-annual
frequency zero coupon
Note: * At end of 2009
Source: Barclays Capital
Argentinean
NTN-Bs, NTN- South Africa
Generic name Udibonos Government BCU TES Galil, ILCPI TURKGB POLGB KTBi
Cs Index-Linked
Inflation-Linked
No. bonds
11 9 5 16 6 10 4 4 1 1
outstanding*
Market value
BRL253.176bn
outstanding MXN360.579 bn ARS57.540bn CLP4,412bn COP15,784bn ILS140bn ZAR139.613bn TRY37.483bn PLN10.113bn KRW1,954bn
(liquid NTN-Bs)
bn*
Market value
outstanding $134.383bn $27.619bn $15.129bn $8.699bn $7.726bn $38.149bn $18.960bn $25.006bn $3.534bn $1.678bn
$bn*
First issue date
in current May 00 May 96 December 03 September 02 October 02 June 06 March 00 February 07 September 03 February 07
format
Unidas de CER Consumer UF Consumer UVR Consumer Israel South Africa CPI
Linking Index IPCA, IGPM Turkish CPI Polish CPI Korean CPI
Inversion (UDI) Price Index Price Index Price Index CPI NSA
BZPIIPCA
Linking index
Index, ISCPINM POCPIYOY
Bloomberg MXUDI Index ARCPI Index CLUFUF Index COCPI Index SACPI Index TUCPI Index KOCPI Index
IBREIGPM Index Index
ticker
Index
Up to 1.5
Up to 4
th th months,
weeks, T-5, T-10 to 1 month to 9 1 month to 15
Indexation lag Up to 2 weeks adjusted on 3-4 months 2-3 months 2-3 months 2-3 months
includes ACERCER Index of month of month
inflation
forecast
release
Coupon and
principal par
Floor? No floor No floor No floor No floor No floor floor (Galils), Par floor Par floor Par floor No floor
No floor
(ILCPI)
Coupon Monthly or
Semi-annual Semi-annual Semi-annual Monthly Annual Semi-annual Semi-annual Annual Semi-annual
frequency semi-annual
Note: * At end of 2009
Source: Barclays Capital
Figure 320: US 10y TIPS real yield Figure 321: UK 10y real yield
5.0% 4.5%
US TIPS 10yr CM real yield UK 10yr CM real yield
4.5% 4.0%
4.0% 3.5%
3.5%
3.0%
3.0%
2.5%
2.5%
2.0%
2.0%
1.5%
1.5%
1.0% 1.0%
0.5% 0.5%
0.0% 0.0%
1997 1999 2001 2003 2005 2007 2009 1997 1999 2001 2003 2005 2007 2009
Figure 322: France OAT€i and OATi 10y real yields Figure 323: Germany DBR€i 10y real yield
3.5% OAT€i 10yr CM real yield 3.0% Germany 10yr CM real yield
OATi 10yr CM real yield
3.0% 2.5%
2.5%
2.0%
2.0%
1.5%
1.5%
1.0%
1.0%
0.5% 0.5%
0.0% 0.0%
2002 2003 2004 2005 2006 2007 2008 2009 2006 2007 2008 2009
Figure 324: Italy 10y BTP€i real yield Figure 325: Greece GGB€i 2025 real yield
4.0% 6
BTP€I 10yr CM real yield
GGB€I 2025
3.5%
5
3.0%
2.5% 4
2.0%
1.5% 3
1.0%
2
0.5%
0.0% 1
2004 2005 2006 2007 2008 2009 2010 2003 2004 2005 2006 2007 2008 2009
15 March 2010
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Figure 326: Australia 10y real yield Figure 327: Canada 2021 real yield
Figure 328: Japan 10y real yield Figure 329: Sweden 10y real yield
5.0% 5.0%
JGBi 10yr CM real yield Sweden 10yr CM real yield
4.5% 4.5%
4.0% 4.0%
3.5% 3.5%
3.0% 3.0%
2.5% 2.5%
2.0% 2.0%
1.5% 1.5%
1.0% 1.0%
0.5% 0.5%
0.0% 0.0%
2004 2005 2006 2007 2008 2009 1997 1999 2001 2003 2005 2007 2009
Figure 330: Israel 10y real yield Figure 331: South Africa 10y real yield
6.5% Israel 10yr CM real yield 4.5% South Africa 10yr CM real yield
6.0%
4.0%
5.5%
5.0% 3.5%
4.5%
3.0%
4.0%
3.5% 2.5%
3.0%
2.0%
2.5%
2.0% 1.5%
2004 2005 2006 2007 2008 2009 2010 2004 2005 2006 2007 2008 2009
15 March 2010
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Figure 332: Argentina 10y real yield Figure 333: Brazil 10y real yield
70% Argentina 10yr CM real yield 12% Brazil 10yr CM real yield
60% 11%
50% 10%
40% 9%
30% 8%
20% 7%
10% 6%
0% 5%
2004 2005 2006 2007 2008 2009 2010 2004 2005 2006 2007 2008 2009 2010
Figure 334: Chile 10y real yield Figure 335: Colombia 10y real yield
5.0% Chile 10yr CM real yield 10% Colombia 10yr CM real yield
9%
4.5%
8%
4.0%
7%
3.5% 6%
5%
3.0%
4%
2.5%
3%
2.0% 2%
2004 2005 2006 2007 2008 2009 2010 2004 2005 2006 2007 2008 2009 2010
Figure 336: Mexico 10y real yield Figure 337: Turkey 5y real yield
6.5% Mexico 10yr CM real yield 25% Turkey 5yr CM real yield
6.0%
20%
5.5%
5.0% 15%
4.5%
10%
4.0%
3.5%
5%
3.0%
2.5% 0%
2004 2005 2006 2007 2008 2009 2010 2007 2008 2009 2010
15 March 2010
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Figure 338: US 10y TIPS breakeven vs realised inflation Figure 339: UK 10y breakeven vs realised inflation
6% 6% UK 10yr CM breakeven
5% 5% UK RPI y/y
Figure 340: France 10y OATi breakeven vs realised inflation Figure 341: Euro €i 10y breakeven vs realised inflation
2% 2.0%
1.5%
1% 1.0%
0% 0.5%
0.0%
-1%
-0.5%
-2% -1.0%
2001 2002 2003 2004 2005 2006 2007 2008 2009 1999 2001 2003 2005 2007 2009
Figure 342: Canada 10y breakeven vs realised inflation Figure 343: Sweden 10y breakeven vs realised inflation
2% 2%
1% 1%
0% 0%
CANRR2021 breakeven
-1% -1%
Canada CPI y/y
-2% Rolling 3yr average inflation -2%
1998 2000 2002 2004 2006 2008 2010 1997 1999 2001 2003 2005 2007 2009
15 March 2010
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Barclays Capital
5 North Colonnade
London E14 4BB
Strategy
Alan James Tim Bond Michael Pond Khrishnamoorthy Sooben
Global Inflation-Linked Strategy Head of Global Asset Allocation US Fixed Income Strategy Inflation-Linked Strategy
+44 (0)20 7773 2238 +44 (0)20 7773 2242 +1 (212) 412 5051 +44 (0)20 7773 7514
alan.james@barcap.com tim.bond@barcap.com michael.pond@barcap.com khrishnamoorthy.sooben@
barcap.com
Stefan Liiceanu Chris Bettiss Chirag Mirani Henry Skeoch
Japanese Strategy Inflation-Linked Strategy US Fixed Income Strategy Inflation-Linked Strategy
+81 3-4530 1554 +44 (0)20 7773 0836 +1 212 412 6819 +44 (0)20 7773 7917
stefan.liiceanu@barcap.com chris.bettiss@barcap.com chirag.mirani@barcap.com henry.skeoch@barcap.com
Economics
Index Products
15 March 2010
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
Analyst Certification(s)
We, Alan James, Khrishnamoorthy Sooben, Michael Pond, Chris Bettiss, Stefan Liiceanu, Marcelo Salomon, Guilherme Loureiro, Roberto Melzi, Jimena
Zuniga, Guillermo Mondino, Sebastian Vargas, Arko Sen, Daniel Hewitt, Christian Keller, Jeff Gable, Bulent Badsha, Wai Ho Leong, Matthew Huang, Scott
Harman, Anand Venkataraman, Brian Upbin, Jenna Myers, Rahul Sharma, Michalis Christodoulou, Yuan Tian, Marcela Barreto Rivera, Jose Mazoy, Xiaonan
Jiang, Chirag Mirani, Henry Skeoch, Nick Verdi, Julian Callow, Christian Broda and Mimi Yang, hereby certify (1) that the views expressed in this research
report accurately reflect our personal views about any or all of the subject securities or issuers referred to in this research report and (2) no part of our
compensation was, is or will be directly or indirectly related to the specific recommendations or views expressed in this research report.
Important Disclosures
For current important disclosures regarding companies that are the subject of this research report, please send a written request to: Barclays Capital
Research Compliance, 745 Seventh Avenue, 17th Floor, New York, NY 10019 or refer to https://ecommerce.barcap.com/research/cgi-
bin/all/disclosuresSearch.pl or call 212-526-1072.
Barclays Capital does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that Barclays Capital
may have a conflict of interest that could affect the objectivity of this report. Any reference to Barclays Capital includes its affiliates. Barclays Capital and/or
an affiliate thereof (the "firm") regularly trades, generally deals as principal and generally provides liquidity (as market maker or otherwise) in the debt
securities that are the subject of this research report (and related derivatives thereof). The firm's proprietary trading accounts may have either a long and /
or short position in such securities and / or derivative instruments, which may pose a conflict with the interests of investing customers. Where permitted
and subject to appropriate information barrier restrictions, the firm's fixed income research analysts regularly interact with its trading desk personnel to
determine current prices of fixed income securities. The firm's fixed income research analyst(s) receive compensation based on various factors including,
but not limited to, the quality of their work, the overall performance of the firm (including the profitability of the investment banking department), the
profitability and revenues of the Fixed Income Division and the outstanding principal amount and trading value of, the profitability of, and the potential
interest of the firms investing clients in research with respect to, the asset class covered by the analyst. To the extent that any historical pricing information
was obtained from Barclays Capital trading desks, the firm makes no representation that it is accurate or complete. All levels, prices and spreads are
historical and do not represent current market levels, prices or spreads, some or all of which may have changed since the publication of this document.
Barclays Capital produces a variety of research products including, but not limited to, fundamental analysis, equity-linked analysis, quantitative analysis,
and trade ideas. Recommendations contained in one type of research product may differ from recommendations contained in other types of research
products, whether as a result of differing time horizons, methodologies, or otherwise.
15 March 2010
Barclays Capital | Global Inflation-Linked Products – A User’s Guide
This publication has been prepared by Barclays Capital, the investment banking division of Barclays Bank PLC, and/or one or more of its affiliates as
provided below. This publication is provided to you for information purposes only. Prices shown in this publication are indicative and Barclays Capital is
not offering to buy or sell or soliciting offers to buy or sell any financial instrument. Other than disclosures relating to Barclays Capital, the information
contained in this publication has been obtained from sources that Barclays Capital believes to be reliable, but Barclays Capital does not represent or
warrant that it is accurate or complete. The views in this publication are those of Barclays Capital and are subject to change, and Barclays Capital has no
obligation to update its opinions or the information in this publication. Barclays Capital and its affiliates and their respective officers, directors, partners and
employees, including persons involved in the preparation or issuance of this document, may from time to time act as manager, co-manager or
underwriter of a public offering or otherwise, in the capacity of principal or agent, deal in, hold or act as market-makers or advisors, brokers or commercial
and/or investment bankers in relation to the securities or related derivatives which are the subject of this publication.
The analyst recommendations in this report reflect solely and exclusively those of the author(s), and such opinions were prepared independently of any
other interests, including those of Barclays Capital and/or its affiliates.
Neither Barclays Capital, nor any affiliate, nor any of their respective officers, directors, partners, or employees accepts any liability whatsoever for any
direct or consequential loss arising from any use of this publication or its contents. The securities discussed in this publication may not be suitable for all
investors. Barclays Capital recommends that investors independently evaluate each issuer, security or instrument discussed in this publication and consult
any independent advisors they believe necessary. The value of and income from any investment may fluctuate from day to day as a result of changes in
relevant economic markets (including changes in market liquidity). The information in this publication is not intended to predict actual results, which may
differ substantially from those reflected. Past performance is not necessarily indicative of future results.
This communication is being made available in the UK and Europe to persons who are investment professionals as that term is defined in Article 19 of the
Financial Services and Markets Act 2000 (Financial Promotion Order) 2005. It is directed at, and therefore should only be relied upon by, persons who
have professional experience in matters relating to investments. The investments to which it relates are available only to such persons and will be entered
into only with such persons. Barclays Capital is authorized and regulated by the Financial Services Authority ('FSA') and member of the London Stock
Exchange.
Barclays Capital Inc., US registered broker/dealer and member of FINRA (www.finra.org), is distributing this material in the United States and, in
connection therewith accepts responsibility for its contents. Any U.S. person wishing to effect a transaction in any security discussed herein should do so
only by contacting a representative of Barclays Capital Inc. in the U.S. at 745 Seventh Avenue, New York, New York 10019.
Subject to the conditions of this publication as set out above, Absa Capital, the Investment Banking Division of Absa Bank Limited, an authorised financial
services provider (Registration No.: 1986/004794/06), is distributing this material in South Africa. Absa Bank Limited is regulated by the South African
Reserve Bank. This publication is not, nor is it intended to be, advice as defined and/or contemplated in the (South African) Financial Advisory and
Intermediary Services Act, 37 of 2002, or any other financial, investment, trading, tax, legal, accounting, retirement, actuarial or other professional advice
or service whatsoever. Any South African person or entity wishing to effect a transaction in any security discussed herein should do so only by contacting
a representative of Absa Capital in South Africa, 15 Alice Lane, Sandton, Johannesburg, Gauteng 2196. Absa Capital is an affiliate of Barclays Capital.
Non-U.S. persons should contact and execute transactions through a Barclays Bank PLC branch or affiliate in their home jurisdiction unless local
regulations permit otherwise.
In Japan, foreign exchange research reports are prepared and distributed by Barclays Bank PLC Tokyo Branch. Other research reports are distributed to
institutional investors in Japan by Barclays Capital Japan Limited. Barclays Capital Japan Limited is a joint-stock company incorporated in Japan with
registered office of 2-2-2, Otemachi, Chiyoda-ku, Tokyo 100-0004, Japan. It is a subsidiary of Barclays Bank PLC and a registered financial instruments
firm regulated by the Financial Services Agency of Japan. Registered Number: Kanto Zaimukyokucho (kinsho) No. 143.
Barclays Bank PLC Frankfurt Branch is distributing this material in Germany under the supervision of Bundesanstalt fuer Finanzdienstleistungsaufsicht
(BaFin). This material is distributed in Malaysia by Barclays Capital Markets Malaysia Sdn Bhd.
IRS Circular 230 Prepared Materials Disclaimer: Barclays Capital and its affiliates do not provide tax advice and nothing contained herein should be
construed to be tax advice. Please be advised that any discussion of U.S. tax matters contained herein (including any attachments) (i) is not intended or
written to be used, and cannot be used, by you for the purpose of avoiding U.S. tax-related penalties; and (ii) was written to support the promotion or
marketing of the transactions or other matters addressed herein. Accordingly, you should seek advice based on your particular circumstances from an
independent tax advisor.
© Copyright Barclays Bank PLC (2010). All rights reserved. No part of this publication may be reproduced in any manner without the prior written
permission of Barclays Capital or any of its affiliates. Barclays Bank PLC is registered in England No. 1026167. Registered office 1 Churchill Place, London,
E14 5HP. Additional information regarding this publication will be furnished upon request.
15 March 2010
GILUG Cover 2010.qxp 12/03/2010 13:42 Page 2