Professional Documents
Culture Documents
Current
Date
Timing of Settlements
Periodic Settlement Dates
M-T-M
Exposure
I
-T
Swap
Origination
Date
Source: Keith Brown and Donald Smith, "Default Risk and
Innovations in the Design of Interest Rate Swaps," Financial
Management, vol. 22, no. 2 (Autumn 1993):94-105.
N
Va = L (Fa,N - F-T,N+T) (NP)(l + rtf' ,
1=1
where the FO,N is the new rate; F-T,N+T is the original
rate for period - T to N + T; rt are the zero-coupon,
risk-adjusted discount rates appropriate for each of
the remaining N cash flows; and NP is the notional
__F_ir_m_A I - o ( - - - - - - ~ . I __F_ir_m_B __
rate established when the deal was originated, how-
ever, the swap will take on a positive value for one
of the counterparties. At any date, the value of the
swap is determined by the difference between the
new swap rate and the original swap rate times the
notional principal of the deal discounted back at,
presumably, a series of zero-coupon rates (Le., a dif-
ferent discount rates for each settlement period into
the future).
To see this in more detail, consider the swap
transaction illustrated in Exhibit 1. The top panel of
this display assumes that at Date - T, Firm A agreed
to pay cash flows based on a fixed rate of F-T,N+T to
Firm Bin exchange for cash flows that vary with the
floating rate index, 1. The maturity of the swap is
N + T periods. Consistent with the earlier discus-
sion, assume that F-T,N+T was chosen at Date -T so
that the initial value of the swap was zero. The lower
panel shows that if T periods have passed so that we
are now at Date 0, the value of this swap can be
measured bycomparing the fixed rate on a newswap
maturing at Date N with F-T,N+T. Letting FO,N repre-
sent this new rate, the economic value of the swap at
Date is given by:
Measuring Actual Exposure on aSwap
When originated, a par-value swap does not re-
quire that either counterparty make an initial pre-
mium payment. The swap therefore starts with a
market value of zero; in fact, the swap's fixed rate is
selected to ensure this is true. Once market condi-
tions start changing relative to the locked-in fixed
worry about what happens to the option holder once
he or she receives the premium. The possibility of
having one's counterparty fail to perform on the
agreement at a time when economic conditions have
moved in an adverse fashion is what creates default
exposure.
For any particular participant, default exposure
on a swap is predicated on two events. The first is a
specific movement in interest rates that creates eco-
nomic value for the swap, which would essentially
make it an asset from that person's standpoint. The
second is that the participant's counterparty must
actually default.
One way to measure the current exposure in a
swap transaction is to calculate the economic conse-
quence of having to replace a swap counterparty.
For example, suppose you did an interest-rate swap
in which you are paying a fixed rate of 6 percent to
your counterparty. Two years after the deal was
originated, market conditions change, and now the
fixed rate for a newswap witha comparable maturity
to the existing one is 8 percent. If the counterparty
defaulted on the agreement, you would have to start
paying cash flows based on 8 percent to replace what
was previously being paid at 6 percent. Conse-
quently, the extent of the exposure can be thought of
as the present value of what it would cost over the
remaining life of the agreement to replace a 6 percent
swap with an 8 percent swap.
Current swap exposure has two distinct aspects:
the actual exposure and the potential exposure. The
actual, or mark-to-market, exposure is the financial
consequence of changes in economic conditions be-
tween the swap origination date and the present.
This can always be measured with certainty. We
know, for example, how swap fixed rates have
changed for a five-year swap done two years ago.
The potential, or fractional, exposure is the potential
mark-to-market exposure that could occur as a result
of changes in future economic conditions. The po-
tential exposure will always have to be estimated,
because future rate movements are unknown. Of
course, this makes determining the cost of a potential
default more difficult. What if, for instance, the
counterparty in the last example waited for another
year and then defaulted when swap rates had risen
even higher?
21
principal on the swap. Notice that Va will be positive
when F-T,N+T FO,N; under these conditions, the swap
represents an asset to Firm A (the fixed-rate payer)
and a liability to Firm B.
This formulation is essentially a very straightfor-
ward bond-pricing problem. Using the earlier exam-
ple, if we did a five-year swap paying a fixed rate of
6 percent two years ago, and the new market rate for
what now is a three-year swap is 8 percent, the
economic value of the swap would be the difference
between the new8percent and the old 6percent rates
times the notional principal of the deal discounted
back to the present. This amount is actually the
present value of an annuity, and the calculation can
be done at any point based on current market condi-
tions. That is why calculating the mark-to-market
default exposure on a swap is easy if the counterpar-
ties can agree to such things as the appropriate dis-
count rates.
Notice, however, that the actual risk exposure of
either firm is only half of this economic exposure
calculation for the swap, because at any point in time,
only one party can be hurt by a default of the other.
Although the economic value of the swap can be
either positive or negative, it will only represent an
exposure to Firm A if a default by Firm Bcosts Firm
A something and vice versa. For instance, if rates go
from 6 percent to 4 percent, Firm A would like to
have Firm Bdefault, because it could replace Firm B
and pay a lower rate. Conversely, Firm B would
benefit if Firm A defaulted when the swap fixed rate
rose to 8 percent. Thus, the mark-to-market risk
exposure exists for the fixed-rate payer only if rates
rise and to the fixed-rate receiver if rates fall. When
swap rates rise, the fixed-rate payer of the existing
contract holds what amounts to an asset. In that case,
the fixed-rate receiver has the liability. That situation
is reversed when rates go down. More formally, the
actual risk exposure (RE) on the swap for both Firm
A and Firm Bat Date 0 can be expressed as:
REA,o = LI(Fo,N - F- T,N+T)(NP)(l + flr
l
if
FO,N> F-T,N+T, Oif otherwise;
and
REs,o = LI(Fo,N - F-T,N+T)(NP)(l + fl) -I if
FO,N < F-T,N+T, 0 if otherwise.
Once again, the important point here is that these
exposures depend on the joint occurrence of two
events: the default by the counterparty and an ad-
verse movement in interest rates.
Measuring the Potential Exposure
Calculating potential default exposure is more
problematic than measuring the actual exposure be-
22
cause it involves estimating how much worse the
interest rate situation could get. Most institutions
engaging in swap transactions have their own formu-
las for estimating potential exposure. Such estimates
should account for three general factors: the per-
ceived volatility of future rate movements; the time to
expiration of the agreement (the longer the remaining
maturity, the more volatility potential and the higher
the risk allowance one needs to make); and the rela-
tive creditworthiness of the counterparty. The period
over which this potential exposure must be estimated
is shown in the lower panel of Exhibit 1.
The potential default exposure is sometimes ap-
proximated as a fixed percentage of the contract
amount. For example, the allowance for potential
exposure might be in the 2-4 percent range for an
interest rate swap in which the principal is notional.
For a currency swap, the allowance might be 10-15
percent because principal is exchanged so the stake
is larger in the event of default. Even if more sophis-
ticated technologies are used, however, measuring
potential exposure is always guesswork.
Managing Swap Credit Risk
Several methods have been developed for reduc-
ing credit exposure on a swap. These are:
Collateral arrangements. These arrangements
are similar to margining in the futures market. One
of the great advantages of the OTC markets relative
to the exchange-traded markets is that they have not
historically required collateralization. Thus, users of
OTC derivatives have not had to monitor and man-
age the hedged position to the same extent as in the
futures market. That policy is changing, however,
with increased focus on the risk exposure of the
products, and collateralizing deals has become im-
portant. Certain swaps that have been transacted
recently had no collateral posted initially, but if some
catalyzing event occurs, such as a doubling of inter-
est rates or a downgrading of the creditworthiness of
a counterparty, then a clause in the contract would
be triggered requiring collateral to be posted.
Captive swap subsidiaries. A second method
for reducing credit exposure is to create special-pur-
pose vehicles for trading swaps within a trading
organization. The idea of captive swap subsidiaries
is to generate an entity that has a much higher credit
rating than the parent firm to act as a counterparty to
the transaction. Usually, these entities will have re-
strictions onwhomtheycan deal with, the creditwor-
thiness of the counterparties, whether transactions
must be matched, and so forth.
Enhanced netting agreements. The idea of en-
hanced netting agreements is to eliminate "cherry
picking." An example of the problem that cherry
Table 1. Assumed Pattern of Future Swap Rates
Basis Risk
Source: Keith Brown and Donald Smith, "Default Risk and
Innovations in the Design of Interest Rate Swaps," Financial
Management, vol. 22, no. 2 (Autumn 1993):94-105.
As defined above, basis risk exists whenever the
volatility of the market prices on a hedge position
and the underlying security do not move in perfect
concert with one another. To understand the basis
risk manifest in a swap-based financial strategy, one
must understand the pricing of swaps and the mar-
ket forces that cause these prices to change. In the
"plain vanilla" swaps we have been considering, the
swap spread is the main pricing variable. This sec-
tion describes the results of some ongoing empirical
research that Van Harlow, Don Smith and I have
done on how swap spreads have varied over time.
As illustrated in Exhibit 2, the fixed rate of inter-
9.00%
8.50
9.50
10.00
8.75
Prevailing Swap
Rate
Necessary Swap
Maturity (years)
--------
2.0
1.5
1.0
0.5
o
1
2
3
4
Settlement
Date
must pay an additional $69,049.
Where is the quid pro quo? The company will
immediately get into another swap for the remaining
three periods, payinga fixed rate of 8.5 percent rather
than 9 percent. The idea is that this additional
$69,049 payment is just the present value of what
they will be saving by paying a lower swap rate for
the remaining time to maturity. This technique elim-
inates the accumulated exposure for the coun-
terparty. In later periods, the rate movements will go
the other way, and the company will be receiving the
unwind payment (e.g., $93,301 at Date 2).
If the company had simply done a swap in which
it paid 9 percent for two years, its effective cost of
funding would have been 9 percent. With the mark-
to-market swap-based transaction, its internal rate of
return, or cost to fund, will also be 9 percent. It will
not know exactly what its cash flows are during the
period, however. This process is exactly like margin-
ing in the futures market except it is done every six
months rather than on a daily basis. The company
can still guarantee the 9 percent fixed rate that existed
on the original two-year swap, but its periodic cash
flows will be a little more uncertain. The benefit is
that it can reduce its credit risk by wiping out the
accumulated mark-to-market exposure every settle-
ment period.
picking creates can be seen by considering a firm that
does two swaps with the same counterparty-one to
pay fixed and one to receive fixed. When market
rates move subsequently, one of those swaps will be
a liability and one will be an asset. If the firm is in
financial distress, it will have the incentive to seek
relief from the swap that is a liability and protect the
one that is an asset. In a netting agreement, the party
cannot default on the swap that is a liability and force
someone to continue to pay onthe one that is an asset.
The economic values will net out on deals done
between the same two counterparties. Netting alone
often reduces exposure by as much as 50 percent in
doing multiple deals with counterparties.
Mark-ta-market swap agreements. A mark-to-
market swap arrangement would be similar to set-
ting up a margin account from which each party
would pay up its economic exposure as it occurs over
time. For example, two counterparties might agree
to make settlement payments on the swap every six
months for the next five years. As we have seen,
when market conditions change, actual exposure
builds up on the contract. One way to mitigate that
exposure is to settle in cash whatever one period's
worth of rate movements have been.
In a mark-to-market swap, on the first settlement
date, two things happen: The parties make the ex-
change of rates according to the original swap agree-
ment and they immediately unwind the existing
swap and cash out its market value according to the
formulas provided earlier. After unwinding the ini-
tial swap, the parties then enter into a new swap that
covers the remaining time to the original maturity
date at the prevailing new market fixed rate. This
procedure, which actually is executed with a single
agreement, is repeated every settlement period until
the swap matures.
To see the benefit of this arrangement, suppose
a company wants to raise $10 million for two years.
It issues floating-rate debt with semiannual coupons
indexed to LIBORand then swaps this two-year debt
into a fixed-rate obligation. An assumed pattern of
future swap rates is shown in Table 1. At the first
settlement date, the company, as the fixed-rate payer,
must make a $450,000effective payment on the swap;
that is, six months' worth of the 9 percent it obligated
itself to, times $10 million. Because rates have fallen
from 9 percent to 8.5 percent, the swap would repre-
sent a liability to the company and an asset to the
counterparty. The company must make an unwind
payment on the old deal, which will be equal to the
present value of the difference between 9 percent and
8.5 percent, times one-half the principal (because it is
a semiannual payment), times the notional principal
discounted back to the present. Thus, the company
23
Exhibit 2. Typical Plain Vanilla SwapTransaction
T+SS
Fixed-Payer on the Swap
Fixed-Receiver on the Swap
FirmA ""
FirmB
Index
Cash flows
(e.g., LIBOR)
Cash flows
on fixed-rate T+BS(A)
Index + CS(B) on floating-rat
debt
debt
, It
Fixed-Rate Bondholders
Floating-Rate Bondholders
e
Firm A's Net Cost of Funds
=T + BS(A) + Index - (T+SS)
=Index + BS(A) - SS
Firm B's Net Cost of Funds
= Index + CSB + T + SS - Index
=T + CS(B) + SS
Source: Keith Brown, W. Van Harlow III, and Donald Smith, "The Determinants of Interest Rate Swap
Spread," Boston University Working Paper (1992).
est on a domestic interest rate swap has two compo-
nents. The first is the yield to maturity for aT-bond
with the same maturity as the swap. In addition, the
market maker will add on some basis points, or risk
premium, which is referred to as the swap spread.
This spread is like a bond spread in the securitized
debt market or a credit spread in the bank debt
market. The swap spread is the critical variable in
determining the movement of swap prices, not the
Treasury rate, because that rate will be a constant
benchmark for rates determined in virtually every
financial market.
Figure 1 traces the historical pattern of the ten-
year swap spread. From the mid-1980s to mid-1991,
the spread ranged from about 40 basis points up to
about 110 basis points over the Treasury rate. If the
risk premium on a swap is compared with the risk
premiumon debt of comparable creditworthiness, as
in Figure 2, the picture is even more choppy. In fact,
the display shows that the two spreads are far from
Figure 1. Historical Pattern oftheTen-Year Swap
Spread
1.4,..--------------------,
1.2
1.0
~
:;; 0.8
2
JrO.6
0.4
0.2
OL.---'-------L----''----'-------L----''----'-----'
1/4/85 10/11 /857/18/86 4/24/87 1/29/88 1/4/88 8/11 /89 5/18/90 2/22/91
Source: Salomon Brothers.
24
matching.
That mismatch is the source of basis risk.
Whether swap prices go up or down is not as import-
ant as how they move relative to other economic
variables that they will be matched against in a
hedge. The risk premium, or spread component, is
not in perfect concert with the underlying security in
the bond market.
Figure 2. Historical Pattern ofthe Ten-Year
Bond-SpreadlSwap-Spread Differential
0.4 ,..-------------------,
0.2
~ 0 fJilt lltll:-i------
"Cl
2
0..--0.2
Ul
--0.4
--0.6 '--_-'--_-'-_---'---_-'-_--'-_---,-L_-----'_-----'
1/4/8510/11/857/18/864/27/87 1/29/88 11/4/88 8/11/895/18/90 2/22/91
Source: Salomon Brothers.
In our study, we tried to obtain a sense of
whether we can counter the effect of the mismatch
between the swap spread and bond yields. The ques-
tion is what explains movements in this critical com-
ponent of a swap's price? We developed a "perfect-
markets" model to see just what, in theory, should be
true about swap spreads' behavior. For this simple
model, we removed transaction and information
costs, credit risk differentials between the two coun-
terparties, and imbalances between the receive-fixed
and pay-fixed sides of the market. The index is three-
Interest rate swaps represent a legitimate innovation
in financial markets, but for all the benefits they
provide to their users, swap agreements also carry
certainrisks. In this presentation, I have explored the
nature of both the default and basis risk aspects of
swap contracting. What I hope that I have been able
to show is that these risks are real and, further, that
they are dynamic in nature. Consequently, the end-
users of swaps and other OTC derivative products
need to understand them for what they are. In par-
ticular, the ability to weigh the undeniable benefits
of these agreements against their sometimes hidden
costs is paramount to maintaining the financial
health of participating firms.
ity for the one- and three-year swaps was substan-
tially greater than for the longer periods, with R
2
values of about 60 percent as opposed to 5 percent.
Thus, the perfect markets model is less than a perfect
explanation for swap movements.
We refined this basic model by adding explana-
tory variables to account for several market "im-
perfections" such as dealer hedging costs, default
risk, and the supply of new debt. Of these additional
factors, the repurchase agreement rate, which served
as a proxy for the price that a swap dealer would have
to pay to hedge a swap position, provided the most
dramatic increase in explanatory power. In particu-
lar, the "repo" rate was significantly negatively cor-
related with swap spread movements, especially for
the longer term swaps. Further, this relationship
appeared to weaken over time. The credit risk vari-
able, which we could only measure to a limited ex-
tent, was also significantly related to swap spread
movements. In contrast, variations in the supply of
new corporate debt had virtually no impact on the
way swap prices were set.
The study yielded three primary conclusions.
First, the basis risk in using OTCderivatives depends
on several factors, with the maturity of the swap
having the biggest effect. Second, short-term swap
prices will be far more predictable than long-term
swap prices because short-term swaps are easier to
duplicate in exchange-traded futures markets. Fi-
nally, the basis risk in the swap product can be sub-
stantial, with about 20 or 30 percent of the volatility
in the swap spread left unexplained, at least by our
model.
Conclusion
EN(LIBOR)
SSN
EN(T-bill)
TN
EN(TED spread)
month LIBOR. Under these assumptions, the equi-
librium condition is that the pay-fixed and receive-
fixed sides of the swap pay the same expected
amount over the life of the agreement. That is,
EN(LIBOR) = TN + SSN =
EN (T-bill) + EN (TED Spread),
SSNt = Yo + Yl(ZNt - TNt) + yzTEDS7 + '!'Nt'
where
so that
SSN = [EN(T-bill) -TN] + EN(TED spread),
where
= Overall expectation of LIBOR,
= Swap spread,
= Expectation of future T-bill yields,
= Current Treasury yield, and
= Expectation of the difference
between Eurodollar and
T-bill yields.
Thus, the swap spread should be equal to the differ-
ence between the expectation of future T-bill yields
and the Treasury yield currently, as well as the ex-
pectation of the TED spread.
To test this formulation, we used the following
regression:
(Z - T) =Difference between zero-coupon and par
Treasury yields, and
TEDS* =Average difference between Eurodollar
and T-bill futures yields.
The variable (Z - n amounts to a measure of the _
slope of the Treasury yield curve. The two compo-
nents will be the same in a flat yield curve environ-
ment, and Z will be higher than Tin an upward-slop-
ing yield curve environment. The hypothesis would
be that we should get positive correlation between
the two independent variables and the swap spread.
More important, they should both be equal to 1 and
the intercept term should be equal to zero.
We used about eight years of data and investi-
gated five different swap maturities: one and three
years on the short end and five, seven, and ten years
on the long end. We also ran the regression for
different subperiods. A l t h o u g ~ the regression coef-
ficients on (Z - T) and TEDS were positive and
significantly different from zero, they were also sig-
nificantly different from 1. Further, the predictabil-
25
Question and Answer Session
Keith C. Brown, CFA
Question: Because most custom-
ers talk to a number of swap deal-
ers on all transactions, doesn't
competition among dealers ensure
that the discount rate used to es-
tablish the mark-to-market value
converges to the market discount
rate?
Brown: If potential swap partic-
ipants talk to multiple coun-
terparties before starting a deal,
they can generally count on re-
ceiving the most competitive
quote available in the market.
Once a deal has been originated
with a particular swap coun-
terparty in a particular bank, how-
ever, the counterparties are com-
mitted to one another. Getting
out of a deal after that point is not
a trivial matter. You can talk to
26
different market makers than the
one you originally used, but if
you end up with an offsetting
swap against the one you have,
economically it will be a wash,
but now you have exposures to
two different counterparties. In
the past, the International Swap
Dealers Association (ISDA) had
strict language governing the un-
wind. Rather than getting dis-
count rates, one approach would
be to poll different market mak-
ers to see what they are willing to
pay for the swap or what they
would demand as compensation.
This approach would then throw
out the high and the low quotes
and average the rest. In this
sense, then, the discount rate is re-
ally set by a panel as opposed to
just picking a rate.
Question: With many swap pro-
viders and ISDA provisions that
allow assignment, isn't it a bit
overly dramatic to say one's only
choice to unwind is with the origi-
nal counterparty?
Brown: Of course, but the hy-
perbole is valuable if it serves the
purpose of underscoring the fact
that deals negotiated in the aTe
market are not nearly as simple to
unwind as they are in exchange-
traded markets. Although assign-
ment provisions make liquidating
an existing swap position some-
what easier, it is still more diffi-
cult and time consuming than,
say, reversing out of a future con-
tract at a price determined by a
central market mechanism.
market is frequently a step ahead of the guidelines.
Liquidity. If the liquidity is better, a manager
will consider using a swap. Chances are, however,
that liquidity will not be better in the swap market,
because swaps are typically less liquid than other
instruments.
Convenience andflexibility. In some cases, es-
pecially in index investing, swaps can simplify port-
folio management dramatically.
When the market began in 1982 or 1983, swaps were
used as an asset/liability management tool. In par-
ticular, savings and loans were using interest rate
swaps because they had short-termliabilities (depos-
its) and long-term assets (mortgage securities) that
they needed to match more closely. In the late 19805,
insurance companies also started to use them, and
now money managers have embraced swaps. Mu-
tual funds and other money managers use the vast
array of swaps to perform all kinds of asset transfor-
mations and transactions that they might not be able
to do without swaps.
As money managers have progressed from
thinking in terms of individual securities and diver-
sification to viewing portfolios as a whole, they have
moved toward transacting entire portfolios. This
approach is now also embodied in the swap market.
You can swap against an entire portfolio of equities
or an entire portfolio of, say, the Lehman Brothers
Government/Corporate (LBG/C) Index.
Pricing in the swap market changed markedly in
the mid-1980s, when a zero-coupon trader at
Salomon moved to the swaps desk. This trader
looked at swaps just as he looked at zeros-as a series
Using Swaps in the Fixed-Income Portfolio
Robert W. Kopprasch, CFA
Senior Vice President
Alliance Capital Management
During the past ten years, swaps have proliferated and evolved into versatile tools for
fixed-income management. Unique forms have developed to serve a variety of specific
purposes. Although some forms are complex, in general, swaps simplify money mana-
gement and entail little credit risk.
This presentation describes how swaps are used in
fixed-income portfolios. In this new market, how-
ever, the term fixed-income is not being used in the
traditional sense, because cash flows are no longer
fixed. You can swap between sectors-from equity
to fixed-income, for example-you can change cur-
rencies, or payments can be kept in the same cur-
rency but can be based on interest rates in different
currencies.
To most people, the reason swaps are attractive
is not readily apparent. The first time someone de- ---------------------
scribed an interest rate swap to me was in 1983. We Evolution of Fixed-Income Swaps
were told that a client could finance with fixed-rate
debt but have someone else make the fixed-rate pay-
ments and then the client would make floating-rate
payments. I asked, "Why would anyone want to do
that?" Now, $4 trillion later, I concede that it might
be a viable market.
Fixed-income swaps are attractive to money
managers for several reasons:
Returns. Swaps can be designed to provide
better returns than other alternatives.
Unique fit in the portfolio. Swaps can be de-
signed to fit the unique needs of a particular portfo-
lio.
Inability to buy the underlying investment be-
cause of portfolio guidelines. If investment constraints
prevent a manager from making a particular invest-
ment, the manager may be able to enter a swap that
embodies some of the desired characteristicsthat are
not otherwise available. For example, a consh'aint
might forbid the use of futures, but exactly the same
end can be accomplished with a swap. Also, if the
use of leverage is not permitted, swaps can provide
leverage as quickly as borrowing money can. Con-
straints are not consistent in portfolios, but the swap
27
of cash flows. As far as he was concerned, everycash
flow was a zero-coupon bond, so he began pricing
swaps the same way he had always priced zeros.
This outlook changed the investment industry's per-
spective on the swap market. Now, everything is
priced off the term structure. If you want a cash flow
in 1997, that is no problem; it can be priced as the
differences between a series of other cash flows.
Swaps are an integral part of the larger market
containing cash investments, futures, and options.
Swaps also connect disparate markets, allowing in-
vestment managers to go, for example, from a fixed-
income portfolio to an equity return. Swaps can
cross the Atlantic or Pacific and put international
returns into a portfolio.
Types of Swaps
A number of types of swaps are possible. One of the
most basic structures is an interest rate swap. The
structures of two types of interest rate swaps are
shown in Exhibit 1. In the first, the money manager
is paying a fixed-rate and receivingfloating-rate pay-
ments. The fixed payment goes to the swap coun-
terparty, typically a dealer, and the floating payment
comes back to the manager. In the cash market, the
manager might own a corporate bond or an asset-
backed security, for example, and receive a fixed-rate
coupon. By paying out a fixed rate on the swap and
receiving LIBOR, the manager converts the fixed-rate
security into a floating-rate position. Assuming the
assets are appropriately matched, the manager now
has sensitivity to LIBOR and not to fixed rates. The
fixed-rate security has effectively been turned into a
floating-rate security with a significantly lower du-
ration. In the second type of interest rate swap, the
manager converts a floating-rate security into a
fixed-rate position.
Swaps are also used as pure volatility instru-
ments that do not have to be linked to specific secu-
rities. You do not have to be hedging to enter a swap.
The obligation to receive fixed and pay floating re-
turns adds volatility to a portfolio. If the value of the
swap increases and adds value to the portfolio as
rates go down, the swap has added duration. This
positionis virtuallyidentical tobuyinga government
bond and financing 100 percent of the cost.
Inaddition to the basic interest rate swap, several
types of complex swaps exist. I will describe index,
constant-maturity, index-amortization, differential,
and arrears swaps. Other types of swaps include
leveraged, zero-coupon, step-up coupon, and real
estate swaps.
Index Swaps
Index swaps allow money managers or firms to
compete in markets in which they do not have long
track records and with which they are not totally
familiar. You may manage mortgage-backed securi-
ties, for example, but want equity market exposure.
You can get the equity exposure by entering a swap
in which you payout LIBOR and receive equity
index returns. For example, you can put your money
in short-term investments, receive LIBOR or LIBOR
plus, payout LIBOR on the swap, and receive the
total return on the S&P 500 Index every quarter
(which may be more or less than LIBOR). With this
type of swap, managers tum their apparent cash
investments into S&P 500 investments. Their invest-
ment market exposure is now equity, regardless of
what their actual securities are, because they are
receiving equity returns.
Why would anyone want to give you the S&P
500 and get LIBOR back? Considering the equity
market over time versus LIBOR, such a trade does
not sound like a good one. It is attractive ina number
of situations, however. A securities dealer who is
holding a large portfolio of equities, for example,
receives the returns on whatever weighted portfolio
is held and finances it with short-term borrowed
money, which usually tracks LIBOR closely. This
dealer is a natural counterparty. The dealer can re-
Exhibit 1. Basic Interest Rate Swap Structures
Fixed
Fixed
Swap Counterparty Money Manager Operant Investment
.....
Floating
Floating
Swap Counterparty Money Manager
Floating
Operant Investment .....
~
Fixed
Source: Alliance Capital Management.
28
ceive S&P 500 returns on the portfolio, pay them out
on the swap, and then receive LIBORback to finance
his or her position. In this case, the swap is an
asset/liability management tool for that dealer, not
a market bet that the S&P 500 will do worse than
LIBOR. The dealer is perfectly matched-long eq-
uity in the cash market, short in the swap market-
paying LIBORon his financing and receiving LIBOR
on the swap.
Normally on this type of index swap, the no-
tional face amount of the portfolio is adjusted as if
the underlying equity portfolio were held. For exam-
ple, if you have $100 millionin the S&P500 Index and
it goes up 20 percent in a quarter, the portfolio is
worth $120 million at the end of the quarter. The
swap notional face amount would then change from
$100 million to $120 million. If the S&P does not go
up, the dollar returns on the $100 million swap
would not equal the dollar returns on an actual port-
folio that is as low as $120 million.
Index swaps can become quite complicated, as
Exhibit 2 shows. Here, the money manager holds a
mortgage portfolio and receives whatever returns
that portfolio generates. Suppose the manager, by
structuring the duration correctly, earns the mort-
gage-backed security (MBS) index plus 100.
Exhibit 2. Index Swaps
This type of approach can also be used to obtain
equity exposure. Recently, for example, a pension
plan had real estate assets it did not want to sell in
the current market but it no longer wanted the real
estate exposure. The board wanted togo into equi-
ties, so it did a swap in which the plan pays the
returns onthe Russell National Council of Real Estate
Fiduciaries Property Index and receives LIBOR.
Then, it did another swap in which it pays LIBOR
and receives the returns on the S&P 500 (or some
other equity index of its choice). Thus, it was able to
swap the returns on its real estate portfolio for the
returns on an equity portfolio, transferring that real
estate for a time into a stock market investment.
Alliance has a number of investors in short-du-
ration mortgage portfolios (that have LIBORplus 100
as their target) who swap LIBOR for the S&P 500
return. They attempt to capture the incremental re-
turn on the mortgage side, thereby beating LIBORby
100. Meanwhile, they get equity exposure with an
index swap that gives them the S&P 500 returns. In
these "passive" portfolios, swaps provide the basic
index returns and the enhanced-cash portfolio pro-
duces LIBOR or a spread over LIBOR.
You can also use swaps in the international mar-
ket, in which actual securities transactions can be
LBG/C
Counterparty Returns Money Manager
MBS
Mortgage Portfolio
~
~
Returns
LIBOR
LIBOR
t ~
MBSlndex
Returns
Counterparty
Source: Alliance Capital Management.
Look at this transaction as two index swaps. The
money manager pays out the MBSindex and receives
LIBOR on one swap. At the same time, the money
manager pays out LIBOR and receives the LBG/C
Index. Assuming no spreads other than earning the
100 basis points over the MBS index, the manager
now has a portfolio that receives the LBG/C Index
plus the extra 100 basis points earned agamst the
MBS index. The manager's mortgage expertise is
transferred into a sector in which he or she does not
specialize-the LBG/ CIndex. When the other sector
is the S&P 500, we sometimes call this double swap
the S&P 600 strategy, because it produces the S&P
500 plus 100 if done correctly.
very expensive. Instead of trading directly in those
markets, you can enter into a swap with a dealer in
NewYork or anywhere and receive the returns of the
foreign market without dealing with all the im-
plications of the underlying transactions.
Constant-Maturity Swaps
Constant-maturity swaps, sometimes known as
yield curve agreements, have become very popular.
They are actually "floating-for-floating" swaps in
which the rate paid and the rate received both float,
based on two different points in the yield curve. An
example of such a swap is given in Exhibit 3. In this
29
-----------------------------------------------------
Exhibit 3. TheConstant-MaturitySwap
Index-Amortization Swaps
Negative convexity could not be confined to the
callable corporate market and to mortgage markets,
and now it has come to the swap market. An index-
amortizationswap is one in which you receive a fixed
rate and pay LIBOR, but the amount on which you
receive and pay (the notional face amount) amortizes
according to a schedule based on how some index,
typically LIBOR, changes over time. If LIBOR goes
up, a five-year swap might extend to seven years,
because its amortization schedule slows down. If
rates decline instead, the notional principal amount
swap, on each payment date, the money manager
pays the counterparty the then-current two-year
Treasury rate plus some spread and the counterparty
pays the money manager the then-current ten-year
Treasury rate. As an example, in the first two weeks
of February 1993, the Federal Home Loan Bank
(FHLB) sold $1.5 billion worth of floating-rate notes
with coupons based on the ten-year constant-matu-
rity rate. (One issue paid the ten-year constant-ma-
turity rate minus 200 basis points, while another paid
half of the ten-year rate plus 150.) Using this type of
swap, FHLB then swapped into LIBOR-based fund-
ing. Its net financing cost was LIBOR less 41 basis
points by performing this swap. Considering the
implications of the forward curve and so on, would
the FHLBprefer to issue a bond that paid the floating
rate, ten-year Treasury less 200, or would it prefer the
deal it did the following week, which paid half of the
ten-year Treasury plus ISO? Actually, the FHLB did
not care, because it swapped its bond payments into
LIBOR using this swap.
This type of trade puzzles many people. How
can FHLB perform this convoluted trade and net
such good financing when the market will not pro-
vide such attractive financing to begin with? Some-
thing must be left on the table somewhere by some-
one in this transaction, but the FHLB Board is walk-
ing away happy, the swap dealer is walking away
happy, and investors like the bonds they bought.
There must be some slippage somewhere, or else
these combinations of structures allow investors to
achieve positions that they might not be able to oth-
erwise.
Source: Alliance Capital Management.
may payoff earlier, and the swap might fully amor-
tize after two years instead of five years.
This type of swap sounds like a mortgage secu-
rity, but the advantage over mortgage securities is
that this swap is a pure interest rate play. A mort-
gage security has prepayment risk, and although
prepayments are related to interest rates, they are not
determined uniquely by interest rates. Home prices,
general economic activity, lending practices, con-
sumer confidence, and many other factors also affect
prepayment rates. In a LIBOR-based index-amorti-
zation swap, LIBOR alone determines prepayment
(amortization). Investors who decide to hedge this
risk can easily do so with Eurodollar futures.
The swap usually has a nominal fixed maturity,
although the actual payment stream may be longer
or shorter. This swap has no long tail of cash flows
like that of a mortgage security. The swap also has a
lockout period, so you do not have to worry about it
being called before that period. If you think of this
as a synthetic mortgage security or synthetic collat-
eralized mortgage obligation, you have the comfort
of knowing that you have a perfect prepayment
model, because the amortization schedule is set up
directly by LIBOR.
Differential Swaps
Differential, or "diff," swaps are one way to play
foreign currency rates without dealing in foreign
currency securities. Suppose you think French inter-
est rates will come down. You could pay six-month
PIBOR (Paris Interbank Offered Rate) minus 300
basis points in dollars for a two-year swap. With
PIBOR at 8 1;S this morning, you could begin with an
initial payment of 5 liS and receive six-month LIBOR,
which was 3 5/16 this morning. If PIBOR came down
enough and LIBOR went up, you would make
money on that swap as the relative interest rates, or
the differential in rates, change. As another example,
you could have paid deutsche mark LIBOR in dol-
lars. If the DM LIBOR rate in Germany was 8 per-
cent, you would pay 8 percent in dollars and receive
U.s. dollar LIBOR plus 144 for three years.
Many investors do not like the idea of paying
six-month PIBOR minus 300 and receiving six-
month LIBOR, because in the beginning, they are
paying out more money than they are receiving, and
nobody wants to get into the penalty box right away.
To solve this "negative carry" problem, the swap
could be restructured so that the swapper receives
11.9 percent minus PIBOR and pays LIBOR. In this
swap, the coupon received increases as PIBOR de-
clines. If PIBOR declines faster than LIBOR rises, the
cash flow on the swap increases. This swap would
use the same diff swap already discussed but then
Money Manager
(Both float)
2-Year Treasury
plus Spread
,
10-Year Treasury
Counterparty
30
add two fixed-for-floating U.s.-based swaps in the
opposite direction. Such a swap creates a completely
different position in a portfolio from the first diff
swap; it adds a lot of duration. The two fixed-for-
floating swaps that were added to the original diff
swap provide the long positive duration. These two
related, but completely different, positions provide
some evidence of the flexibility in structuring that
swaps provide.
Arrears Swaps
Currently, the yield curve is upward sloping,
and the implied forward rates are all high. If you
believe that the forward rates are wrong, that is,
future rates will not be as high as the forwards today,
you can do an arrears swap. In a LIBOR arrears
swap, you would receive today's three-month
LIBOR plus about 20 basis points (for a quarterly
swap) and you would pay the three-month LIBOR
that existed at the end of the quarter. That is, you
would receive today's LIBOR three months later and
pay the counterparty based on the LIBOR that pre-
vailed then. If rates move up less than 20 basis points
a quarter, the swap will generate positive cash flow
all the time. This swap allows the profit and the risk
to be constrained to changes in LIBOR over a three-
month period.
Swap Credit Risk
Credit risk in a swap is the risk that the counterparty
will default when the swap has positive value. In
swaps-both theoretically and empirically-eredit
risk is quite small and can be further reduced by
many methods. To reduce credit risk, parties can put
up collateral as a good faith deposit or they can mark
it to market. Several brokerage firms are now setting
up special-purpose vehicles with sufficient capital to
be Aaa rated to perform swaps on favorable terms.
Swap credit risk depends upon a number of
factors, including the type of swap, its maturity,
whether "negative carry" is built in, and so forth.
Naturally, the credit quality of the counterparty is
important as well.
Most people do not think that the counterparty
they are considering for a swap will go bankrupt
tomorrow; otherwise, they would not consider doing
the swap. The expected probability of default by a
counterparty is therefore low at the time a swap is
transacted; you can easily get some sense of the
counterparty's financial stability at that time. You
may not be so sure, however, that the counterparty
will be free of problems in three years. The probabil-
ity of default increases over time.
Another aspect of credit risk that is also affected
by the type of swap involves the volatility of rates
and how a particular rate move affects the swap's
value. (Of course, higher levels of volatility result in
potentially larger rate moves.) The impact of a given
rate move depends upon the type of swap and its
remaining maturity. In general, at least for fixed-for-
floating interest rate swaps, a rate move in one direc-
tion will cause the swap's value to become positive
and a move in the other direction will cause the value
to become negative, as shown in Exhibit 4.
Exhibit 4. Exposure to Rate Movement, One
Standard Deviation
Time
-- Favorable
- - - Unfavorable
Source: Alliance Capital Management.
Notice in Exhibit 5 how maturity affects the
swap's value. With a three-year swap, the credit risk
is small; however, even if rates move substantiallyby
the end of two years, only one year of interest rate
payments is left in the agreement and the swap value
is the present value of the difference in the payments.
For a seven-year swap, the payment period is longer,
and at the end of two years, five years of payments
remain and the value will reflect the present value of
five years of payment differences. Because the swap
involves a notional amount, not actual principal,
Exhibit 5. Exposure to Price Change on Interest
Rate Swap, One Standard Deviation
~ -
----
-- 7-Year Swap
- - - 3-Year Swap
Source: Alliance Capital Management.
31
there is virtually no credit risk at the end.
Other types of swaps have different levels and
timing of credit risk. For example, an index swap has
credit risk only during one period because the parties
settle up entirely every period: one party pays the
return on the index, the other party pays the return
on LIBOR, and they are "flat" again. The only differ-
ence is that they might have started out with a $100
million swap and now have a $120 million swap. In
general, you do not have more than one period's
worth of credit risk. At the opposite extreme, credit
risk for a currency swap probably reaches its maxi-
mum amount at the end of the swap period, when a
large payment, based on changes in value of the
entire notional amount, must be made.
Conclusion
The swap market has changed dramatically in the
past ten years. The fact that some people of 40-plus-
years are talking about swaps today shows that it has
become a mature market. (But those 40 year olds
started in swaps when they were young, ten years
ago.)
32
Swaps are widely used. New York City, for
example, is asking New York State for more author-
ity to use swaps in its financing. RJR Nabisco Hold-
ing has $3.6 billion in outstanding swaps. The FHLB
examples earlier show how many new issues are
swap related.
For money managers, the swap market presents
a tremendous opportunity-and potential problems
as clients and dealers become competitors. Clients
do not need managers for S&P500exposure; they can
achieve it themselves with an index swap. One ad-
vantage of the index swap market for managers is
that it eliminates turnover and rebalancing. Manag-
ers no longer have to do all the chores required to
manage an index fund. They simply get the returns
on the S&P 500 and devote their attention to earning
incremental return on the short-term portfolio.
But the opportunities are clearly available, from
tweaking return patterns to completely unhinging
the returns on a portfolio from the actual underlying
assets. As the market matures further and additional
swap types are developed, it will be important to
exploit these opportunities.
Question and Answer session
Robert W. Kopprasch, CFA
Question: Please provide addi-
tional detail on the FHLB con-
stant-maturity swap. Why do
you believe there is some slip-
page in the transaction?
Kopprasch: If the FHLB pays
the investor the ten-year constant-
maturity Treasury (CMT) rate
less 200, and we know its net cost
was 40 basis points under LIBOR,
it must have been able to do this
swap at essentially LIBOR as op-
posed to receiving the ten-year
CMT less 160. The 40 basis points
come from the fact that investors
are willing to buy the bond at a
coupon that is 40 basis points
lower than the swap market al-
lows. These investors probably
cannot do such swaps, because if
they were able to, they would in-
vest in a Euro-floating-rate note
and do the swap. The 40-basis-
point differential is the slippage.
In a bond, investors are willing to
take a lower return than they
could get in a swap, perhaps be-
cause they do not know a market
for that type of swap exists or per-
haps they are not permitted to
enter into that kind of swap. So
they do the next best thing, which
is 40 basis points less return.
One thing to keep in mind
with all these issues is that there
are no free lunches. These pat-
terns and structures exist because
one party is selling an option he
or she does not value highly (the
seller may not realize the extent
of value that is there) or is mak-
ing a bet on the forward curve.
The forward curve may be an in-
accurate forecaster of rates, but it
provides a mechanism by which
the parties to these contracts can
hedge their exposure to interest
rate risk. By engaging in a type
of trade that pays an above-mar-
ket rate today, investors are in ef-
fect forecasting that the forward
curve is. wrong, that rates will be
stronger in the future or weaker
in the future than the forward
curve suggests. Depending on
the structure of the instrument,
they will pay for the higher rate
today with a lower rate in the fu-
ture if their forecasts of future
rates are incorrect. When consid-
ering these transactions, you
must forecast rates and then de-
termine how a particular struc-
ture fits with that forecast; if your
forecast is correct, you will have
an opportunity to make more
money than you otherwise would.
Gastineau: Many people do not
realize exactly what they are
doing in some of these swaps. If
they did, or if they had invest-
ment flexibility, they would see
that they could use any number
of strategies in the futures market
to take advantage of the differ-
ence between their forward-rate
forecasts and the market's for-
ward-rate forecasts. Trying to
capture a higher return by trad-
ing futures sounds speculative,
but this transaction is not really a
hedge, just a swap, and it looks
like it adds value. People look at
it two different ways, when in
fact, it is the same transaction.
Kopprasch: I talked to an ac-
count once who decided to leave
an investment manager even
though the manager had been
earning 100 basis points over the
index. The account left to go
with a swap that guaranteed 75
basis points over the index
(rather than hoping for 100 basis
points from the previous man-
ager). The new manager could
make the guarantee by doing a
swap against the same sector and
putting the cash in a Euro-float-
ing-rate note. The client ignored
the fact that now the firm had
two credit risks (two swaps),
whereas previously, the portfolio
was a mortgage portfolio that
was all agency securities. Now,
they have a swap counterparty
who is a dealer and a floating-
rate note well below Aaa. It was
not a free lunch; clearly, the credit
characteristics changed.
Question: How is this /I dis-
guised leveraging" we are calling
swaps any different from specula-
tion with zero margin?
Kopprasch: Swaps are no dif-
ferent from futures except futures
involve a small amount of mar-
gin. As with futures, swaps
allow you to add or subtract vola-
tility from a portfolio. Presum-
ably, investment guidelines, regu-
latory structure, or something
else prevents an investor from
doing swaps to the point of disas-
ter, because you could easily
blow yourself up by levering
with swaps, just as you could by
buying or shorting too many fu-
tures contracts.
Question: Who are the logical
counterparties to receive the MBS
index?
Kopprasch: Any natural MBS
investor might be a logical coun-
terparty to receive the MBS index.
The investor would eliminate all
of the inconvenience of MBS in-
vesting: monthly principal and in-
terest on numerous pools, diffi-
cult settlement procedures, and
33
so forth. The swap would have
absolutely nothing to do with the
real estate market. The investor's
counterparty would probably be
a dealer who is carrying an inven-
tory of MBS and financing that
position.
Question: Is the value of
34
swaps coming from lower trans-
action costs, spanning (complet-
ing markets), or risk transfer?
Can't existing products achieve
the same results (span the same
payoff space)?
Kopprasch: At least part of any
advantage swaps may have come
from an increase in credit risk.
An MBS portfolio might consist
of all agency securities. The alter-
native-a swap-is really just a
promise to pay from a dealer, and
that might be combined with
lesser rated short-term securities.
egy involving derivatives must be tailored to suit the
liabilities. Otherwise, it is just an asset strategy.
I will describe options embedded in insurance
contracts that could be hedged and how the deriva-
tives-pricing technology can be applied to value the
liabilities. I will also discuss how to hedge these
liabilities with assets and with other liabilities.
The textbook objective of portfolio management is
managing the mean and variance of a portfolio's
end-of-period market value. This objective can be
simplified to obtaining risk-adjusted excess expected
returns. The expected returns can be specified by an
equilibrium asset-pricing model or by a passively
managed benchmark portfolio. The portfolio
manager's task is to add value by choosing a portfo-
lio or trading strategy that has higher returns than,
but the same risk as, the benchmark portfolio. The
asset/liability manager's task is to specify the bench-
mark portfolio.
In practice, excess expected returns seldom
count as much as excess actual returns, which may
be measured by accounting earnings rather than by
the change in total market value. I prefer to treat
practical features such as accounting earnings and
loss avoidance as constraints to the standard optimi-
zation problem. The objective remains to maximize
risk (variance)-adjusted expected return. The loss-
avoidance target becomes a probability constraint
requiring the chance of negative accounting earnings
in any year to be less than, say, 5 percent. These
issues are fleshed out by Martin Leibowitz in the
proceedings of AIMR's seminar on managing
Asset/Liability Management: Implications for
Derivative Strategies
Henry M. McMillan
Director, AssetlUabilityManagement
Transamerica OCcidental Ute Insurance Company
In asset/liability management, derivatives can help identify and quantify risks. Once
the risks are specified, derivatives can then be used to hedge those risks. They are also
useful in matching asset and liability duration and convexity within portfolios.
Asset/liability management is the management of a
firm's net worth position. The purpose of asset/lia-
bility management is to enhance the risk-return
trade-off on net worth rather than on the assets or
liabilities alone. Although the problems of asset/li-
ability management are similar to those for enhanc-
ing the risk-return trade-off on assets alone, the so-
lutions of asset/liability management must reflect
the peculiarities of a firm's liabilities. In financial ----------------------
institutions, asset/liability management focuses on Objectives of AssetlLiability Management
interest rate risk, but more broadly, asset/liability
management includes credit risk, operating risk, for-
eign exchange risk, and equity risk.
Derivative securities are securities for which the
cash flows are defined by the cash flows of other
securities. A futures or an option contract is a deriv-
ative security because its cash flows depend on the
value of the underlyingstockor bond at a future date.
Derivative securities can be priced by eliminating
arbitrage of cash flows between the derivatives and
other traded securities. The pricing techniques and
portfolio strategies for derivative assets have been
and continue to be widely discussed.
This presentation focuses on the risks implied by
liabilities and on how derivatives can be used to
hedge those risks. The liabilities of my organization
are life insurance contracts, annuity contracts, and
investment contracts. The contracts range from rel-
atively simple to quite complex. Most relevant to the
asset/liability management process, these contracts
typically have embedded interest rate options, many
of whose values depend on the history of interest
rates. A clear understanding of the liabilities is cru-
cial to developing a successful asset/liability man-
agement program. An asset/liability hedging strat-
35
asset/liability portfolios.
1
The solution to this problem for debt-oriented
liabilities and assets involves choosing a risk expo-
sure measured by duration, convexity, and cash flow
matching properties. The measures should be op-
tion-adjusted measures because the liabilities typi-
cally contain embedded interest rate options.
Basic Liability and Asset Structure
Insurance contracts provide for payments to policy-
holders contingent on death, disability, retirement,
or other events. The size of the payments may be
fixed or depend on current or past interest rates. The
premiums the policyholder pays may be fixed by the
contract or depend on current or past interest rates,
or they may be at the discretion of the policyholder.
Types of Liabilities
Insurance companies have several different
types of liabilities, including their insurance con-
tracts, annuities, and investment contracts.
Insurance. Life insurance products can be
classified into three types. Term insurance provides
life insurance for short periods of time, typically 5-10
years. Term insurance products seldom have a sig-
nificant asset/liability management problem, be-
cause annual premiums for a given group of policies
just about cover death benefits paid to the group and
expenses and profits for running the operation. That
is, term insurance has substantially more mortality
risk than interest rate risk.
Traditional whole life insurance provides life
insurance for an extended period of time (i.e., as long
as the policyholder lives and keeps the policy in
force). The annual premiums are fixed for life or to
some specific age, such as 65, or for some specific
period, such as 20 years. Because people are more
likely to die when old than when young, the premi-
ums are more than adequate to cover mortality risk
in the early policy years. The "extra" premiums are
invested to provide for insurance coverage in later
years. These policies build up substantial reserves
and cash values, which produce assets to invest and
interest rate risk to manage. A nonparticipating
whole life policy guarantees an interest rate on the
asset buildup. A participating whole life policy
guarantees a minimum rate but retrospectively pro-
vides higher interest rates if justifiedbyactual invest-
ment performance. Mutual life insurance companies
sell the majority of participating whole life insurance
in the United States. For them, the determination of
1"Setting the Stage," in Managing Asset/Liability Portfolios, ed.
Eliot P. Williams, CFA (Charlottesville, Va.: Association for
Investment Management and Research, 1992):6-13.
36
the policyholder dividends scale is a key asset/liabil-
ity management issue.
Universal life insurance was a key 1980s innova-
tion in life insurance coverage. Universal life is
whole life insurance with flexible premiums and
flexible credited interest rates. Basically, the policy-
holder can choose, within limits, how much to pay
each year. The insurer deducts the cost of insurance
during the current year and leaves the rest of the
funds in an account for the policyholder. That ac-
count is credited with a competitive interest rate.
The credited interest rate does not depend solely on
past investment performance, as in participating
whole life insurance. Universal life policies also have
guarantees as to minimum credited rates and maxi-
mum costs of insurance.
Annuities. Insurance companies also insure
the risk of living through annuities that pay while the
policyholder lives or for specified time periods. The
single-premium immediate-annuity market pro-
vides a guaranteed rate of return. Life expectancy for
these annuitants can easily be 20 years. For struc-
tured-settlement products, the guarantees can run
even longer.
Investment contracts. Insurance companies
also sell investment contracts in various packages.
The deferred-annuity market provides a specified
return for various time periods without any cash
flows to the policyholder. The contract can be a
single-premium deferred annuity or a flexible-pre-
mium deferred annuity with many premium pay-
ments by the policyholder. Guaranteed investment
contracts have become a popular investment vehicle
for pension saving. These contracts provide guaran-
teed returns with withdrawal and "dump-in" pro-
visions.
Other liabilities. The procedures discussed
here could apply to defined-benefit pension plan
liabilities, health insurance liabilities, the liabilities of
property and casualty companies, or interest rate
liabilities of industrial firms.
Asset Universe
The assets insurers choose are largely fixed-in-
come securities: bonds, pass-through mortgage-
backed securities, collateralized mortgage obliga-
tions (CMOs), direct mortgages, real estate, equities,
and cash assets. The bonds may be investment grade
or not and callable or not. The CMOs come from the
alphabet soup of tranches: A, B, C, Z, PAC, TAC, 10,
PO, and so forth. The direct mortgages may be on
commercial property, residential property, or agri-
cultural property. Direct real estate investment may
be allocated similarly. Equities are a smaller share of
insurance company investments largely because of
10.3(150,000) - 9.4(133,912) = 18
16,088 .
The interest rate risk of a long-term annuity can
be managed with the long bond and cash. An im-
munized portfolio would have $28,000 in cash and
$122,000 in the long bond.
The interest rate risk can also be adjusted with
futures contracts. The hedge ratio for the futures
contract should be
further that the required surplus is 3.75 percent of the
reserve or $5,370.
Assume that the annuity is initially funded with
a 30-year, 9 percent coupon noncallable bond pur-
chased at its par value of $150,000. The initial net
worth of the position, at market value, is $16,088 (that
is, $150,000 - $133,912). The initial durations of the
bond and the annuity are 10.3 and 9.4 years, respec-
tively. The duration of the net worth position is,
therefore,
where n is the number of futures contracts, F is the
face value of the futures contract, L is the market
value of the liabilities, B is the market value of the
bonds, and DL, DB, and OF are the durations of the
liabilities, bonds, and futures contract, respectively.
The example shown in Table 1 assumes that the
cheapest-to-deliver T-bond is a 20-year, 7.5 percent
coupon bond; its nominal value would be $95, and
its duration would be 10.1. Shorting 300 contracts
produces an equity position duration of 0.1. This
position hedges the market value of net worth for
small changes in interest rates.
This position does not hedge the book value of
surplus. The table also shows the duration of surplus
with and without futures contracts. The futures
hedge turns a long-duration surplus position into a
short-duration surplus position. Neither the
annuity's nor the bond's book value changes as the
DLL-DaB
FDF
n =
Consider first the case of cash flows, independent of
interest rate changes. The best examples are immedi-
ate annuities and structured-settlement payments.
The total liability cash flows could be stochastic,
because we do not know for sure who will die when,
but we will assume that the mortality risk is indepen-
dent of interest rate changes.
Assume an insurance company's only assets are
noncallable bonds and futures contracts. Its liabili-
ties are insurance contracts payable on some contin-
gency that could be life, death, or disability. If the
payments are unrelated to interest rate movements,
then classic immunization methods can hedge inter-
est rate risks. Three such methods are as follows:
Measure the duration and convexity of
the liability cash flows and choose assets
with durations and convexities to match
those of the liabilities you wish to im-
munize.
Select a portfolio with duration and con-
vexity to obtain the desired risk-return
trade-off in light of your objectives. This
strategy can be carried out with bonds
and cash.
Buy the bond portfolio that you want
and use T-bond futures to adjust the
portfolio's duration and convexity. Fu-
tures contracts can be especially helpful,
for example, if bonds are not available in
the right amounts, prices, or durations.
As an example, consider an immediate annuity
of $1,000 a month payable for 30 years. It is priced at
8.50 percent and reserved at 7.75 percent. Conse-
quently, its market value is $133,912, but the book
value of the reserve liability is $143,191. Suppose
regulatory restrictions on the amount of them an
insurer may own.
Asset/Liability Management for Fixed Cash
Flows
Table 1. Duration of SurplUSwith and without Futures
Without Futures With Futures
Asset/Liability Value Duration Value Duration
Bond $150,000 10.3 $150,000 10.3
Futures NA NA Short 300
contracts;
$95/ contract 10.1
Annuity (market) 133,912 9.4 133,912 9.4
Net worth (market) 16,088 18.0 16,088 0.1
Annuity (reserve) 143,191 9.8 143,191 9.8
Net worth (surplus) 6,809 21.2 6,809 (21.2)
Source: Henry McMillan.
NA = not applicable.
37
Liabilities
Basic insurance liability
Policyholder options
Withdrawal put option
Dump-in call option
Guaranteed rate floor
interest rate changes, but the future contract's book
value does change. The gains or losses on the futures
position may be marked to market or amortized over
the life of the bond, depending on the particular
accounting treatment. The market value hedge thus
reduces the variation in the market value of net
worth but increases the variation in the book value
of net worth.
Futures contracts can also be used to lengthen a
portfolio's duration. This use is especially effective
as the chance of bond calls increases. In Table 1, the
bonds were assumed to be noncallable. In practice,
the bonds are likely to be callable in 5 or 10 years.
Calls shorten a bond's duration, of course, and pose
reinvestment risk for insurers with long liabilities.
Dynamic futures strategies or options onfutures con-
tracts can be used to offset this call risk.
Asset'Liability Management for
Interest-5ensitive Cash Flows
Insurance cash flows are seldom independent of in-
terest rate changes. Rather, operating cash flows are
typically interest sensitive. Certain policy provisions
provide policyholders and insurance companies
with interest rate options embedded in insurance
products.
Policyholder Options
Policyholders have withdrawal rights. They can
withdraw their funds at book value less a surrender
charge. The surrender charge may be waived if the
withdrawal is partial (say, up to 10 percent) or if a
"bail-out" provision has been triggered because the
credited rate has dropped by the trigger amount or
below the trigger level. Alternatively, policyholders
can withdraw funds without terminating their poli-
cies by borrowing against their own accounts
through a policy loan. The loan may be at a fixed rate
or at a fixed spread over the credited rate.
Policyholders have credited rate guarantees,
which are either minimum annual rates (say, 3 per-
cent) or minimumcumulative guarantees (say, 5 per-
cent) for the life of their contracts. Policyholders for
flexible-premium products have the option to con-
tribute more or less money in any policy year. This
dump-in privilege allows unscheduled funds to earn
the same credited interest rate as scheduled premium
payments.
Insurer Option: Rate Reset
An insurer can change its credited interest rate.
Although this reset is typically done annually, the
insurer has some flexibility in selecting the time and
38
the size of the rate reset. The rate reset option is
different from a floating-rate contract in that the
insurer in a rate reset is not required to change the
credited rate. A floating-rate contract automatically
puts policyholder options in the money or allows
them to be exercised. If the credited rate is indexed
or tied to the performance of a particular mutual
fund, the product is termed a "variable" product and
is subject to additional regulations and reporting
requirements.
Assets, Liabilities, and Embedded Options
Assume that the assets and liabilities of the in-
surance company in the previous section have been
modified as follows to include the embedded interest
rate options:
Assets
Bonds
- Bond call option
Mortgage-backed securities
- Mortgage prepay option
Insurer rate reset option
The company follows the standard call-or-put
convention: As interest rates fall, fixed-rate securities
and interest rate call options increase in value; con-
versely, when interest rates rise, fixed-rate securities
and interest rate call options decrease in value but the
value of interest rate put options increases. The in-
surance firm is short both put and call options. It has
a short straddle. The options effectively are written
on current interest rates. The value of the options
varies with policy year and with the level of interest
rates.
When interest rates decline, bonds are called,
mortgages are prepaid, and CMO cash flows are
accelerated. The asset cash flows are reinvested at
low interest rates, which do not provide sufficient
spreads for the product to be profitable. Accounting
income declines, and surplus problems could
emerge.
If market interest rates fall far enough, credited
rate guarantees become binding. The credited rate
becomes super-competitive, which results in dump-
ins and extra premium income. Declining interest
rates can present problems, however, even if the
guaranteed rates are not reached. The insurer may
use its rate reset option too slowly or too little. Mar-
keting departments may argue for short-term losses
in order to continue to sell new business. Investment
departments may predict that interest rates will re-
bound, so lower rates now will be reversed later.
Computer systems or legal issues may limit the
insurer's ability to adjust credited rates. For exam-
ple, suppose that systems and regulatory constraints
permit rate resets only at the policy anniversary date,
that a l-month lead time is needed between a deci-
sion and its implementation, and that 3 months are
needed to determine that the lower interest rates are
here to stay. In this case, 16months would be needed
to reduce credited rates for all in-force policies.
When interest rates rise, withdrawals increase if
credited rates on existing accounts do not keep up
with market rates. Surrender charges can soften the
blow, but frequently these charges have graded
down to zero on in-force business. Even if a policy
does not lapse, it may become dormant as its pre-
mium flows dwindle. If cash flows tum negative,
then assets must be sold at anaccounting loss, further
eroding book surplus. If the insurer raises its cred-
ited rates to match the market, credited interest ex-
penses can increase dramatically because the higher
credited rate is paid on all deposits, not just marginal
deposits.
An Integrated Asset'Liability Management
Process
The traditional asset/liability management
problem is to design an asset portfolio to hedge
interest rate risk for a fixed cash flow liability. The
modem asset/liability management problem has
three components: to design an asset portfolio to
hedge the put and call options of the short straddle,
to design and implement a rate-crediting strategy to
optimize the value of the rate reset option, and to
design marketable products that reflect the cost of
embedded put and call options. Solutions to these
three components form an integrated asset/liability
management system. The use of derivatives in the
asset portfolio should be coordinated with the rate-
crediting strategy and product design. Only by look-
ing at the liability side can the active fixed-income
portfolio manager adjust the overall exposure to in-
terest rate movements to the desired exposure con-
sistent with predicted interest rate movements.
Modeling Issues
The question now is how many and which deriva-
tives to buy. To quantify the options on the liability
side, both asset cash flows and insurance cash flows
must be modeled. The product liability options are
path-dependent options: Their value depends not
just on current interest rates but also on the history
(the path) of interest rates. Credited rates and
competitors' credited rates are also path dependent.
Monte Carlo simulationprocedures workbetter than
binomial lattice procedures in such cases.
Several issues must be resolved before running
the simulations, namely, how to define the market
interest rate, how to model the cash flows, and what
the investment strategy will be.
Market interest rates. Yields on corporate
bonds and mortgage-backed securities are typically
modeled as spreads over T-bond yields. The issues
are how future Treasury yield curves will be gener-
ated and which ones to use.
Much management and regulatory cash flow
testing is done with deterministic interest rate sce-
narios. New York Regulation 126 specifies seven
scenarios. Historical or best-estimate scenarios gen-
erate useful explanations for management.
To value embedded product options, however,
stochastic interest rate scenarios are necessary. One
approach is to use a model that generates short (91-
day) and long (la-year) rates separately with an al-
lowance for correlation. The short and long rates
have separate volatility and mean-reversion coeffi-
cients. Most models assume a constant volatility
factor throughout, although there are good reasons
to relax this assumption. The modeler can include an
inversion adjustment to reduce the frequency and
extent of yield curve inversions. After all is said and
done, the yields should be adjusted to obtain an
arbitrage-free set of spot rates.
Cash flows. The model for asset cash flows
should account for defaults, bond call behavior, and
mortgage prepayments. Calls depend on the present
value of cash flows relative to call price. Mortgage
prepayments, expressed as multiples of Public Secu-
rities Association prepayment rates, depend on cur-
rent market mortgage rates and book rate.
In modeling insurance cash flows, the first task
is to split cash flows into interest-sensitive and non-
interest-sensitive categories. The non-interest-sensi-
tive cash flows might be fixed expenses and mortal-
ity. Even these variables can have interest-sensitive
components, however. To the extent that expenses
and interest rates both reflect inflation, expenses
should be correlated with interest rates. Mortality
could be sensitive to interest rates if healthy policy-
holders let their coverages lapse but unhealthy poli-
cyholders do not. (These indirect effects will not be
considered in this discussion.)
The interest-sensitive cash flows are everything
that remains. First, you must model your credited
rate strategy: What will you pay policyholders if
interest rates increase by 100 basis points? By 200
basis points? Second, you must model your
competitors' credited rate strategy and define the
market rate. Some competitors pay a current interest
rate (a competitor could be a one-year T-note rather
than another insurance company). Some competi-
tors pay an average of past interest rates. Because the
money tends to go to the highest alternative rate, the
most sensible approach is to define the market rate
39
Simulation and Interpretation
In a simulation, cash flows are calculated for
each period for each interest rate scenario. The pres-
ent value of cash flows for a given scenario is calcu-
lated using discount rates taken from the scenario's
interest rate path. The average across scenarios of all
present values is the fair value of the product, given
the assumed product options, reinvestment strate-
gies, policyholder behavior, and so on.
Problemscenarios can occur and should be iden-
tified. For example, a problem may be defined as the
surplus-to-reserve ratio dropping below 5percent in
any period of the simulation. The problem scenarios
are examined to determine the causes of the problem
and possible solutions. Could another asset portfolio
strategy prevent the problem? Could alternative
product options prevent the problem? How could
derivatives be used to prevent the problem?
product was sold-through an agency or through a
broker. Typically, fewer lapses occur through the
agency system than through the brokerage system.
Some of the discussion in the previous sections can
be clarified by using a simple example of hedging the
interest rate spread for a UL product. Consider an
insurer that is evaluating the spread risk in its uni-
versal life product over an interest rate cycle. The
product has three primary competitors, whose rate-
crediting strategies are depicted in Figure 1. One
competitor, termed a "short-rate follower," follows a
new-money method and pays the short rate less a
spread. A second competitor, a "yield averager,"
follows a portfolio method and pays a moving aver-
age of recent earned yields. The third competitor, a
Hedging a Universal Life Product
Comparison with Modeling Mortgage
Prepayments
A comparison of the modeling of policy lapses
and mortgage prepayments should be instructive.
Table 2 lists some common features.
Both mortgage prepayments and insurance cash
flows depend on an interest rate differential. For _
insurance products, the key differential is between
the policy credited rate and market credited rates.
Mortgage prepayments and insurance products can
have seasonal components. Insurance policies are
more likely to lapse around policy anniversaries, and
policy anniversaries can be clustered if new sales
tend to occur at the end of the year. The prepayment
probability of a mortgage depends on the length of
time since it was written and the time remaining to
maturity. Similarly, for an insurance product, the
time since the policy was issued and the time until
policy expiration affect lapse rates. The closest thing
to a mortgage prepayment burnout effect is an ulti-
mate lapse rate, which differs depending on how the
as the highest rate offered by all of your competitors.
The differential between your own credited rate
and the market credited rate has implications for
lapse and dormancy rates, premiums received, and
new business written. These factors must also be
modeled.
Payout-reinvestment strategy. The invest-
ment strategy allocates cash flows to various alterna-
tives, including dividends to stockholders and poli-
cyholders and income tax payments.
The portfolio allocations may be fixed through-
out the simulation or be sensitive to the interest rate
environment. Specifying legitimate strategies is im-
portant. If active turnover strategies are chosen,
make allowances for transaction costs.
Table 2. cash RowModeling Comparison: Mortgagesand Insurance Products
Item Mortgages Insurance
Cash flow source
Key interest differential
Seasonality
Time
History
Type
a
Mortgage prepayments
Couponvs.
current market
Summer-fall
Time since issue
Time to maturity
Burnout
GNMA
FNMA
FHLMC
Lapse, dormancy,
premium flow
Credited rate
vs. competitor
Policy
anniversaries
Time since issue
Attained age
Ultimate rate
Agency vs. broker
Source: Author's classification.
aGNMA = Government National Mortgage Association (Ginnie Mae).
FNMA =Federal National Mortgage Association (Fannie Mae).
FHLMC = Federal Home Loan Mortgage Corporation (Freddie Mac).
40
Figure 1. Competitor Rate-crediting Strategies
I '\
V .'",,'
. ';".
/ \ .
/ \
/ : \
/ : \
/ :
'. /.'
.. -/.'
/
-'"
V"
"
.""
"".
\ .
\
\ '.
\
\
\
\
"
/
/
/ :
/ .'
6
8
Figure 2. Market versus Insurer's Credited Rates
10
/
..../ .....
.' /
. /
/
/
./
"-
"-
"-
"-
'. "-
.... ),....
'-
8
10
Short-Rate Follower
Yield Averager
Mean Reverter
2
o --:7=3:---::!:85:----::9::7--7
10
'1
Months
Source: Henry McMillan.
Months
121 136 101 81 61 41 21
2
-1.0 '-------'----'---'----'-----'---'------'
1
The insurer would like to reduce its downside
deficiency without changing its basic investment or
credited rate strategies. The investment department
suggests a floor, but should the floor be written on
the short rate, the long rate, or an average of the two?
What should be the strike rate? Figure 4 plots the
Source: Henry McMillan.
0.6
-0.6
oL----.J_---.l._---L_--L_--L_---'-_--'----_--'---'
1 13 25 37 49 61 73 85 97 101
Months
-- Market-Credited Rate
1.0,----------------------,
- - - Insurer's New-Business Rate
...... Insurer's In-Force Credited Rate
Source: Henry McMillan.
Figure 3. Total Spread Sufficiency
The total spread is the difference between the
total earned rate and the in-force credited rate. The
target spread is 150 basis points. The total spread
sufficiency equals the total spread less the target
spread, and it is plotted in Figure 3. During some
periods, the insurer earns more or less than the target
spread, with the range being plus or minus 80 basis
points.
0.2
] 01---- - , -.
-0.2
0..
if]
"mean reverter," splits the difference between cur-
rent new-money yields and its perceived long-run
credited rate of 6.5 percent. The market rate is the
highest of the three competitors' credited rates. Be-
cause the three strategies differ so much, each com-
petitor pays the market rate at some point in the
interest rate cycle.
The gap between the market credited rate and
the insurer's credited rate affects the flow of new
premiums and the persistence of business already in
force. The assumed crediting strategy is intended to
model the most likely future rate-erediting strategy.
Credited rates are separated into the rate on new
business and the rate on in-force business. The new-
business rate is more sensitive to the current invest-
ment environment and competitor credited rates.
This rate splits the difference between the market
credited rate and the yield on new investments less
the target spread of 150 basis points; however, it will
never be more than 100 basis points above or below
the market rate. The average in-force credited rate is
a weighted average of the previous in-force rate and
the new-business credited rate. Consequently, the
in-force and new-business rates would converge to
the same rate if the yield curve and the new-business
rate remain constant for a long period of time.
Figure 2 shows the new-business rate, the in-
force rate, and the market rate. The new-business
rate leads the in-force rate both up and down. The
new-business rate leads the market rate down, but
tracks the market on the way up.
This hypothetical insurer has invested in a mix
of short and long bonds. The investment strategy
allocates one-third of investable funds to short bonds
and two-thirds to long bonds when the long rate is
above the short rate. The proportions are reversed
when the yield curve is inverted. The average earned
rate is a weighted average of the previous earned rate
and the rate on new investments.
41
156
Sufficiency without Hedge
Sufficiency with Hedge
61 81 101 121 141
Months
41 21
"-
/
/
Figure 5. Sufficiency with and without Floor Hedge
1.0
Source: Henry McMillan.
0.6
0.2
"Cl
0
"'
-0.2
0-
CfJ
-0.6
-1.0
1
1.0
121 49
'.,
73 97
Months
-- Short Rate
- - - Long Rate
Total Sufficiency
25
Figure 4. Comparison of Market Yields and Total
12
<J>
8
"Cl
QJ
6 ;>:
.....
OJ
4 ..I<:
....
"'
:::E
2
0
1
Source: Henry McMillan.
total sufficiency against the short and long rates. The
total sufficiency tracks the long rate more closely
than the short rate, which suggests that the floor
should be written on the long rate.
Figure 5plots the total sufficiencywithand with-
out the floor hedge. The cost of the hedge is assumed
to be 25 basis points per unit per period. The strike
rate is 8 percent. The worst deficiency in the hedged
portfolio is about 50 basis points. In this case, the
hedge provides too much protection when rates are
very low. An alternative hedge consisting of a series
of floors with decreasing strike rates could be written
that would establish a maximum deficiency of 50
basis points and cost less than 25 basis points.
The key point to take from this example is that
the appropriate hedge depends on the rate-crediting
strategy, competitors' rate-crediting strategies, and
the existing investment strategy. Alternative credit-
ing strategies would require different hedges. The
mean reverter or the short-rate follower would
choose a different floor because their crediting strat-
egies differ from the hypothetical insurer's crediting
strategy. The appropriate floor also depends on the
basic investment strategy, The appropriate deriva-
tive strategy thus depends on assets and liabilities,
not just one or the other.
Conclusion
Derivatives are useful in asset/liability management
in several respects. First, they can help identify risks.
That is, the theory underlying derivative pricing lo-
cates the risks and allows a determination of the
qualitative characteristics of those risks. Identifying
risks may be the most important use of a derivative
strategy; it allows one to think clearly about what the
risks really are.
Second, derivative pricing techniques can help
quantify risks. Before hedging a risk, you need to
know how much a risk costs today and how its cost
changes with interest rate changes.
Third, derivatives can be used to rectify or adjust
the risk exposure of the net worth position. Most
people think of this use when they consider using
derivatives to hedge a specified risk. The first two
uses help specify the risk to be hedged.
42
Question and Answer Session
Henry M. McMillan
Question: You mentioned that
your structured-settlement liabili-
ties extend to 50 years. How do
you deal with the asset reinvest-
ment risk in this fund?
McMillan: Reinvestment is a
problem that anyone in the busi-
ness has to address. There are
several methods. You can deal
with it up front when the product
is priced by requiring substantial
interest margins. You can in-
crease the surplus supporting the
product, which will tend to
lengthen your surplus position.
Or you can buy futures contracts
to lengthen your assets; this is the
method that I described earlier in
the talk. If you use margins, you
should make certain that the mar-
gins are real. When this business
was really up and rolling, many
companies funded their struc-
tured-settlement liabilities with
junk bonds, in the belief those
margins would persist. Some of
those companies are not in busi-
ness any longer.
Question: Why immunize abso-
lute value of surplus rather than
the surplus ratio? Why choose
one immunization method over
the other?
McMillan: The dollar value ap-
pears in monthly or quarterly
earnings. If you have guaranteed
a particular dollar amount of in-
come in the next quarter, you will
want to choose the dollar
amount. You might want the dol-
lar amount immunized if certain
people focus on dollars instead of
ratios or percentages in the short
term. Rating agencies, however,
care about the ratio of surplus to
assets, and so forth. You should
look at both dollars and ratios.
Question: You mentioned that
your holdings are mostly corpo-
rate bonds and Ginnie Maes, and
you obviously have sold options
with those. You did not mention
convexity. Is that something you
pay much attention to in a quanti-
tative sense?
McMillan: I view convexity as
more important to somebody in-
volved in a total return environ-
ment with a relatively short-term
focus. What matters to me is
what happens when interest rates
change a lot over long time peri-
ods. By the time you finish calcu-
lating all the option values, you
have everything you want, so
you do not have to bother with
convexity. That is, for the liabili-
ties with which I work, convexity
may be less useful than option
values as a way of measuring and
explaining risks. Convexity is a
means to an end.
Question: In conducting a cash
flow analysis using the seven sce-
narios required by the New York
Insurance Commission, suppose
you noticed a much lower level of
profits on your single- or flexible-
premium deferred annuities in a
rising-interest-rate environment.
Would you usually change your
crediting strategies rather than
your investment strategies?
McMillan: Choosing a crediting
or investment strategy with pre-
knowledge is not fair. For exam-
ple, you cannot legitimately set
an investment strategy specific-
ally for one scenario in which the
yield curve shifts up just because
you know that the yield curve
will shift up in this scenario. The
investment and crediting strate-
gies should be feasible and rea-
sonable to use.
The crediting strategy can
nonetheless be a complicated
function. In the example pre-
sented here, "the competitor" dif-
fers depending on whether inter-
est rates are rising or falling. The
basic strategy is to credit the port-
folio yield minus 150 basis points
but stay within 100 basis points
of the market. When the market
is going up and new-money
yields are higher than portfolio
yields, the strategy implicitly
changes the competitor to a new-
money company. The change is
built into the crediting strategy
function, the computer code, and
all the calculations. The invest-
ment strategy in the example also
depends on whether or not the
yield curve is inverted. These
strategies are legitimate because
the necessary information would
be available at the time the credit-
ing or investment decision is
being made.
Question: I have an impression
that most state insurance regula-
tory agencies are ill informed
about derivatives; they may be
somewhat more knowledgeable
in California. Many insurance
companies want to use deriva-
tives, but cannot because the regu-
lators are afraid of them. What
can you tell us about that?
McMillan: The regulators are
concerned about people using
strategies they-either the regula-
tors or the companies-do not un-
derstand. The concern seems rea-
sonable in light of the regulators'
charge to protect the public. This
43
problem, if it is one, is probably
not confined to state regulators.
Some observers believe that
Franklin Savings of Kansas City
44
was shut down because some fed-
eral regulators did not under-
stand Franklin's elaborate hedg-
ing techniques. Well-reasoned,
well-understood, and well-ex-
plained derivative strategies have
a legitimate use in an insurance
company's investment portfolio.
Embedded Options: Understanding
Callable / Putable Bonds
Lloyd McAdams, CFA
Chairman
Pacific Income Advisers
Derivatives can be used several ways to help manage fixed-income portfolios. To
manage a bond portfolio successfully, however, requires a thorough knowledge of
evaluating embedded call options. Listed options can be used to shift durations, time
the market, liquidate a portfolio, and hedge prepayment risk.
If bond portfolio managers began to use listed option
strategies as equitymanagers do, brokers would wel-
come it as a potential gold mine of newcommissions.
Such use is not likely, however, because bond port-
folio managers do not have as many consequential
opportunities to use listed derivatives as do equity
managers. The primary reason is that there are no
derivatives linked to the most commonly used bond
indexes. As a result, bond managers have not been
able to use strategies such as equity index spreads,
which have provided an efficient technique to en-
hance equity portfolio yields.
Bond Management Uses of Derivatives
The primary listed-derivative strategy that has been
of value to bond managers is duration shifting using
the futures market. With its long Treasury contract,
the futures market provides managers responsible
for large portfolios with nearly instant liquidity and
the ability to control the level of portfolio duration
cost-effectively.
The ability of a portfolio manager to use futures
to control other important bond portfolio risks such
as convexity, prepayment, and credit risk is either
limited or nonexistent. Only the swaps market has
the long-term potential to meet this need.
Managing Convexity Risk
An example of a bond portfolio risk is the nega-
tive convexity in portfolios of adjustable-rate mort-
gages (ARMs). Major financial institutions have es-
tablished ARM money market mutual funds. The
relatively poor results of several of these funds man-
aged by supposedly the most sophisticated organi-
zations are a testimony to the difficulty of sufficiently
controlling the uncertainty associated with the con-
vexity risk of these ARM securities.
Compared with other government agency secu-
rities of similar price volatility, an ARM portfolio
appears to provide an attractive incremental yield.
This apparent extra yield is primarily the result of
selling a long string of call options. As an example,
when buying a ten-year six-month reset ARM, we
have sold 21 call options embedded in this security.
If we could cost effectivelyhedge those options using
listed securities, we would have a good chance of
significantly enhancing the yield of a money market
fund.
Later on, we will discuss possible ways to hedge
such a security and the methodology that seems
likely to solve a large part of this problem. Although
this solution may not be the ultimate one, it is at least
a good start toward solving the problem. More im-
portant, however, it is a good example of how bond
managers can try to control complex risks, often in
indirect ways, using derivatives.
Managing Duration and Liquidity
Derivatives' most significant contribution to
bond portfolio management is the ability to control
duration. Simply put, go long a futures contract and
extend portfolio duration or go short a futures con-
tract and shorten portfolio duration. A second sig-
nificant contribution is increased liquidity. Not con-
sidered among the more significant contributions is
the use of futures contracts to add value directly as
do the equity managers who trade the basis between
45
an index's cash market and the futures market.
A successful interest rate forecaster must be able
to shift duration very quickly. The difficulty for the
successful large portfolio manager is not when to or
when not to shift duration; it is that he or she often
manages too much money to execute such large
trades in the cash market without significantly in-
creasing transaction costs. Small portfolio managers,
of course, usually have more-than-sufficient liquidity
to execute their smaller trades very near the last sale
price in the Treasury or even corporate cash market.
Therefore, they do not to the same degree need the
futures market for shifting portfolio duration.
unique, apparently simple, often overlooked and de-
ceptively complex idea: Bonds mature. Unlike
stocks, which never mature, each bond's volatility
will dramatically decline over its lifetime. Adding to
the complexity confronting the bond manager, the
volatility of a bond's yield appears to be more sensi-
tive to the business cycle than the volatility of an
equity security.
Figure 2 illustrates the effect of yield volatility on
future prices of a bond maturing in five years. The
bigger ellipse shows the price possibilities for a 14
percent volatility level. The smaller one is for a 10
percent volatility level.
5 4 3 2
---
--------
-------
85 L--__....L-__--L .l--__-L__-l
o
The Importance of Accurate Valuations
The tool most bond managers use to evaluate the
effect of an embedded option is the option-adjusted
spread. With this concept, a portfolio manager can
Most bonds are callable through options embedded
in their indentures. The only major exceptions are
most Treasury securities. Mortgage-backed securi-
ties, because of the uncertainty of homeowners' refi-
nancing practices, are callable in a highly unpredict-
able pattern. Prepayment rates of the underlying
residential mortgages determine both the timing and
the amount of the mortgage bonds that will be'called.
Since 1987, almost all of the return difference
among managers has come from evaluating the true
worth of the embedded options in callable bonds;
practically none has come from being able to use and
evaluate listed options. Understanding and prop-
erly evaluating embedded options in bonds has be-
come a crucial skill to manage a bond portfolio suc-
cessfully.
Years
110
-- 14% Volatility
- - - 10% Volatility
115..--------------------,
Source: Salomon Brothers.
90
Figure 2. 95 Percent Confidence Interval of Future
Prices
1i 105
(l)
~ 100
u
;t 95
Embedded Options
Deciding how to measure bond volatility is a
complex and important issue. There are many ac-
ceptable ways to measure volatility, and each pro-
duces a slightly different result. Among these are
historical versus projected, one-year versus three-
year versus five-year, and so forth. Stock market
analysts when measuring volatility use different
time horizons, but they also incorporate issue-spe-
cific volatility into their measures. For bonds, how-
ever, a similar measure of yield volatility is often
used for securities of the same maturity. Choosing
the correct time horizon over which to measure vol-
atility is therefore very important.
Bond managers must also deal with another
Years
Source: Salomon Brothers.
Figure 1. 95 Percent Confidence Interval of Future
Rate Levels
Managing Volatility and Uncertainty
To add value with fixed-income derivatives, a
manager must understand the nuances of yield vol-
atility measures. Figure 1 illustrates future rate lev-
els around today's yield of 8 percent; yield volatility
is assumed to be 20 percent. The 95 percent confi-
dence band reflects the expected range of future
yields for the next three years based on this level of
volatility. The elliptical shape is the result of the
standard deviation statistic being a function of a
squared number.
18
16
14
~ 12
~ 1 O
OJ
;;:: 8
6
4
2
0 2 3
46
170
150
Source: Salomon Brothers.
13 11 7 9
Yield (%)
-- Noncall Bond to Maturity
70L-------'--------'--------'---------'
5
bond is callable and the investor gets the benefit of
the "proceeds" of the sale of the options embedded
in the callable bond.
Compared with buying noncallable U.S. Trea-
sury bonds, buying a callable corporate bond is an
enhanced-yield strategy. When buying a callable
corporate bond, the investor sells a call option and
assumes an extra credit risk. These sold call options
increase the yield of the bond to the investor but also
significantly reduce the price sensitivity if market
yields decline and prices increase. As happens when
an investor sells a call option and prices decline, the
investor will hold the bond until the call's expiration,
but if prices increase, the bond is called away by the
issuer at the first opportunity.
Figure 3 illustrates the price-yield relationship
of a noncallable bond. The curve is convex. When
this 10 percent bond yields 5 percent, its price is
approximately 175.
Figure 3. P r i ~ Y i e l d Curve of a Noncallable Bond
(10 percent semiannual coupon, 28-year
maturity)
evaluate the yield of a callable bond as if it were a
noncallable bond. When an analyst eliminates the
effects of the uniqueness of the multitude of call pro-
visions from a group of callable bonds, the evaluation
of several different bonds can take place in a more
apples-compared-with-apples environment.
A bond portfolio manager must become profi-
cient in evaluating the option-adjusted spread to be
successful. Bond managers in the top quartile in
long-term rate of return are only 40 basis points
better than those in the bottom quartile. Often port-
folio managers lose more than 40 basis points by not
properly evaluating the worth of an embedded op-
tion inside a callable bond. The 1992 mortgage mar-
ket is a testimonial to this fact.
Several bond managers suffered in 1992 because
of their inability to evaluate embedded call options.
Frequently, these investors unwittingly sold an ex-
traordinary amount of overvalued and embedded
naked puts. So what happened to investors who sold
naked puts in the stock market in October 1987 hap-
pened to bond portfolio managers five years later.
There was a reason these bonds were yielding 13
percent at the time, but none of these investors appar-
ently stopped to figure out what they should have
yielded. If these bonds yielded in the 25 percent
range, the investor's reward for selling such a put is
fair.
Potential problems are still out there, and they
are significant problems that cannot be minimized.
If you are managing a bond portfolio and selling or
buying embedded call options inside the securities
you deal with, you must evaluate what you are
doing. Gone are the days when a broker could call
and offer a portfolio manager a new issue of General
Electric, 3D-year maturity and 10-year noncallable,
trading at 64. The manager might say "Well, a Gen-
eral Electric five-year is trading now for 55. I wonder
if 64 is enough? Yeah, I'll buy them." Those days are
over because the analytical technology necessary to
make these calculations is, although expensive,
available to everyone.
Callability
The increased risks of greater callability raise the
yields and lower the prices of bonds. Compare, for
example, the values of callable and noncallable
bonds of the same issuer. The noncallable bond has
a 3D-year maturity, 8 percent coupon, 8 percent yield
to maturity, and a price of 100. The callable bond has
the same maturity and coupon, but its yield is higher
because its callability makes it riskier. In this exam-
ple, the callable bond provides a 9 percent yield, so
its price is only 89. The yield spread between these
two bonds is 100 basis points because the corporate
Figure 4 illustrates the concept of the duration-
price relationship. When yield goes down, duration
goes up because the present value of the cash flows
increases because the discount factor of future cash
flows is also lower. Hence, duration, which is being
used to measure portfolio risk, is changing as yields
move. Although this difficulty in managing dura-
tion is not difficult to overcome, it is an important
example of the dynamic nature of the bond market
that bond managers must overcome.
Figure 5 illustrates two different price-yield
graphs that together represent price-yield behavior
47
13
called.
The price-to-worst curve in Figure 6 represents
the appropriate price-yield curve for this callable
bond. As a technical note, the actual price-to-worst
line is a smoothing of these two curve segments. As
an example of this situation, suppose a broker of-
fered a bond trading at 115, callable in July, with a
yield to first call of 30 basis points more than com-
mercial paper of the same quality. This pricing looks
like a 30-basis-point "free lunch." Yet, even with
only six months to maturity, there is still a finite
probability that yields will go up so dramatically that
the issuer does not call this bond. A portfolio man-
ager uses the option-adjusted spread to determine
the value of that probability. If the probabilistic
value of this happening were 50 basis points, the
bond yielding only an extra 30 basis points would be
overvalued. 13 11 7
7 ~ - - - - - ' - - - - - - - ' - - - - - ' - - - - _ - - - I
5
Figure 4. Modified Duration of a Noncallable Bond
(10 percent semiannual coupon, 28-year
maturity)
Source: Salomon Brothers.
Figure 5. Price-Yield Curve of a3-Year Noncallable
Bond and a28-Year Bond Noncallablefor
3 Years
(10 percent semiannual coupon)
for a bond with 28 years to maturity that can only be
called after 3 years at 108. The steeper curve repre-
sents a noncallable 28-year bond, and the flatter
curve reflects the behavior of a 3-year noncallable
bond. Together, these two curves describe the price
behavior of the callable 28-year bond. In low interest
rate environments, this bond will likely be called in
Year 3, and in higher interest rate environments, it
will likely mature in Year 28. In between, the bond's
price always incorporates the likelihood of its being
13 11 7 9
Yield (%)
-- Noncall Bond to Maturity
- - - Noncall Bond to First Call Date
70 ~ --'- ---L. - - - - J ~ __----'
5
130
Most people assume the worst case is not likely
to occur, and they are correct. But they are wrong in
believing the worst case can never occur and there-
fore they can ignore it. In October 1987, that event
did occur in the stockmarket in only a matter of days,
and to a lesser degree, it occurredin the bondmarket.
The bond market rallied, but investors with callable
bonds did not fully participate in the gains.
- Price-to-Worst Curve
..... , Price-to-Yield Curve
'"
150
170
Source: Salomon Brothers.
Figure 6. Price-Yield and PriC&-to-Worst Curves of
a Callable Bond
13
---
9 11
Yield (%)
--
7
---
170
150
130
QJ
u
.;::
0..
110
90
70
5
Source: Salomon Brothers.
Convexity
The practical effect of negative convexity is that,
48
Source: Salomon Brothers.
12% U.s. Treasury Bond
(maturity date 8/15/2013)
12% Southern California Edison
Corporate Bond
(maturity date 11/1/2012)
...
.........
. .....
...
I I
80
7
L------'9-----1.,Ll--------::173-----'
Yield of Treasury Bond (%)
periods of rising prices.
A 3D-year bond, noncallable for 5 years, is noth-
ing more or less than being long a 3D-year bond and
short a call option. The effective duration of this
bond is that of the 3D-year bond times the delta of the
embedded option. Hence, two 3D-year bonds, one
noncallable and one callable in5years, have different
durations because, on a probabilistic basis, they have
different expected maturities, price volatilities, and
deltas. The duration of a 3-year non-callable bond is
approximately 11 years, and the duration of a 3D-year
bond, noncallable for 5 years and priced at par, is
approximately 7 years.
Source: Salomon Brothers.
Figure 8. Historical Price sensitivity to Changes in
Interest Rates, weekly data, August
1986
'#
140 f-
120 f-
,..
..
<lI
U
c
...
w.
100 f-
13 11 7 9
Yield (%)
- - - Price-to-Yield Curve
70 L- --'-- -'-- --'-- ----'
5
90f-
80
100f-
120f-
--, . ./ Negative
1101-
.....
.....
.....
.....
.....
.....
.....
.....
.....
poo;A,
when yields go down, bond prices do not go up
commensurately. Positive convexity occurs when
yields go up or down and bond prices move faster.
Figure 7 illustrates the regions of positive and nega-
tive convexity in a price-yield curve. Convexity is
another concept to assist portfolio managers in eval-
uating the potential effect of embedded options.
Figure 8 is a price graph of two bonds during the
rapidly changing interest rate environment of Au-
gust 1983 through August 1986. It compares the
convexities of a Treasury bond and a callable corpo-
rate bond issued by Southern California Edison. The
price changes of the Treasurybond demonstrate pos-
itive convexity, while the SouthernCalifornia Edison
bond demonstrates negative convexity. As rates
moved lower, the price increase of the Edison bond
did not match the price increases of the noncallable
Treasury bond.
Figure 7. Regions of Positive and Negative
Convexity
Duration of aCallable Bond
The concept of effective portfolio duration is
portrayed in Figure 9. Alert bond managers must
know the durations of their portfolios. This is partic-
ularly important when they want to be duration
matched to a benchmark index. For the noncallable
bond, duration goes up as yields go down; for the
callable bond, the opposite is often true. The dura-
tion of a callable bond changes dramatically when
interest rates decline and the likelihood of its being
called increases. A callable bond's declining dura-
tion during a bull market is among the most signifi-
cant sources of poor investment returns during these
Options Strategies
Listed options are another tool for shifting durations,
timing the market, and liquidating a portfolio. Be-
cause most investors think futures can do these jobs
better, portfolio managers think of bond options as
not particularly interesting except in the context of
embedded options. Table 1 provides a strategy
framework for using listed options to anticipate mar-
ket changes in rates, spreads, or volatility. If a port-
folio manager anticipates a future level of rate vola-
tility less than the market consensus, he or she would
sell call options (overvalued based on the manager's
forecast). Conversely, if the manager projected the
49
..................................................
12
percent yield, 30 years to maturity, and a spot price
of 100. The one-year financing rate is 9 percent-an
investor buying this bond on credit will owe the
lender 109 at the end of one year. To break even, the
value of the bond plus the coupon income one year
hence must be sufficient to retire the loan. Therefore,
for the one-year forward price of this bond to be in
equilibrium with the spot market, the forward price
must be 99. The forward price is lower than spot for
a positively sloped yield curve and higher than spot
for a negatively sloped yield curve.
With knowledge of the forward price, the man-
ager can calculate the dollar change in price expected
for each change of 1basis point of yield (DV01). The
next step is to identify which Treasury bond is cheap-
est to deliver (CTD) against the futures contract. The
DV01 of the CTD determines the volatility of the
Treasury futures contract.
To identify the CTD bond, the manager calcu-
lates its forward price and divides that bythe conver-
sion factor. As interest rates move, the CTD bond
will have a changing DVOl. When another eligible
bond becomes CTD, the DV01 can change dramati-
cally, as shown in Figure 10. The hedge ratio traders
use has a probabilistic nature. In Figure 10, it is a
solid line reflecting the potential for a newCTDissue
to occur.
Example of a Hedge Strategy
A real-world example of a hedge strategy is out-
lined in Table 2. Suppose a mutual fund buys $20
million of Federal Home Loan Mortgage Corpora-
tion (FHLMC) 1419-A, a hypothetical ARM-backed
security. The ARMs reset every six months at LIBOR
plus 100. All the manager must do is make sure the
ARM security constantly resets and earns 100 basis
13 11
-------
7 9
Yield(%)
-- Noncall Bond to Maturity
- - - Callable Bond
. . . . .. Noncall Bond to First Call Date
2'-- -L..... --L '--__---l
5
14r-------------------,
Bond Hedging with Futures
A portfolio manager uses the forward price of a
bond to calculate the dollar value of a single-basis-
point change in yield. Assume a bond has a 10
/'
/'
/'
/'
./
./
/'
/'
/'
rate volatility to be greater than the consensus, he or
she would not sell call options, which by his estimate
were undervalued. These projections of future rate
volatility involve a high degree of uncertainty, and
most portfolio managers known to make this forecast
have not been noticeably successful.
Source: Salomon Brothers.
Figure 9. Effective Duration of a28-Year callable
Bond and a28-Year Bond Noncallable for
Three Years
(10 percent semiannual coupon)
Table 1. Options Strategies for Bond Managers
Strategy Motivation
Practical
Concerns/ Concepts
Adding exposure
Hedging
Portfolio restructuring
Fee generation
Synthetic assets
Anticipate market
changes in rates/spreads/
volatility
Risk reduction
Sell options to grow
or shrink bond
portfolio
Sell options to
increase yield
Change portfolio
return pattern
Simple/bounded risk
Availability/
reasonable cost
Flexibility
Right market
environment
Short-term
technique; attack
long-term embedded
bond options
Source: Salomon Brothers.
50
Source: Salomon Brothers.
80 L----L_...L-_.L----.JL---l_--'- ---'--------' 0.080
-240 -200 -160 -120 -80 -40 0 40
Yield Curve Move (Basis Points)
Figure 10. Price sensitivity of December 1990 Bond
Futures: Tracking the eTC Bond
delta, is 0.71 percent, or $142,000, which is the aver-
age of $134,460 (from the 100-point drop) and
$149,540 (from the 100-point increase). The gamma
is $15,080, or the difference between $149,540 and
$134,460. I have not yet found an off-the-shelf com-
puter program that does this calculation, but such a
spreadsheet is obviously easy to make.
Now we know what is going to happen when
interest rates go up or down 100 basis points. The
50-point change does not look bad, but we cannot
take a $149,000 loss from a 100-point increase, be-
cause we would then have the worst performing cash
management fund. If we get the $134,000 from a
100-point decline, then we would be the best per-
forming.
Instead of focusing on directly hedging the
rather unpredictable prepayment risk, we look for
derivative securities that might naturally go in the
opposite direction. To do this, we use the Eurodollar
futures contract, but we need a return pattern that is
slightly skewed like the ARM. Long a Eurodollar
futures contract (Euro-ED22) and short a listed put
option on this contract (96 PUT-EDH3) does the job.
To calculate the number of futures contracts and
put options needed to properly hedge the ARM, we
need the deltas and gammas of both securities.
Using either an optimizer or trial and error, we can
quickly identify a solution that almost eliminates
both delta and gamma risk. Specifically, we hedge
the $20 million ARMby shorting 47 futures contracts
and buying 27 puts. The aggregate delta of all three
securities is analmost nonexistent negative $200, and
....
0.085 ;
Q
_ 1.00
<lJ
_ 0.095 ~
o
o
_ 0.090 ~
<lJ
0..
--13.875% U.s. Bond
(maturity date 5/15/2011,
call date 5/15/2006)
- - - 13.25% U.s. Bond
(maturity date 5/15/2014,
call date 5/15/2009)
. . . . .. 7.5% U.s. Bond
(maturity date 11/15/2016)
points over LIBOR. The only uncertainty is prepay-
ment risk: The manager is unsure when this security
will mature-in three months, six months, or 30
years.
The security now trades at par. If interest rates
go down 100 basis points, mortgage prepayments
will change by some amount and the price will in-
crease to 100.67. If interest rates go up 100 points, the
price will go to 99.25. The average price change, or
100
ti
95
'" .b
~ """
c
0
U
90
....
<lJ
0..
....
0
> 85
Q
,
Table 2. Hedge Strategy Using Eurodollars and Short Puts
Rate Change (basis points)
Item -100 -50 0 50 100 Delta Gamma
Floater
FHLMC 1419-A
Price $100.67 $100.34 $100.00 $99.63 $99.25 -$142,000 -$15,080
Price change 0.672 0.336 0.000 -0.374 -D.748
Total change $134,460 $67,230 $0 -$74,770 -$149,540
Hedge
Euro-ED22P
Price $97.35 $96.85 $96.35 $95.85 $95.35 $117,500 $0
Price change 1.00 0.50 0.00 -D.500 -1.00
Total change -$117,500 -$58,750 $0 $58,750 $117,500
Number of contracts = -47
96 PUT-EDH3
Price $0.01 $0.04 $0.16 $0.40 $0.93 $24,300 $41,850
Price change -D.15 -D.12 0.00 0.24 0.77
Total change -$10,125 -$8,100 $0 $16,200 $51,975
Number of contracts = 27
Total hedge -$127,625 -$66,850 $0 $74,950 $169,475 $141,800 $41,850
Difference $6,835 $380 $0 $180 $19,935 -$200 $26,770
Source: DLJ Mortgage Research.
51
the gamma is a very attractive $26,770. Note that
delta and gamma are measured in dollars, not per-
centages. Although the use of dollars is not a com-
mon procedure in the academic literature, many
portfolio managers find them a helpful measure of
portfolio risk.
In this example, all five cumulative numbers are
positive. In Las Vegas, you would call this strategy
a jackpot. Now, we have a synthetic money market
security that has a significantly above-average cou-
pon and reduced negative convexity risk. Using this
methodology, the risk is primarily limited to using
significantly inaccurate projected PSA speeds with
the projected changes in interest rates.
52
Conclusion
The described methodology for applying derivative
theory to most portfolio management problems, al-
though known, does not as yet come in an easy-to-
use package. We are relatively certain, however, that
software vendors will soon introduce new programs
that can easily perform these tasks.
Most bond portfolio mangers know what to buy
and usually have a good sense of what risks they
need to avoid or hedge. Even if it is through trial and
error, many of these managers are now using deriv-
ative theory to address and solve complex problems.
Question and Answer Session
Uoyd McAdams, CFA
Question: How do you deter-
mine what is the right put con-
tract? Is there some computer-
aided process to find the right
one?
McAdams: The Black-Scholes
formula seems to continue to be
the most suitable valuation
model. To minimize transaction
costs, you will want the put to be
as long as possible.
Question: How could you out-
perform a bond index without
forecasting interest rates if you
didn't use options and duration
tweaking?
McAdams: If yield curve move-
ments were always parallel, it
would be very difficult to outper-
form the market without an inter-
est rate forecast. Because the
yield curve torques put it in some-
thing of a long-term disequilib-
rium, there is an opportunity to
take advantage of this disequilib-
rium on a short- to intermediate-
term basis. Embedded options
are an effective tool to implement
this strategy.
Question: What do you think of
client restrictions that specifically
prohibit the use of options and
other derivatives?
McAdams: Whether they know
it or not, practically all bond man-
agers use options and derivatives.
They are managing portfolios not
knowing what is in their portfo-
lios. Even though they have in-
vestment guidelines that prohibit
the use of options, the portfolio
has many embedded options hav-
ing the same effect.
Question: Some municipal
bonds have been sold with de-
tachable call options. Do you see
this developing into an important
instrument for portfolio manag-
ers?
McAdams: The municipal mar-
ket is relatively illiquid, so the
call options would probably have
the same degree of illiquidity. If
an active secondary market devel-
ops in these options, then the an-
swer would be yes.
53
Adding Value with Equity Derivatives: Part I
Gary L. Gastineau
Vice President
Swiss Bank Corporation
Derivatives are useful in both strategic and tactical applications, and they are particularly
appropriate in international investing because they are cheaper and more flexible than
direct investment. Performance evaluation is difficult, however, because traditional
statistical tools may not apply.
Portfolio insurance. When investors use op-
tions, they make a strong commitment to a particular
portfolio structure that deviates from the traditional
equity or other market exposure they would get if
they had a symmetric return. This commitment is a
strategic decision. Portfolio insurance is an example
of such a strategy; it deals with situations in which
the return is different from the traditional bench-
mark. The return pattern is asymmetric.
Asymmetric asset allocation. This category re-
fers to any asset allocation structure with a built-in
option payoff pattern. A decision to use such an
allocation is a strategic one. Asymmetric asset allo-
cation is not common today. It is used when asset
and liability management are integrated or when the
risk of return shortfall is unacceptable. Portfolio in-
surance is a special case.
Tactical decisions are those that have a very specific
application. A number of so-called options strate-
gies are really tactical decisions.. The tactical deci-
sions include the following:
Dividend capture. Dividend capture was
popular in the early 1980s. Corporate treasurers who
could benefit from the intercorporate dividend-re-
ceived exclusion hired dividend capture managers to
buy stocks paying high dividends, sell calls against
them or sell index calls, and obtain tax-favored divi-
dends for the corporations. Many corporations com-
mitted a significant amount of their short- or inter-
mediate-term cash management to these programs.
In 1987, selling a call against a stock was dearly
not providing the kind of downside protection many
corporate treasurers believed they had. A handful of
Tactical Decisions
Everything done with derivatives is referred to as a
strategy. This terminology is a holdover from a book
by Malkiel and Quandt written in the days of the
OTC option market that refers to the various option
positions as option strategies.
1
Contrary to conven-
tion, in my discussion of various derivative applica-
tions, I will differentiate between strategic and tacti-
cal applications of derivatives, because the differ-
ences are significant. I will also discuss performance
measurement and option payoff patterns.
IBurton G. Malkiel and Richard E. Quandt, Strategies and
Rational Decisions in the Securities Options Market (Cambridge,
Mass.:The MIT Press, 1969).
Strategic Decisions
Strategic decisions are those that deal with the big
picture. The strategic applications of derivatives in-
clude the following:
Statistical arbitrage and market making. The
decision to become an arbitrageur or a market maker
is a strategic one. A statistical arbitrageur tries to ----------------------
improve return on capital by buying something un-
dervalued and selling something overvalued in a
related market. The people engaging in statistical
arbitrage are usually large market participants, and
the arbitrage is a key part of their strategy. Any
market-making activity in which someone acts as a
specialist on an exchange, as a market maker in in-
terest rate swaps, or as a market maker in any other
security is a deliberate strategic decision. These peo-
ple decided to be in that business. The return pattern
arbitrageurs' and market makers' experience can be
very different from what would be expected in a
typical investment programmanaged and evaluated
relative to a benchmark.
54
managers "exploded" and lost a substantial amount
of money for their clients. Also, changes in the tax
law reduced the amount of dividends that could be
excluded from taxable income. The combination of
these two events almost ended the dividend capture
business. Even money managers with extremely
good records and customers who were not hurt
badly during the 1987 crash found dividend capture
to be an unpopular product. Although almost a
historical artifact now, in a period of high dividend
yields and high option premiums, this tactic may
make a comeback.
Other tax arbitrage. Many other tax-arbitrage
applications are practiced using options. One is tax
arbitrage of dividends to avoid withholding taxes.
Unless cross-border investors fall into a very specific
category in certain countries, they pay a withholding
tax, which for U.s.-based investors averages a mini-
mumof 15 percent of the dividend paid in the foreign
country. Many countries have dividend tax credits
that are not available to investors domiciled outside
the country. The return to a U.s.-based investor,
even a tax-free investor, is below the dividend-ad-
justed index return that a tax-exempt fund located in
the foreign market can earn. By using derivatives,
cross-border investors can usually avoid the impact
of the withholding tax. An equity derivative exe-
cuted across a national boundary is usually a mech-
anism to profit from withholding tax arbitrage.
Taxes on stock transfers are an important issue
for many derivatives. As illustrated in Joanne Hill's
transaction-cost examples? Japan and the United
Kingdom have transfer taxes for domestic security
transactions but essentially no such fees for deriva-
tive transactions (Japan has a small fee, but it is
immaterial relative to the transaction tax on stocks).
An offshore transaction sometimes avoids these
taxes, but doing the transaction with a derivative
usually avoids them entirely.
Taxes periodically playa role in index arbitrage.
They determine what kind of entity finds an index-
arbitrage transaction most useful. Extending the
dividend tax example, in Germany, a domestic in-
vestor who pays no withholding tax and receives a
dividend tax credit can buy stock, sell futures below
their fair value to a tax exempt investor and still
improve on a money market return. The key is the
use of the dividend tax credit. Apart from these
situations, tax authorities gradually close in on tax
arbitrage and eliminate most of the fun. More tax
arbitrage opportunities existed several years ago;
now, we are down to the transactions that are very
difficult for various tax authorities to get at.
Zsee Ms. Hill's presentation, pp. 62-71.
Regulatoryarbitrage. Although the world has
many futures contracts, not all of them can be used
by U.s.-based investors under the current crazy quilt
of regulations governing some of these issues. By
using OTC derivatives-embedding the return on a
futures contract in a swap, note, or option-an in-
vestor can get participation in these futures markets
through regulatory arbitrage.
Canada has severe restrictions on foreign invest-
ment. As a result, some of the most enthusiastic
users of derivatives are Canadian. A common prac-
tice for a Canadian issuer of debt is to buy from a
financial intermediary an option that provides par-
ticipation in some non-Canadian stock market index.
The option is embedded in a note the Canadian
company issues. That note becomes a Canadian se-
curity and falls under permitted domestic invest-
ments, so the Canadian investor can have interna-
tional diversification without going beyond the ap-
proximately 18 percent of assets a pension fund can
invest offshore.
Many investors cannot buy options. Others can-
not take a position in futures or do swaps, usually for
regulatory or tax reasons. These forbidden deriva-
tives can be embedded in notes. Equity-linked notes
are increasingly popular with highly regulated in-
vestors. The notes can be used to embed fixed-in-
come products as well as equity returns. This market
is growing rapidly. If investors see an application
they like today, they might ask their friendly finan-
cial intermediaries how they can embed that return
pattern in a note. Few investors are unable to buy a
note or something that looks exactly like a medium-
term note but has some features that go beyond
single interest rate return patterns.
Yield enhancement. Yield enhancement is
probably the most popular derivative application.
The term is a code word. Usually, someone has sold
an option and the option premium is added back to
the yield. The option may be hard to find in the
complicated structure that has beencreated, but ifyou
look hard enough, you can find it. When people
mention using interest rate swaps to increase their
yields or reduce their banking costs, occasionally they
have found a slight bargain in the swap market or an
inappropriate credit differential; these opportunities
are relatively infrequent, however. Most of the yield
enhancement or yield pickup people discuss comes
from options embedded in the transaction.
Investors must understand what they are giving
up in exchange for yield. Sometimes, despite the fact
that the option being sold may be fairly valued, the
transaction may make a great deal of sense. Reach-
ing this decision requires a fair amount of care, how-
ever. An investor can be less concerned with valua-
55
tion if the option is the most efficient way to imple-
ment a market forecast or if the resulting return
pattern has a higher utility than the return without
the option component.
Equity collars. An equity collar is a combina-
tion of a long put and a short call used to bracket an
equity return. Joanne Hill used the collar as an ex-
ample of year-end tactics to provide downside pro-
tection and, in return, cap upside return. As the end
of the year approaches, a manager has met his or her
bogey; the manager is ahead of the index or bench-
mark enough to want to provide downside protec-
tion and give up participation beyond a certain level.
The manager can use a collar as a tactic to bracket the
year-end return.
Symmetric international diversification. In this
tactic, synthetic futures are substituted for actual
positions in the common stocks; the synthetic long
call-short put is substituted for a long international
stock position, or the synthetic short call-long put is
substituted for the short international stock position.
The symmetric return pattern is distinguished from
the asymmetric when an option is involved. In the
asymmetric pattern, an option truncates the return.
International Applications
Most international diversification with derivatives is
much cheaper than holding or trading the securities
outright. You have to be careful in estimating costs,
however. If Goldman, Sachs & Company could exe-
cute a $25 million Financial Times Stock Exchange
(FTSE) 100 Index futures trade with a lO-basis-point
market impact, it would have 100 percent of the
market. That futures cost figure is fine for a contract
rollover, but for a real market-impact trade, it is not
possible to trade that cheaply very often.
For many investors, international diversification
is impractical without derivatives. Consultants have
urged U.S. investors for years to diversify internation-
ally. The investors look at the advantages and say
they understand the thesis for international diversifi-
cation but the costs are too great. Many of those costs
are substantially reduced with derivatives.
Derivatives provide greater flexibility than di-
rect investment in equities to get in and out of the
market, so asset allocators can change exposures to
various international markets. The application of
derivatives encourages indexation, but indexation is
not essential. An investor can still do derivatives
based on baskets of specifically selected stocks. The
greatest cost reductions, however, usually occur in
connection with indexation.
56
Performance Measurement and Option Payoff
Patterns
Option performance measurement is widely
misunderstood. This section will focus on Fischer
Black, one of the great financial minds of this gener-
ation. Black's entire career can be viewed as provid-
ing extensions to, and thus extending the useful life
of, the capital asset pricing model (CAPM). Each of
Black's major contributions falls into that category.
Black did much of the early work on CAPM
formulation. He wrote the original papers dealing
with restricted borrowing, short selling, and the ex-
tension of the CAPM to international diversification.
He is best known for the option model, which is a
direct extension of the CAPM and integrates options
into the market model.
3
If an investor has a portfolio of stocks and sells
stock options or anindex option against each of those
positions or against a portfolio and creates a delta-
neutral position, the Black-Scholes model says the
investor should earn the risk-free rate of return if the
options are fairly priced. That is the essence of the
Black-Scholes model: That one can create a portfolio
with fairly priced options and the underlying stock
and the risk-return trade-off falls on the market line,
which is shown in Figure 1.
Figure 1. Options and the Market Une
~ 100% Stocks
~ Risk-Free Rate
0'-------------------------'
Risk
Source: Gary L. Gastineau, The Options Manual, 3rd ed. (New
York: McGraw Hill, 1985).
Empirical Results
The empirical results when an investor sells op-
tions against a portfolio or buys calls and sells stocks
3Fischer Black and Myron Scholes, 'The Pricing of Options
and Corporate Liabilities," The Journal of Political Economy
(May/June 1973):637-54.
18 .----------------------,
Figure 2. The Merton-Scholes-Gladstein
Simulation Results
(136-stock sample)
short are not necessarily what the Black-Scholes
model predicts. Figure 2is a graph based on a study
by Merton, Scholes, and Gladstein.
4
This study
showed option-buying strategies falling the
market line and covered-writing strategies falling
below it. This finding surprised many people be-
cause the popular view is that there is something
magical about being a call-option writer.
Several studies purport to show some advan-
tages in writing call options. Figure 3 illustrates
advantages to call writing with different assump-
tions and different data. Obviously, the conclusion
implied by these two studies cannot both be right. In
fact, neither conclusion is correct.
Figure 4 illustrates a Monte Carlo simulation of
option buying and option selling using standard
deviation as the measure of risk. When call options
on varying percentages of the portfolio are sold, the
expected return moves down but the reduction in
standard deviation is greater than expected. If put
options on varying percentages of the portfolio are
bought, the result is a reduction in expected return,
but that reduction takes place at a greater rate than
expected. In effect, the standard deviation does not
fall as rapidly as would be expected when puts are
4R.c. Merton, M. Scholes, and M.L. Gladstein, "The Returns
and Risk of Alternative Call Option Portfolio Investment
Strategies," Journal of Business, vol. 51, no. 2 (April 1978):183-242.
24 21 18 15
- Bonds
12
1
2
3
4.
5. -
, B
A
9 6 3
8 lL-_...L__-L_---l__-'-----_...L-_-L_---l__
o
16
0
25
/
e
50
.2
14
75 "Y 25
<Ii 50
100
tf' 75
"0
12
100
u
<Ii
0-
x
10
18
16
14
l::
...
12
.2
&
10
8
_ Cash
6
0 2 4
Standard Deviation (%)
6 8 10 12 14 16 18
Standard Deviation (%)
-- Call Options
- - - Put Options
Source: Richard Bookstaber and Roger Clarke, "Problems in
Evaluating the Performance of Portfolios with Options,"
Financial Analysts Journal (January/February 1985):48---{;2.
18.-----------------------,
Problems with the Standard Deviation
The nature of the standard deviation problem is
revealed by examining the probability distributions
of return from several option strategies or tactics.
Figure 5 shows a normal distribution. The standard
deviation intervals are given at the bottom of the
figure, together with the percentage of the observa-
tions that fall within each interval. About 34 percent
of the observations fall on each side of the mean
Figure 4. Standard Deviations and Retums of
Portfolios with Call and Put Options on
Varying Percentages of Portfolio
Source: Morgan Stanley.
purchased. This result is at variance with expecta-
tions created by the Black-Scholes model and the
CAPM.
Figure 3. Performance of Asset Allocations,
1985-90
30 25
-- Stock Only
10 15 20
Standard Deviation (%)
5
...... ;"
16 /'/' 10 /
/' I
/'
14
Option-Buying Strategies / /' / /
/ /
/ /
12
/
/
/
-9
/
l::
/
/
...
/ /
.2
10
/ /
<Ii
/ /
<e
/
/
/
_8 /
;::l
8
/'
l:: / /'
l::
/ 7 /'/'
.
6
_
<Ii
Y6- ) CIl
r""" /
4
/' 5_ /
/ 4_ /'
(3- ...... /' C d W .. S .
2
_ ...... overe - ntmg trategles
2
Source: Gastineau, The Options Manual.
57
Figure 5. The Nannal Distribution
-2
Standard Deviation (number)
-1 0 2
-60 -40 -20 o
Standard Deviation (%)
Return Distribution
20 40 60
Standard Deviation
<-4 -4 to-3 -3to-2 -2 to-1 -1 to 0 oto 1 1 to 2 2 to 3 3t04 4+
Interval
Probability (%) 0.2 2.1 13.6 34.1 34.1 13.6 2.1 0.2
Source: Swiss Bank Corporation.
Note: Mean is 3.5 percent; standard deviation is 21.21 percent.
within one standard deviation, or about 68 percent
within one standard deviation on either side. Within
two standard deviations on both sides of the mean,
the percentage is up to about 95, and at three stan-
dard deviations, it is up to 99. This pattern is the way
the normal standard deviation distribution works.
If an option is added to the basic distribution of
returns on stocks, the picture is different. Figure 6
illustrates the distribution associated with selling an
out-of-the-money call option. Obviously, the call
changes the probability distribution dramatically.
The probability between plus and minus one stan-
dard deviation is only about 30 percent instead of 68
percent. The probability on the upside is greater, but
the lowest observations on the downside are much
lower.
Figure 7 illustrates the distribution for a long
at-the-money call; it shows the dispersion on the plus
side. A large number of observations are in the in-
terval of zero to minus one standard deviation, and
participation in the extreme areas on the upside is
more than in Figure 5.
Figure 8 diagrams the distribution for a collar.
All observations in this particular example fall be-
tween plus and minus one standard deviation. The
standard deviation is fairly broad, however. In ef-
fect, no matter how the investor moves the strikes on
58
a zero-premium collar up or down, within reason,
this kind of relationship will hold. The standard
deviation will be sizable or very small, depending on
the location of the two option strike prices.
These figures show that options do not fit neatly
into the standard deviation model. The standard
deviation is not as standard as one assumes. Stan-
dard deviation does not mean quite what we were
taught in school when options are involved. Mea-
sures such as skewness and kurtosis are not mean-
ingful numbers. Traditional statistical tools can be
overused and abused in analyzing options.
Other Risk Measures
Because of the problems with standard devia-
tion, the development of other risk measures is being
emphasized. When Harry Markowitz initially intro-
duced the concept of diversification, which became
the basis of modern portfolio theory, he did not use
standard deviation to measure risk. He used semi-
variance, and many people have refocused on that. I
amnot sure this focus adds a great deal to the picture,
but it does provide a different perspective. Semivari-
ance focuses on the one-direction character of many
kinds of risk, something the standard deviation
misses.
Other people focus on the risk of shortfall,
Figure 6. Out-of-th&Money call
Standard Deviation (number)
-3 -2 -1
-50 -25
o
o
Standard Deviation (%)
Return Distribution
2
25
3
50
Standard Deviation
Interval <-4 -4 to-3 -3 to-2 -2 to-l -1 toO oto 1 1 to 2 2 to 3 3 t04 4+
Probability (%) 3.1 4.4 8.0 12.1 15.3 16.1 18.1 23.0
Normal(%) 0.2 2.1 13.6 34.1 34.1 13.6 2.1 0.2
Source: Swiss Bank Corporation.
asset/liability matching, lower partial moments, and
the Sortino ratio. All of these measures focus on the
downside risk. These measures are important be-
cause many financial products today place greater
emphasis on asymmetrical return patterns than
products did in the past.
Option valuation should be key to any decision
option investors make; it can assure that they are not
paying too much or receiving too little for the options
they use. In addition, however, investors must look
at their expectations for future patterns of behavior
in the underlying rate or instrument they are model-
ing and the return they are creating. Utility is also
important. If the risk of an absolute shortfall is un-
acceptable, the investor must buy a put or create
some structure that prevents losses beyond a certain
point.
More customized risk measurements and differ-
ent approaches to the question of option perfor-
mance measurement will appear. Currently, there is
no good way to measure option performance. So far,
AIMR has wisely sidestepped the issue of option
performance in its Performance Presentation Stan-
dards. AIMR requires a discussion of option appli-
cations, but it does not yet prescribe a particular
standard for option inclusion.
The focus on the exposure of flawed option strat-
egies, which we have only touched on here, will
increase. In finding a flawed strategy, you must not
look at the traditional rate-of-return calculations or
the mean-standard deviation relationship; instead,
focus on option valuation and on the fit of a particular
strategy or tactic to the structure you are trying to
create.
Some alternatives to the standard ways of deal-
ing with options are available-the Procrustes ap-
59
Figure 7. Long At-the-Money Call
Standard Deviation (number)
-3
-50
-2
-30 -10 10 30 50
60
Standard Deviation (%)
Return Distribution
Standard Deviation
Interval <--4 --4 to-3 -3 to-2 -2 to-1 -1 toO ato 1 1 t02 2 t03 3t04 4+
Probability (%) 54.3 24.7 14.3 5.2 1.2
Normal(%) 0.2 2.1 13.6 34.1 34.1 13.6 2.1 0.2
Source: Swiss Bank Corporation.
Note: Mean is -2.65 percent; standard deviation is 14.8 percent.
Figure 8. Risk Reversal
Standard Deviation (number)
o 1 -2 -
Ii
f--""
-25 -20 -15 -10 -5 o 5 10 15 20 25
Standard Deviation (%)
Return Distribution
Standard Deviation
Interval <-4 -4 to-3 -3 to-2 -210-1 -1 toO oto 1 1102 2 to 3 3 t04 4+
Probability (%) 54.20 45.80
Normal(%) 0.2 2.1 13.6 34.1 34.1 13.6 2.1 0.2
SOUfce: Swiss Bank Corporation.
Note: Mean is 5.7 percent; standard deviation is 14.9 percent.
proach, for example. It is named for a Greek bandit
who put people on a bed of one size no matter
whether they fit or not. If they were too short, he
stretched them. If they were too tall, he cut off their
feet. In options, the Procrustean solution is the ap-
plication of standard mean-variance models
whether they fit or not. It is not a particularly satis-
factory approach.
The LIFFE approach is similar to what many risk
managers use; that is, a scenario analysis that deter-
mines what portfolio results look like if underlying
returns go very high or very low. In effect, this
approach provides a distribution of returns that can
be related to the distribution of prices or rates on the
underlying asset.
The BARRA solution is usable only to evaluate
managers; it does not work well in evaluating a
portfolio. Other new models may crop up, but a
universal option-portfolio benchmark is probably a
great deal like the Holy Grail.
61
Question and Answer Session
Gary L. Gastineau
Joanne M. Hill
Question: In light of the high
volatility of the Nikkei 225 and
possible market manipulation of
the index, how good or appropri-
ate are tracking models? Are
there any references to the opti-
mal adjustment of tracking bas-
kets?
Hill: The Japanese market pres-
ents two problems. First, the
index has gone from 14 percent to
28 percent volatility in a few
years. With that amount of
change in risk, the tracking mod-
els, which are often based on five
years of historical data, are under-
going changes. The numbers for
tracking error move. For exam-
ple, the BARRA tracking num-
bers, because they are based on
historical data, gradually work
their way up to adapt to the new
volatility environment. Because
of this change, it is hard to know
whether the tracking has gotten
worse or the volatility has gotten
higher.
Second, because the Nikkei 225
is a price-weighted index, build-
ing a tracking basket for it is very
hard because small, less liquid
stocks can influence the index lev-
els and need to be represented.
People deal with this problem in
several ways. One is to use a dif-
ferent benchmark. Managers
must decide whether using a par-
ticular benchmark like the MSCI
Index or the FT is more important
than constructing a capitalization-
weighted benchmark out of the in-
dexes that underlie the futures.
Global investors must start think-
ing more about how to get expo-
sure and less about how close
they are to a country's index. Be-
cause what is the best index is not
72
clear in many of these countries,
global investors have more toler-
ance for tracking risk.
For rebalancing, managers
must look at the most recent data
and use statistical regression more
than fundamentals in developing
global tracking baskets.
Question: Do time-decay issues
with regard to collars affect the
true payoff at expiration? If not,
why worry about time decay?
Hill: An investor with a specific
horizon who keeps to that hori-
zon does not need to worry about
closing out the options strategy
prior to the short option positions
experiencing their time decay.
The horizon does not affect the
payoff at expiration. If an in-
vestor has a specific horizon and
does not care what happens in the
interim, he or she can put the in-
vestment in place and go away for
the six months or a year until it is
over. That individual is hard to
find, however. Most people will
need to unwind the strategy be-
cause of new circumstances. Sup-
pose the index has risen sharply,
and the investor now wants to
buy exposure. The investor wants
to see what the cost will be to buy
these options back and sell them
in the interim. Something like
that can always happen.
Options, like bonds, force you
to choose a maturity, but people
who buy 3D-year bonds do not
necessarily have 3D-year horizons.
They might trade in and out of
long-termbonds but still want
them for their duration character-
istics. You might have the same
type of strategy or tactic with op-
tions.
Question: Which non-U.S. op-
tion contracts (exchange traded
or OTC) are approved for use by
U.s. investors?
Gastineau: The issue on the fu-
tures contracts is buried in the leg-
islation covering the renewal of
the Commodity Futures Trading
Commisssion (CFTC) authoriza-
tion that preceded the late 1992
reauthorization. When Congress
renewed the authorization, it
added a complex procedure by
which the SEC and CFTC must
both approve an index. The ex-
change in which the futures con-
tract is traded must subject itself
to the authority of the CFTC, and
a few other things must be done.
No such requirement is in place
for options contracts. If U.S. in-
vestors want to buy options con-
tracts on the DAX or Hang Seng
indexes (for which they are not
permitted to use futures con-
tracts), they can, because options
do not fall under CFTC jurisdic-
tion.
Question: Please explain "mar-
ket impact."
Hill: Market impact is a trader's
estimate of how much a transac-
tion would effect the market or
the size of the spread. It is an em-
bedded cost of trading. Perold
wrote an excellent article recom-
mending taking a snapshot of a
portfolio before transacting, hav-
ing a trading horizon, and then
looking at what the realized cost
. k' 1
IS to measure mar et Impact.
1Andre F. Perold, "The Implementation
Shortfall: Paper versus Reality," Journal of
PortfoliO Management, vol. 14, no. 3 (Spring
1988):4-9.
Question: Over a long term, say
ten years, would the cost of main-
taining a futures position in in-
dexes such as the EAFE become
more expensive than maintaining
a moderately diversified portfolio
of the same stocks with reason-
able turnover?
Gastineau: First, the EAFE is
not the index you want to match.
Second, doing a continual roll of
futures for ten years is probably
not the way to go on most of
these futures contracts. Ap-
proaches that might be attractive
are a ten-year swap or a sequence
of swaps of shorter duration that
are continually rolled over with
the same financial intermediary
or, by agreement, switched from
one financial intermediary to an-
other so that the withholding tax
is avoided. If the financial inter-
mediary is, in effect, making the
decision for its own account
whether to take the position via
futures or via the underlying
stocks and thereby taking advan-
tage of the tax characteristics, the
swaps will in many cases provide
a better return than taking a posi-
tion in the underlying stock in the
index. Each institution must look
at the economics of each of these
markets for its own particular cir-
cumstances.
For example, most investors
who want to invest in the Ger-
man market today would be well-
advised to go to their friendly fi-
nancial intermediaries. The inter-
mediaries would either swap
them or sell them a return embed-
ded in some other instrument
that would outperform the DAX
over any reasonable time horizon
during which they would want to
take the position.
The DAX is a total return
index, so it includes dividends.
Investors must watch whether
the quotes they are getting on
other indexes include dividends.
Different firms will quote these in-
struments in different ways.
Question: Is stochastic domi-
nance a sure way to choose be-
tween options alternatives?
Hill: The stochastic-dominance
approach might be useful in prov-
ing a strategy that appears to
dominate an alternative strategy
that lies on an efficient frontier in
mean-variance space. St6chastic
dominance looks at how a whole
distribution compares with an-
other. This approach may screen
out strategies that appear to domi-
nate in mean-standard deviation
spaces but in fact do not when
you look at the entire distribu-
tion. This concept is difficult to
apply in practice, however, be-
cause the entire distribution must
be estimated and the stochastic-
dominance criteria can find a
wide range of strategies that do
not dominate one another.
Question: Please comment on
option performance measure-
ment.
Hill: Option performances for a
portfolio manager should be inte-
grated with that of the underly-
ing index or stocks being hedged
with options. The delta of the po-
sition combined with options can
be useful in constructing risk-
equivalent strategies for compari-
son purposes.
73
Adding Value with Equity Derivatives: Part II
Joanne M. Hill
Vice President
Goldman, Sachs &Company
Futures and index options provide a way to participate in international equities markets
at relatively low cost. The choice among option strategies depends on the investment
objective. In general, futures are used to provide equity exposure to a particular country
market, and index options are used to manage risk.
This presentation covers global applications of eq-
uity derivatives. I will discuss specifically the use of
futures for managing equity exposure, what is avail-
able, and how it is done. I will also reviewsome basic
hedging and risk management applications,
illustrated by case studies involving asset allocation
and active management.
Futures Trading
Futures are attractive because they can be used to
achieve the economic equivalent of equity exposure.
Investors can use both stock index futures and equity
index swaps to gain exposure to most international
equity markets. Thus, the three ways of owning an
index fund are:
buy the index and receive the dividends
and capital gains (or losses);
buy the futures contract and invest the
body of the notional amount in a cash
equivalent with the same expiration
date or close to the same term as the
futures contract and receive interest in-
come on the cash-equivalent investment
and the capital gains from the futures
contract;
buy the swap, receive the interest in-
come on the reset dates and the total
return on the index (that is, the capital
gain plus dividends, although if so spec-
ified in the contract, the dividend pay-
ment is netted out of the interest rate
payment).
These equivalencies establish what swap rates
should be and what futures prices should be at all
times. The basic premises are that an investor is
62
indifferent between buying stocks and futures when
the futures premium is equal to the difference be-
tween the interest rate and the dividend yield, times
the index, and the investor is indifferent between
buying stocks and enteringinto a swap when interest
income is equal to the fixed- or floating-rate pay-
ment. If those rates differ from the equivalency-set
rate, then arbitrage transactions occur. The cost and
ease of conducting the arbitrage and the likelihood
of markets being open when needed determine the
point at which arbitrage takes place and how close
the three alternatives track each another.
The primary difference between a futures con-
tract and a swap is that, with the future, the transac-
tor manages the rolling process for expiring futures
contracts during the equity exposure, and with a
swap, the dealer or counterparty manages the rolling
process. When the futures contract expires, the
transactor must sell that contract and buy the next
term's contract. That process introduces a risk-re-
ward trade-off in the calendar spread, or the cost of
rolling. This risk-reward trade-off is imbedded in
the pricing of the swap. The dealer will assess where
the spread is likely to trade for rolling positions and
price the swap accordingly. Some people think fu-
tures are more flexible than swaps for positions with
high turnover, because the transactor is in the
driver's seat in terms of rolling positions and adjust-
ing the amount created in equity exposure.
One of the primary ways derivatives add value
is through the savings in transaction costs. Table 1
lists some estimated round-trip transaction costs in
selected markets as of November 1992. The round-
trip costs do not include custody and settlement fees,
and market-impact costs are estimated by the trader.
Transaction costs vary by market. Derivatives,
Table 1. Estimated Round-Trip Transaction Costs as Percentages of Amounts Invested
Cost United States Japan United Kingdom France Germany
Stocks
Commissions 0.14% 0.20% 0.20% 0.10% 0.10%
Market impact
a
0.55 0.70 0.90 0.55 0.50
Taxes 0.00 0.30 0.50 0.00 0.00
Total 0.69% 1.20% 1.60% 0.65% 0.60%
Average stock price in U.s. dollars
b
$39.10 $27.30 $7.73 $158.60 $385.65
Futures
C
Commissions 0.01% 0.06% 0.02% 0.04% 0.03%
Market impact
a
0.05 0.05 0.10 0.10 0.05
Taxes 0.00 0.00 0.00 0.00 0.00
Total 0.06% 0.11% 0.12% 0.14% 0.08%
Source: Goldman Sachs.
Note: Statistics are representative of a $25 million, capitalization-weighted, indexed portfolio executed as agent; does not include settlement
and custody fees.
aTrader estimate.
bLocal indexes: S&P 500, Nikkei 225, FTSE 100, CAC-40, and DAX, respectively.
cAll contracts are quarterly except for the CAC-40.
Table 2. Countries with Stock Index Futures
Sources: Commodity Futures Trading Commission and
FT-Actuaries World Markets Monthly.
aApproved for use by U.s. investors.
therefore, can add more value for applications in
higher trading cost markets, such as Japan and the
United Kingdom than they do in the lower cost mar-
kets such as the United States.
The magnitude of index futures trading in vari-
ous countries is shown in Table 2. Six of the markets
are approved by the Commodity Futures Trading
Commission for trading by U.S. investors, but U.S.
investors can transact in the inaccessible markets by
using synthetic futures, combinations of a long call
and a short put for a long position (or a short call and
long put for a short position) struck at the same strike
price. Synthetic futures can be structured to expire
close to the expiration of the exchange-traded futures
Australia
a
1.3%
Canada
a
1.8
Denmark 0.3
France
a
3.3
Germany 3.3
Hong Kong 1.4
Japan
a
25.4
Netherlands 1.5
New Zealand 0.1
Spain 0.9
Sweden 0.7
Switzerland 2.0
United Kingdom
a
10.5
United States
a
42.8
Total as of December 31,1992 95.3%
contract. This strategy involves two OTC options,
whichare priced off of the futures price with a spread
to the dealer for creating the transaction. When these
synthetics are included, about 95 percent of the
world's market capitalizationis covered with futures
at this point.
Figure 1 shows the growth in the average daily
dollar volume of various index futures between 1991
and 1992. Although the volumes of some of these
contracts appear small relative to activity in the
United States and Japan, they must be taken in the
context of percentage of the world's capitalization.
In Europe, the growth in trading activity from 1992
to 1993 was 50-80 percent. Japan experienced a big
drop in futures trading, consistent with the decline
in equity trading volume during this period.
For all the countries listed, the equity index fu-
tures now trade in dollar volumes that are close to or
more than trade in the underlying stock market on
an average trading day. To put this volume into
perspective, remember that a futures trade is really
another form of portfolio trade, or an index trade.
The cash market is dominated by individual stock
traders or active managers; about 80 percent of the
participants trade in individual stocks. In contrast,
futures are used most often to facilitate portfolio
trades. If traders do a portfolio trade in Europe and
quote a charge for that based on the market closing
prices, they will hedge that position with futures.
They will price the portfolio trade based on where
they can establish the hedge in the futures market.
Globally, almost all portfolio trading includes look-
ing to the futures market and determining what a
futures hedge will cost. That cost will help in pricing
the basic transaction.
Another measure of the size of markets is open
Percentage of FT-A
World Index Country
63
Figure 1.
17.5
15.0
12.5
<JJ
...
;:
10.0 0
Cl
'+<
0
<JJ
7.5
~
~
5.0
2.5
0
Average Daily U.S. Dollar Volume for Index Futures around the World
S&P OSAKA SIMEX TOPIX AllOrds FTSE
500
.1992
01991
DAX CAC SMI
Key:
S&P 500 - United. States
OSAKA - Japan
SIMEX - Singapore
TOPIX - Japan
AllOrds - Australia
Source: Goldman Sachs.
FTSE - United Kingdom
DAX - Germany
CAC - France
SMI - Switzerland
interest. Figure 2 shows the growth in open interest
volume for index futures between 1991 and 1992.
Most institutions establish maximum trading posi-
tion guidelines relative to open interest. These com-
Figure 2.
40
35
30
<JJ
...
25
~
Cl
'+<
20
0
<JJ
~
0
::3 15
~
10
5
Average Daily U.S. Dollar Open Interest Volume for Index Futures
around the World
64
S&P OSAKA SIMEX TOPIX AllOrds FTSE
500
.1992
01991
Source: Goldman Sachs.
Note: Please see Figure 1 for key to exchanges.
DAX CAC SMI
panies do not want to try to trade too large a portion
of the open interest because, when they need to roll
their positions, the liquidity may not be sufficient.
Enhanced Index Strategies
Enhanced index strategies combine index futures (or
swaps) with cash management or structured fixed-
income products to earn incremental return over the
index. With futures trading on global stock indexes
and government bond markets, an exposure can be
easilycreated that participates inupside or downside
moves in the underlying security or index. The
funds in excess of initial margin can be invested in
cash-enhanced strategies to add an alpha to the un-
derlying index return. Some of the incremental re-
turn can be compensation for taking on credit risk.
Any cheapness in the futures would be an added
benefit to long-exposure strategies.
Because the futures trader has control of the
notional value, many traders like the flexibility of
denominating that notional value in a chosen cur-
rency. They can be long Chicago Mercantile Ex-
change/Nikkei 225 futures, which have their gains
and losses in dollars, invest in Eurodollar deposits,
and have the yen risk hedged without any explicit
currency transaction. Index futures gains and losses
are denominated in local currencies (with the excep-
tion of the CME/Nikkei and CME/FTSE contracts).
Currency hedging must be done dynamically based
on realized gains and losses if the investor wishes to
have a fully hedged position.
Using these strategies, investors can separate the
management of the currency risk from the equity
exposure. In managing equity exposure, the investor
can try to add value within that notional amount by
such strategies as trading at the high end of the risk
spectrum, using long-and-short strategies, volatility
trading, and taking on credit risk. These strategies
add value through transaction cost savings, return-
ing more than the T-bill rate by taking on some
additional risk, and taking advantage of the cheap-
ness or richness of a futures or swap price. This
opportunity varies over time and among markets.
The S&P 500 Index and CAC-40 Index are the most
efficient in tracking fair value. The Nikkei 225 Index
and the German and Swiss indexes have deviated
from fair value in recent years. An investor posi-
tioned on the other side of that demand for lIquidity
has an opportunity to buy cheap or sell rich.
The five basic reasons for misvaluation of index
futures are as follows:
Stock borrowing costs. The availability of bor-
rowing-a stock loan-is important in the extent to
which futures track fair value. If investors cannot
short stocks in the local market (find shares to borrow
to sell to others), they will use futures to execute a
short position. This practice biases futures toward
trading cheap. For example, when index futures
began trading in Germany, it did not have a well-de-
veloped stock-borrowing market, so futures were
consistently cheap. The degree of cheapness has
declined, however, because the availability of loans
on German stocks has improved.
Dividend treatment. The main reasons for the
mispricing of German futures are the dividend treat-
ment for different entities operating in that market
and the cost of difficulty in borrowing stocks. In a
period of heavy dividend paying for German stocks,
futures are cheap.
Taxes and commissions. The tax status of div-
idends and transaction taxes varies across countries
and influences the price of futures relative to stock.
Convergence. Some index futures (U.s.,
Japan, Germany, France) allow for trading at the
contracts' settlement price, while other contracts
(U.K., Hong Kong) have less convergence because
the settlement price is based on an average of prices
over a time interval.
Demandfor liquidity when amajor market trend
emerges. Because futures offer a central market to
trade aggregate equity exposure, they often feel the
most pressure when new information or an imbal-
ance of market views concentrates buying or selling
pressure. This concentration may make the futures
appear mispriced relative to the (less accessible) un-
derlying stock market.
Four other issues should be considered in assess-
ing the usefulness and risks of index futures: how
well the index tracks its benchmark, liquidity, nearby
futures mispricing, and calendar spread mispricing.
Tracking
One consideration is potential return differences
between the futures and the local index. Most insti-
tutions manage a benchmark or active portfolio.
Tracking their benchmark against the local index un-
derlying a futures contract informs them of the range
of under- and overperformance. The range can be
calculated statistically with regression techniques
available through firms such as BARRA or Vestek.
Tracking errors can be positive or negative and
should balance out over long performance periods.
Some of the measures used to assess how well an
index derivative matches a portfolio are correlation
and analyses of annualized tracking error, the num-
ber of issues in common, and market capitalizations
in common. Table 3 shows tracking error statistics
versus benchmarks for several countries with index
futures markets, along with some volume informa-
65
Table 3. Countries with Stock Index Futures Contracts and Hedgable OTe Options and Index Swap Markets
1993 Average
Daily Volume Tracking Error Tracking Error
Country" Futures Index Stocks in Index (US$ Million) vs.FT-AWI vs.MSCI
Australia All ordinary shares 274 181 2.2% 3.1%
Canada TSE35 35 27 3.2 NA
France CAC-40 40 1,302 1.9 1.9
Germany DAX30 30 1,287 1.8 2.2
Hong Kong HangSeng 33 300 2.5 1.8
Japan Nikkei 225 225 8,032 6.9 7.1
TOPIX 1,229 977 2.3 3.1
Netherlands EOE 25 81 5.5 6.8
Spain IBEX35 35 62 1.9 3.3
Sweden OMX 30 25 2.7 NA
Switzerland SMI 23 229 1.8 4.2
United Kingdom FTSE 100 100 1,210 1.0 1.3
SOUfees: Goldman Sachs and BARRA Country Models of Global Equity Models.
Noles: NA = Not available. In addition to the country indexes, there is limited trading in a European index called "Eurotop 100," which lists
futures contracts on the COMEX in New York. All Indexes are market capitalization weighted except for the Nikkei 225, which is price
weighted, and the EOE, which is close to equal dollar weighted.
tion for the first five months of 1993. The benchmark
indexes most widely used for international markets
are those maintained by FT-Actuaries (FT-A) and
Morgan Stanley Capital International. Tracking
error is quoted in annualized standard deviation
units: for example, with a 2.3 percent tracking error
of TOPIX to the FT-A Japan Index, we would expect
the TOPIX return to be within 2.3 percent of the
return of the FT-AJapan in approximately two years
out of three. Most indexes on which global index
futures are based have tracking error of 1-4 percent
with benchmark indexes.
This type of analysis comparing portfolio to local
futures indexes can be done for all the major markets.
Table 3 shows the tracking of the local indexes in
several countries to the two leading international
benchmark indexes, the MSCI EAFE and the FT-A
Europe.
liqUidity
Liquidity is the ability to execute an order within
an expected time frame. One measure of index fu-
tures liquidity is futures volume in relation to the
stock volume in a given time period. When investors
want to adjust a long or short exposure, they want to
implement it in the market with the best liquidity for
that type of trade. Because the futures market is by
definition a centralized place for exchanging index
exposure, in most countries, it has the greatest liquid-
ity.
Nearby Futures Mispricing
The potential for adverse pricing of futures rela-
tive to a benchmark is another consideration. One
66
way to measure this mispricing is the percentage
deviation from fair value for nearby futures con-
tracts. Figure 3 shows the average percentage devi-
ation from fair value for global nearby futures con-
tracts in 1991 and 1992. Mispricing has been declin-
ing over time for most index futures markets. The
Japanese index futures market, for example, which
was persistently trading rich, has moved closer to fair
value, and the German index futures have moved
closer to fair value from persistent cheapness.
Calendar Spread Mispricing
The potential for added costs from rolling posi-
tions at levels that deviate from fair value is another
consideration. Figure 4 shows the percentage devi-
ation from fair value for different calendar spreads
in 1991 and 1992. Here also, the cheap German mar-
ket has moved closer to fair value but continues to
trade most cheaply in the period in which dividends
are not concentrated. In the United Kingdom, a great
many investors are using FTSE futures as a surrogate
for index exposures (in part because the cost of stock
transactions in the United Kingdom is high com-
pared with futures). That market generally trades
slightly rich. In general, the factors affecting calen-
dar spread pricing are very similar to those in overall
futures valuation.
Rolling Futures Positions
Derivatives users can stay in derivatives by rolling
their futures positions; or they can move into stocks
one by one (as the futures are unwinding); or they
can use an EFP (exchange for physical)-an ex-
change of futures for the portfolio, which is a portfo-
Figure 3. Percentage calendar Spread Deviation from Fair Value for Global Nearby
Futures Contracts
0.75 r----------------------------------,
0.50
c
.. 0.25
. ~
:>
C;
"E
Q) a
~
P...
-0.25
..D
-0.50 '---- ----!
S&P OSAKA TOPIX AllOrds FTSE
500
-1992
D 1991
DAX CAC SMI EOE
Key:
S&P 500 - United States
OSAKA - Japan
TOPIX - Japan
AllOrds - Australia
FTSE - United Kingdom
Source: Goldman Sachs.
DAX - Germany
CAC - France
SMI - Switzerland
EOE - Netherlands
lio that closely tracks the futures-to move from
futures to stocks. An EFP is permitted on futures
exchanges when official trading hours are over; the
price is negotiated between the buyer and seller.
EFPs can be done in most global markets, and the
dealers will quote a cost for moving futures into the
portfolio. No one of the three strategies dominates
the other two.
The advantages of replicating index exposure
with futures are low commissions, little market im-
pact, and no disruption of the portfolio. On the
negative side, futures may become inefficient for
maintaining equity exposure during long periods
because commissions, market impact, and rollover
risk are incurred as one expiring contract is replaced
by another. By monitoring the rate of open interest
decay and net demand for long- or short-roll trades,
an investor can increase return in managing the roll.
The near contract goes away, and the next one grad-
ually builds up open interest.
The advantages of moving futures into stocks
one by one are low initial cost to obtain market
exposure and ability to assemble portfolios carefully
without time pressure. The disadvantage is that the
manager still incurs the bid-ask spread of trading
individual stocks.
Moving into stocks through an EFP creates less
market impact than trading individual stocks and
also involves low initial cost to obtain market expo-
sure. The disadvantages of this approach are that the
manager must trade the portfolio at once rather than
spread it over time and that transaction costs are
higher than rolling futures.
Alternative Cash-Equivalent Strategies
Cash held in synthetic index funds can be invested
in short-termsecurities or other market-neutral strat-
egies. Some of the cash-equivalent strategies people
use are as follows:
T-bills. Safest investment but also lowest re-
turns.
Money market fund. Best liquidity; returns
depend on credit risk and average maturity.
Eurodollar certificates of deposit. Highest
money market returns; good liquidity; closest to rate
implied in futures or swaps because they represent
base for dealer funds cost.
67
Figure 4. Percentage Deviation from Fair Value for Different calendar Spreads
0.75,- _
0.50
o
.: 0.25
o
J
1::
OJ
~
d': -0.25
-0.50
...D
-0.75 L- ---l
S&P OSAKA TOPIX AllOrds FTSE
500
-1992
01991
DAX CAC SMI EOE
Source: Goldman Sachs.
Note: Please see Figure 3 for key to exchanges.
Mortgage- or other asset-backed securities.
Floating-rate paper backed by pools of mortgages or
other collateral; returns depend on credit spreads,
prepayment characteristics, and defaults within
pool; low liquidity.
Market-neutral strategies. Hedged positions
in stocks, equity indexes, or fixed-income securities;
basis risk of hedge; low liquidity.
Hedging with Index Options
An investor can use listed or aTC options on a stan-
dardized index or a customized portfolio to manage
equity market risk without removing all equity ex-
posure. Hedging with index options is appropriate
for managers who want some equityexposure but do
not want full equity risk. A hedging or index options
program can be designed to provide asymmetric
participation in the upside or the downside, where
futures are just the equivalent of owning a portfolio.
To implement an index options program success-
fully, the investor must choose an options strategy,
decide on a time horizon, compare listed and aTC
options, and choose an index. Table 4 lists some
index options available in the U.s. market.
aTC options are privately negotiated options
contracts between two parties who agree on the un-
derlying stock portfolio or index as well as on the
strike price, expiration, and exercise style. In choos-
ing between listed and aTC options, the standard-
ization of the listed options, settlement, and clearing
are issues. If the arrangement is something straight-
68
forward and the strike prices come at fairly tight
intervals, most people prefer the liquidity of the
listed market. For a long-term objective, such as
overperformance relative to a benchmark, they must
go to the aTC market, which has more flexibility but
less liquidity for transactions of small to moderate
size. The major differences between the listed and
aTC markets are as follows:
Maturities. Listed options typically have
shorter maturities than aTC options, although some
longer dated options exist. aTC options can vary
between one month and five years.
Trading procedures. For listed options, each
exchange has its own operating procedures (e.g.,
market maker or specialist. aTC options are traded
in dealer markets with traders at major investment
and commercial banks quoting bid and offer prices
by phone.
Settlement. Each exchange with listed op-
tions has its own settlement and clearing procedures,
whereas settlement of aTC options is negotiated by
buyout terms in a contract entered into by the coun-
terparties.
Underlying assets. Listed options are traded
on individual stocks, major stock indexes, and se-
lected industry groups. aTC options can be based
on any stocks, industry portfolios of stocks, and in-
dexes (domestic or international). They can also be
based on the relative value of two stocks or stock
portfolios.
Exercise. Most listed options can be exer-
cised at any time (American), but some only at expi-
Table 4. Usted Index Options
Long- Short-
Symbol Index Exchanges
a
Dated Dated
SPX S&P500 CBOE X X
OEX S&P 100 CBOE X
VLE Value Line PHLX X X
NYA NYSE Composite NYSE X
XMI XMI Major Market AMEX X
JPN Japan AMEX X X
MID S&PMidCap AMEX X
FSX FTSE 100 Share MERC X
EUR Eurotop 100 AMEX X
WSX Wilshire 250 PSE X
RUT Russell 2000 CBOE X
XOC OTC PHLX X
Source: Goldman Sachs.
aAMEX =
CBOE
MERC =
NYSE
PHLX
PSE
American Stock Exchange
Chicago Board of Trade Options Exchange
Chicago Mercantile Exchange
New York Stock Exchange
Philadelphia Stock Exchange
Pacific Stock Exchange
ration (European). Settlement of OTC options is ne-
gotiated by the counterparties. They can be Ameri-
can or European and can be based on the maximum
or average values of the equity during the life of the
options. They can also contain "knockout" features;
that is, if the asset reaches a certain price before
expiration, the option automatically expires.
Options Applications
A review of some applications of options in the U.s.
markets will illustrate how option strategies can be
constructed to be consistent with a particular invest-
ment objective. In Case A, a portfolio manager wants
to reduce S&P 500 exposure with index derivatives
and tries to choose among the alternatives. In Case
B, a portfolio manager wants short-term upside par-
ticipation. In this situation, the manager has a lot of
cash, does not know what stocks to buy, and uses
derivatives to become invested in something closely
resembling what the portfolio already holds. In Case
C, a portfolio manager wants to benefit from a sector
that is over- or underperforming another sector.
Case A
Scenario: Portfolio Manager A has some stocks
and is fairly bullish on the S&P 500 for 1993 (expects
total returns to be 12 to 15 percent) but is concerned
about the risks associated with recent index gains,
the risk of very negative reaction to President
Clinton's policy in the bond market, and the risk of
slipping back into recession. Futures are trading
within 20basis points of fair value, and pricing favors
long futures positions by 20 basis points. Index op-
tions are priced to expect a slightly more volatile year
in 1993 than 1992, but the prices do not reflect the risk
of the sudden 5-10 percent decline the portfolioman-
ager fears.
Discussion: Some hedging strategies the man-
ager might use and the associated payoff profiles are
shown in Exhibit 1. Strategies range from selling
futures, writing calls, buying puts, using a zero-pre-
mium collar (buying a put and selling a call to elim-
inate any up-front cash investment), and using a
zero-premium put spread collar (buying a put but
selling a put farther out of the money because the
manager is willing to add to the portfolio's equity
exposure if the index drops dramatically).
Portfolio Manager A expects a rising market but
is worried about the downside risk. I would rule out
a zero-premium collar because it limits the upside
too much. The zero-premium put spread collar best
fits the profile of what a wary portfolio manager
might choose to do, because it leaves roomfor upside
in the range of performance the investor expects. The
other possibility would be simply buying the put
itself. The choice would depend on the pricing of the
two strategies.
Table 5 is an example of how costs affect a
manager's decision. When choosing hedging strate-
gies, scenario analysis is important, just as it is for
fixed-income investments. Of course, looking at up-
front cost will not help with the deltas, gammas, and
so forth. It indicates what happens for a marginal
change, but it does not indicate what happens in six
months, which might be the horizon. In that time,
changes are not marginal; they are much larger.
As Table 5 indicates, buying puts has no upside
69
Exhibit 1. Most Common Hedging Strategies
------------------------
___S_t_ra.,t_e-",g-=--y 1 1 M_a_i_n_F.,e_a_tu_r_e 1 1 P_a...;y'--o_f_fs 1 I v_ie,w
t t t t
300
Sell Futures
Writing Calls
Buying Puts
Zero-Premium
Collar
Raising Cash
Compensated for Limiting
Upside
Pay for Limiting Downside
Only
Full Index Participation
within a Range
f
-----------
300 / 400 500
/ 105%)
1//
===='
400 500
105%)
Downside Risk High
Upside Potential Low
I
Trading Range Likely
Downside and Upside
Potential Low
Downside Risk High
Upside Potential High
Trading Range Likely
Risk of Downside HIgh
Zero-Premium
Collar with Upside
0.6
"0 1.0
"0 <1i
Q) <1i
....
0.5 ~
0.2 0..
(fl 0..
'0
0.0
(fl
0.0
'0
~
-D.5
~ -D.2
.. 0
<1i '.0
'S; -1.0
<1i
Q) 'S; -D.6
Ci
-1.5
Q)
Ci
-2.0 -1.0
Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1
1991 1992 1993 1991 1992 1993
CAC-40
0.5 r-----------.......--------,
Q2 Q3 Q4 Q1
1992 1993
"0
~
~ 0.1
'0 0.0
8 -D.1
~ ~
~ -D.3
Ci
-D.5 l.-_-'--_.L-_...l.-_--'--_---'----_--'---_--'---'
Q2 Q3 Q4 Q1
1991
~
0.3
Source: Goldman, Sachs & Co.
what overpriced. Anyone who bought futures con-
tracts and then rolled them forward into the next
maturity date paid something extra. In the latter part
of 1992, when the spread was underpriced, the roll
would have helped performance. The percentage
deviations are not very large-only about 5 basis
points-but rolling forward with an underpriced
spread two or three times a year can affect perfor-
mance to a nontrivial extent. For index funds trying
to track performance to within a few basis points,
purchasing underpriced futures contracts can help
performance (and overpriced futures contracts can
detract from it).
The deviation of the calendar spread from fair
value for the Nikkei 22? was rich in the latter part of
1991 and early part of 1992 and then stayed at fair
value during the latter part of 1992. As investors
were rolling their contracts into the next maturity
84
100...---------------------,
Figure 4. S&P500 Futures, Average Percent of
Total Open Interest of Nearby and Next
Maturities over Five Recent Expiration
Cycles
OL--L----l'------.L_---'---_.L------'_-.L_---'---_-'-------'_--.J
45 43 38 32 28 24 19 16 11 8 3 0
Calendar Days to Expiration
-- Next
- - - Nearby
date in the lattE:r part of 1992, they did it at fair value;
prior to that time, long positions rolled forward paid
more than fair value, which would have reduced
performance somewhat.
For the FTSE, the calendar spread has been fairly
rich in the past 18 months, whereas it was cheap in
the latter part of 1991. The calendar spread for the
DAXhas almost always been cheap, which is related
to the fact that brokers have trouble arbitraging back
to fair value because of the difficulty in selling short.
In France, the spread has been rich rather than cheap
for most of the past two years.
When investors get near the end of the maturity
date of the contract and must roll forward into the
next one, the volume of open interest from one con-
tract to the other becomes important. Open interest
represents the volume of contracts outstanding at the
time. The nearby contract-the one maturing-em-
bodies most of the open interest about 45 days to
expiration. The transfer of open interest away from
the nearby contract to the next maturity occurs most
rapidly about 15 days before expiration. Figure 4
shows the average open interest of nearby and next
contracts as a percent of total open interest for the
past five quarterly expiration cycles in the United
States. Very few contracts are held to expiration.
Most of them are transferred to the next maturity by
Options offer another alternative for implementing
asset allocation strategies. They offer asymmetric
payoffs, whereas futures contracts' payoff patterns
are linear and symmetric. A futures contract goes up
and down in the same magnitude as the market,
whereas an option is asymmetric in the way it be-
haves.
Option prices contain an implied volatility pre-
mium, which varies over time by strike price and by
maturity. The implied volatility of an option is an-
other way to describe its pricing. Options are re-
ferred to as cheap or expensive when implied vola-
tilities are lower or higher than historical volatilities.
When investors purchase options, they incur an up-
front cost for automatic participation in the price
change of the underlying index beyond whatever
the options' strike price is. When investors sell op-
tions, they generate incremental returns if the op-
tions expire worthless at the end of their maturity.
Investors who want to use options for asset allo-
cation must make several decisions: (1) choose the
strike price of the option, (2) choose the term to
maturity, and (3) choose the strategy or configura-
tion of put and call options. They cannot choose the
implied volatility, because the market dictates that-
that is, the market dictates option prices. The im-
plied volatilities of options can differ and may cause
investors to choose one strike price or maturity over
another or even force them to rethink their strategy
configuration, depending on how the market prices
the options.
Options can be used for several purposes in asset
allocation, including creating new market exposure,
protecting existing market exposure, and return en-
hancement.
depending on how quickly people are moving from
one expiration date to the next. If people are holding
on to their contracts and trying to ride them to the
end before they roll over to the next contract, typi-
cally, the spread will be underpriced. When the next
futures contract is in great demand by investors with
long positions, the calendar spread sometimes goes
to a premium near expiration.
Options-Related Strategies
.-
/'
/
/
I
I
/
/
/'
/'
/'
80
20
l:: 60
<lJ
~
<lJ
P... 40
Source: Goldman, Sachs & Co.
rolling forward into the next contract expiration.
For the next contract, open interest builds most
rapidly about two weeks before expiration. Much
gaming occurs in the futures market as contracts
approach expiration. The pricing of the spread be-
tween the nearby and next contracts sometimes goes
to a premium and sometimes goes to a discount
Creation of New Market Exposure
Two option strategies used to create new market
exposure are buying calls and buying call spreads.
To overweight a portfolio in stocks temporarily, be-
cause stocks are attractively priced, the investor can
buy call options. If, for example, extra exposure to
the Japanese equity market is desired, the investor
can buy call options on the Nikkei 225. Of course, the
85
Current Value
Source: TSA Capital Management.
buy a call option to create exposure, they participate
if the market goes up, but if the market goes down,
they will not suffer the full decline. The gap between
the option and futures payoff is the cost of the call
option. If an investor's view about the market is
positive but not excessively bullish, he or she may be
better off to use the lower cost call spread rather than
the single call option to create additional exposure to
a particular market.
The effects of the options are somewhat different
if the market moves right after the options are pur-
chased, before the full effect of the time decay in the
option premium is realized. The bottom panel of
Figure 5 illustrates what occurs before option expira-
tion. The valuation curves smooth out if the market
moves right after the investor puts the position on.
In this case, the options expire in 60 days. With the
call option, the investor participates smoothly as the
market moves away from its current security price,
in this case at 100. On the way up, it will always lag
behind the futures contract, but it will not decline as
sharplyif the market goes down right after the option
is put on. It is not a straight line because the natural
exposure of the option changes as the security price
moves. The call spread creates a flatter profile. It
does not participate as much in the full market move
as the single call option and declines less when the
market falls.
Protection of Existing Market Exposure
A second use of options is to protect existing
market exposure in the portfolio. Strategies that are
used for this purpose include buying puts, buying a
put spread, using a zero-premium collar, and using
a zero-premium put spread collar.
Buying puts. Buying puts may be attractive
if the investor believes the market has high downside
risk and is willing to pay for protection. '"{he price of
protection is the price of the put option.
Buying a put spread. Buying a put spread is
similar to buying a call spread. A put option with a
near strike price is purchased, and a put option with
a lower strike price is sold. The put spread can be
used when someone views the market as having
moderate downside risk. In this case, the investor
pays less than for a single put option.
The zero-premium collar. This strategy is con-
structed by selling a call option with enough pre-
mium to fund the purchase of a put option. As long
as an investor believes the market will stay in a
trading range but with high downside risk, this strat-
egy is attractive. Investors get full market participa-
tion within the trading range; if the market runs
outside of that range, the strategy will underperform
on the upside but do better on the downside.
106 104
./
./
./
./
./
.. --
.y.;.: :,. ...-
--....
96
Creating NewMarket Exposure with
Options
Value at Expiration
8r-----------=---------,-.-,
6 i-
-6
-8 '-----_---l__-----L__--L__---l.-_---'-..L-_--.J
94
8,-------------------,
6
-6 ,...
-8L--_--l..-_---.L_---L__L-_--l..-_-L_---.J
94 96 98 100 102 104 106 108
Security Price at Expiration (60 days)
....... Futures Exposure
- - - Call Option
-- Call Spread
investor must pay a premium to participate in the
upside potential, and that premium represents the
cost of the option.
An alternative strategy would be to buy a call
spread to create the Japanese exposure. With a call
spread, investors buy calls with a lower strike price
than the call options they sell. For example, if the
market has only moderate upside potential, an in-
vestor might buy an option with a strike price at
current market levels and sell an option with a higher
strike price-at a strike price above where the market
is expected to be at expiration. Using a call spread,
the investor will pay less for upside participation
because the sale of the call option reduces the net
premium.
The payoff profiles at expiration of these two
strategies, as opposed to buying a futures contract,
are illustrated in the top panel of Figure 5. If invest-
ors buy futures contracts, they will participate to the
full extent of the market increase or decrease. If they
Figure 5.
./
./
2 ./
I': 0 ././
E ... ././
b-2 ------;-. ...:..---./
.E-4-
86
Figure 6. Protecting Existing Market Exposure
Income Enhancement
Options can also be used for income-enhance-
ment purposes. These strategies involve selling vol-
atility or option premium in the marketplace. One
such strategy is a short strangle. An investor, believ-
ing that the market will stay in trading range, sells
both a put and a call that are out of the money. If that
investor is planning to sell market exposure as the
market rises, the short strangle automatically does
that. The market compensates the investor for mak-
ing that decision in advance.
On the downside, selling the put effectively
means investors will buy more market exposure once
the market falls to a certain level. If they think the
market will stay in the trading range and they are
willing to sell off some of their exposure at the top of
the range and buy more exposure at the bottom of
the range, a short strangle uses options to help them
do that automatically. The short strangle has sub-
stantial risk to big moves in the market. To eliminate
this risk, investors will sometimes buy a long stran-
gle with wider strike prices than the short strangle to
provide protection in case of an unusual event in
which the market runs quickly.
The payoff pattern for the short strangle in con-
junction with underlying market exposure is shown
in Figure 7. If investors hold the short strangle to
expiration and the market stays within a trading
range from roughly 96 to 104, extra income is gener-
ated by the amount of the premium received from
selling the put and the call options. If the market
moves immediately after investors sell their short
strangles, they get a smooth curved line. As the
market moves farther from its central position, they
underperform the broad market.
110 98 100 102 104 106 108
Security Price at Expiration (60 days)
96
-6
-8 '-----_--'----_---L-_------'---_----'__-'--_-c-_-'-_----'
94
Value at Expiration
10
/'
8 /'/' ....
/' .
Q) 6 .. _.-
;:l .....---:::
4 ....
19 2 /'''". ....
:::; / .. '
S 0 . ...
b-2 ;.;? .
:::; ./
-4 _ ../
/'
/'
The zero-premium put spread collar. This strat-
egy is constructed by buying a put spread and selling
a call option, as in the zero-premium collar. This
strategy works well if an investor believes the market
will stay in a trading range and not appreciate or
depreciate greatly. It provides higher upside partic-
ipation but restores market exposure after a substan-
tial decline.
The portfolio protection these strategies provide
is diagrammed in Figure 6. The bottompanel shows
the payoff profile of selected strategies at the time
they are initiated. The futures exposure is a straight
45-degree line. The other strategies have exposure
on one or the other side of the futures line. Using
these strategies changes an equity portfolio's beta or
a fixed-income portfolio's duration. The beta or du-
ration of a portfolio is related to the slope of payoff
profiles. The profiles flatten as the market falls and
offers the portfolio protection against negative mar-
ket returns.
.......
96
"7.'
/, .. '
:: ..
-15 '------'----_-'-_-----'-_---'__.L-_...L..._------'---_---'
92 94
-10
10
Q)
;:l
5
19
:::; 0
Q)
Ei
b-5
.5
Figure 7. Selling Volatility for Return Enhancement
15,--------------------,
108
/'
/'
/'
/'
/'
/'
/' .'
/- ---::: ...:-' - . .
6 .. -
98 100 102 104 106
Current Security Price
96
-6
-8 '---_---"-- '-----_-'-__L-_------'---__.L-_----'
94
Current Value
10 ,---------=-==.::..:='--'-'==-----------,
8
Q) 6
;:l
4
2
S 0
Q)
-2
- - - Futures Exposure
- - - - Zero-Premium Collar
-- Put Spread
....... Put Option
-- Futures Exposure
- - - Short Strangle at Expiration
....... Short Strangle without Time Decay
Source: TSA Capital Management. Source: TSA Capital Management.
87
34,------------------,
Source: Goldman, Sachs & Co.
Figure 9. Strike Price Structure of Implied
Volatilities, March 25, 1993
0'--- ---' ---'- --'
0.84 0.92 1.0 1.08
Strike/Index. Puts Calls -----)...
. .
....
---
---
---
29
14
-- Nikkei 225
--- DAX
....... S&P500
.b 24
;.:::
'';::
ell
19
Each option price has an associated implied volatility,
which varies by option maturity, strike price, and
calendar seasonality. Implied volatility tends to be
upward (downward) sloping if recent volatility is
below (above) average volatility; it tends to be gener-
ally downward sloping because of investor demand
for puts to hedge downside risk. Also, some markets
show declining volatility during midyear months.
Figure 8 illustrates the strike price structure of
the S&P 500 implied volatility. When these data
were compiled, the S&P index was selling at about
440. The call options have somewhat lower implied
volatilities as the strike price increases. Put options,
however, have somewhat higher implied volatilities
as the strike price decreases.
Figure 8. Strike Price Structure of S&P500 Implied
Volatility, March 1993 Expiration
Implied Volatility
18,----------------------,
_ __'__.l.___L_...J....__'______'__--'-___l
405 410 415 420 425 430 435 440 445 450 455 460
maturity of the option, the more expensive it is in
both price and implied volatility. That relationship
may be mostly attributable to liquidity pressure. The
most liquid options are those of the closest maturity.
Farther out, the option price may be priced with a
liquidity premium, which is translated into greater
implied volatility.
The seasonality of implied volatility is shown in
Figure 11. During May, June, and July, market vol-
atility has historically trended downward. This sea-
sonality pattern has implications for the pricing of
options, because implied volatilities tend to follow
the same pattern as actual volatility. Options are Calls Puts
-- January 29,1993
- - - December 31, 1992
Figure 10. Tenn Structure of S&P500 Implied
Volatility At-the-Money Options
Source: Goldman, Sachs & Co.
16 ,--------------------,
Source: Goldman, Sachs & Co.
/"
/"
/"
/"
Dec. July
1994
Sept. June April
1993
Contract Expiration
March
10 L-__'____L-_.---JL-_----'__----'__--'
Feb.
15
-- December 31,1992
- - - January 29,1993
.b
S 14
]
12
11
This implied volatility curve changes over time.
It is usually downward sloping. Most of that bias
occurs because investors are naturally long in their
exposure to the marketplace. They often use put
options to hedge on the downside. There is a natural
hedging pressure to buy put options, which creates
more expensive put options in implied volatility
terms. As Figure 9shows, this characteristic is fairly
consistent in international markets; it occurs in the
Nikkei 225, DAX, and S&P 500. The Nikkei 225
bends up somewhat because this volatility profile
was taken just after a market rally in Japan, so the
out-of-the-money calls suddenly became expensive
as the market started to run on the upside.
Figure 10 shows the term structure of S&P 500
implied volatility for 1992 and 1993. The longer the
88
DAX FTSE CAC-40
100
Nikkei
225
S&P
500
-0.6 L------.J ---L .L.-__--l. --L-----l
o
-0.1
'"
-0.2
l::
0
'.0
..Q
-0.3
Ili
....
....
0
-0.4 U
-0.5
Figure 12. Correlation Coefficients between
Weekly Percent Changes in Implied
Volatility and the Market Index
Nov. Dec. Sept. July May March
0.8 '------'------'--------'----'------'--------'
Jan.
0.9
1.1
1.2
1.3 r----------------------,
.9
~
c:><: 1.0
Figure 11. Ratio of S&P500 Volatility by Month to
Overall Average Volatility
(20 years ending May 1992)
Source: Goldman, Sachs &Co.
Note: Average volatility = 13.4 percent.
October 1991 to December 1992
o January 1990 to December 1992
-----'-----
Source: Goldman, Sachs &Co.
generally cheaper during the summer than in the fall
because actual volatility, and consequently implied
volatility, increases somewhat in the fall. This pat-
tern seems to hold true even when data from October
1987 are excluded from the analysis.
Changes in implied volatilities in options are
contemporaneous with changes in actual volatility.
As the market becomes volatile, the implied volatili-
ties in the options rise, although the implied volatil-
ities rise more as the market declines than when it
goes up, which is consistent with the natural dispo-
sition of investors to carry long positions in the mar-
ketplace. When the market falls, many investors
want to buy put options. The put options are bid up
in price, which translates into an expanding implied
volatility.
Current implied volatilityis generally not a good
predictor of future market direction or volatility.
Figure 12 illustrates the correlation between the
change in implied volatility in options and the
change in the index. The two are generally nega-
tively correlated; that is, as the market rises, implied
volatilities come down, and as the market falls, im-
plied volatilities go up. That is true in all five markets
shown in Figure 12.
Figure 13 shows the correlation coefficients be-
tween historical and implied volatility. The two
have been generally positively correlated in the
United States, Japan, and France, although not in
Germany and the United Kingdom. As the market
becomes more volatile, implied volatilities go up.
They move in tandem contemporaneously. If im-
plied volatility is lagged one period, however,
changes in implied volatility and subsequent
changes in market volatility are uncorrelated or neg-
atively correlated (Figure 14).
Implied volatilities tend to exhibit some mean
Figure 13. Correlation Coefficients between Weekly
Changes in Historical and Implied
Volatility
0.35
0.25
I
'" l::
0
0.15
. ~
]
JJ
....
0.05
0
U
0
I
-0.05
-0.15
S&P Nikkei FTSE CAC-40 DAX
500 225 100
October 1991 to December 1992
o January 1990 to December 1992
Source: Goldman, Sachs & Co.
reversion. If option prices are bid up and implied
volatilities are high, they decay back toward a long-
term position. That is uniformly true across all coun-
tries in this sample. If implied volatilities are too low,
they migrate up to the average over time. Historical
volatilities also show some mean reversion.
Figure 15 traces implied volatilities in the five
subject markets during the period from 1992through
the first quarter of 1993. Implied volatilities in Japan
have been in excess of those in the United States,
although they have been subsiding recently. The
patterns in the United Kingdom, France, and Ger-
many are more like those in the United States, al-
though France has experienced some spikes since
third quarter 1992, about the time the EuropeanMon-
etary Union started having problems. The United
States has been through a period of relatively low
89
-0.6 L------.L --'---__-----.L --'---__-----.L--...J
1993
QI Q4
~ '1
J". 1 . / .
. I- . I
.\" 7.\
I . ,I /\ I' \
,I / \ 1\ \
/ I '. r... '\/ -r 1\ /
1 .\Y:\/'v\
. ,/." . \ -.
I:. .... -.I .." .....'r'.
0" ,. 0"
.\
.' . / " r
.: ......... -./ ........ '1.-
0
/',. -.J' ......... '
Q3
.\
r
1992
II'
\: ./
.\
/
Q2
10
5L-__---L. .L-__---'- ..L....__----'
QI
Figure 15. Implied Volatility Comparison, Five
Markets
50,--------------------,
45
40
.35 .\
:;:::
E 30 \.. i '\/
o \:'
:> 25
] 20
0.. r-
Ei 15 . __ 'v'
~ .'
DAX CAC-40 FTSE
100
Nikkei
225
S&P
500
o
-0.1 -
Figure 14. Correlation Coefficients between
Weekly Changes in Lagged Implied
Volatility and Subsequent Changes
<FJ
-0.2 -
c
.g
..$
-0.3-
(l)
....
....
0
U -0.4 -
-0.5-
October 1991 to December 1992
D January 1990 to December 1992
Source: Goldman, Sachs & Co.
implied and historical volatilities.
Table 5provides 1992data on volatilities in these
markets. The range of implied volatilities has been
small in the United States, wide in Japan, and in the
middle for the European countries. The implied vol-
atilities in all markets tend to be higher than histori-
_ .. - Nikkei 225
- - - CAC-40
-- FTSE100
DAX
S&P500
Source: Goldman, Sachs &Co.
ment the change in a cost-effective manner. A vari-
ety of derivative instruments are available not only
in the United States but also in major countries
Table 5. 1992Global Index Volatility Statistics
Standard
Average Implied Deviation of Ratio of
Historical Implied Minus Implied High to
Index Volatility Volatility Historical Volatility Range Low
S&P500 9.4 13.4 4.0 1.6 10.3 to 17.9 1.7
Nikkei 225 28.5 33.0 4.5 7.0 17.0 to 55.2 3.3
FTSE 100 15.2 17.8 2.6 4.1 9.4 to 34.0 3.6
CAC-40 18.6 22.0 3.4 5.8 15.2 to 41.1 2.7
DAX-30 13.8 14.7 0.9 4.6 8.5 to 36.3 4.3
Source: Goldman, Sachs & Co.
Note: Implied volatilities from nearby contracts.
cal volatilities, which favors option sellers or those
creating option positions dynamically.
Conclusion
If investors want to make asset allocation decisions,
they need a view of future market direction. They
must be willing to act on that view and then imple-
around the world. The various strategies used in
asset-allocation-type trades using options and fu-
tures, swaps, and embedded options have a variety
of payoff patterns. The choice of which strategy is
best depends on the investor's view of market direc-
tion, the cost of the strategy, and the risk-return
preferences of the investor.
90
Question and Answer session
Roger G. Clarke
Question: How does the percep-
tion of fair value for index futures
differ for local investors com-
pared with nonresidents, who do
not receive the full dividend?
Clarke: I assume the question is
referring to the wedge that is
sometimes driven in fair value
pricing with regard to the taxa-
tion of dividends. U.s.-based in-
vestors in Germany, for example,
have no choice. If they buy Ger-
man securities, the dividends are
taxed. To deal with that, people
have tried swaps in which the
broker is a local, who tries to arbi-
trage the tax issue and pass those
savings on to investors outside of
the country. As long as this arbi-
trage can take place and be of-
fered to foreign investors, the no-
tion of fair value should be the
same for both German and U.s.
investors.
Question: Why is the term struc-
ture for implied volatility not flat,
given that the option valuation
model has an explicit term allow-
ing for time to maturity?
Clarke: In its simplest form, the
Black-Scholes pricing model has
several parameters, including an
explicit allowance for time to ma-
turity and a volatility assump-
tion. The volatility assumption
does not need to be the same for a
three-month option as for a six-
month option.
Implied volatilities are calcu-
lated by taking the option price as
given and finding the volatility
that generates that option price.
Because of supply and demand
considerations, prices for options
that are way out of the money or
of distant maturities without
much liquidity can be priced
high, which translates into a high
implied volatility number. That
difference cannot generally be ar-
bitraged away, because the put-
call parity arbitrage relationship
is between options of the same
maturity and strike price. It is
not possible to arbitrage the dif-
ferential implied volatility pricing
of an option with three months to
expiration against one with six
months to expiration. The price
of the six-month option will
never be below the three-month
option price, so there are some
bounds on how large the differ-
ence in implied volatility can be,
but historical patterns have gener-
ally been upward sloping.
Question: Given that the Black-
Scholes model has a variable for
term to maturity, why is there
also a term structure in volatility?
Clarke: There is no arbitrage
mechanism to force implied vola-
tilities of long maturity options to
be equal to those of short-dated
options. The lack of a tight arbi-
trage mechanism allows supply
and demand considerations to be
reflected in option prices and sub-
sequent implied volatilities.
More distant maturity options are
less liquid. The market may ex-
tract a somewhat higher pre-
mium to compensate for the lack
of liquidity, which translates into
a higher implied volatility. In ad-
dition, the Black-Scholes model,
which is typically used to calcu-
late implied volatilities, is a sim-
plified model of the real world.
Real-world violations of the sim-
ple assumptions may induce a
bias in implied volatilities that we
interpret as a term structure.
Question: When an investor cal-
culates the number of futures con-
tracts to change the mix in a port-
folio, should the portfolio's beta
be taken into account?
Clarke: Yes. I used a simple ex-
ample that assumed the beta of
the portfolio was the same as the
beta of the futures contract and
the underlying index. The beta of
the S&P 500 is generally consid-
ered to be 1.0. If an investor has a
slightly less risky portfolio with a
beta of 0.9, the investor would
buy only 90 percent of the futures
contracts to hedge instead of buy-
ing the entire 100 contracts in that
example.
91
Using Derivatives to Manage the Currency
Risk in Global Investment Portfolios
David F. DeRosa
Director, Foreign Exchange Trading
Swiss Bank Corporation
Global investors can be at substantial risk if they fail to manage their currencyexposures.
These exposures can be hedged through currency forward contracts, currency futures
contracts, or various options ranging from those on actual currency to exotic options such
as knock-out or barrier options.
This presentation sketches out some of the practical
aspects of managing currency risk.
1
Many people
have argued that, because foreign exchange is a zero-
sum game, investors do not have to worry about
currency risk when managing international equity
portfolios. I disagree. Take a look at Table I, which
shows the annual percentage gains or losses to a U.s.
dollar investor exposed to the British pound, the
German mark and the Japanese yen from the end of
1973 to the end of 1992. It shows that, left un-
managed, currency fluctuations can generate sub-
stantial gains and losses. On balance, those of us who
are dollar-based have had a pretty easy time of it
since 1985 because the dollar has been in a secular
downtrend. Investors on the other side, however,
would have suffered considerably. During 1985,
1986, and 1987, the dollar fell by more than 25 percent
each year! No wonder one of the first questions from
Japanese and European investors about any new
investment strategy is "what are you going to do
about the foreign exchange risk?"
You could argue that the risk might wash out
over time, especiallywithexposure to manydifferent
currencies, but Figure 1 says otherwise. It shows the
history of the USDX FINEX index of the value of the
U.S. dollar. This index closely tracks the behavior of
the currency exposure of the MSCI EAFE index. Fig-
ure 1 shows that the dollar index's fluctuations are
IFor further reading, see David F. De Rosa,"An Option-Based
Approach to Currency Risk Management" in Strategic Currency
Investing, edited by A. Gitlin (Chicago: Probus Publishing, 1993);
and David F. DeRosa, Managing Foreign Exchange Risk (Probus
Publishing 1991) and Options on Foreign Exchange (Probus
Publishing, 1992).
92
large and can run in secular trends for years at a time.
I believe that investors can be at substantial risk if
they do not manage their currency exposures.
Foreign Exchange Trading
Foreign exchange is traded on both a spot and
forward basis. Spot means delivery in two business
days (except for U.s. dollar/ Canadian dollar trades,
which settle in one business day). If you buyGerman
marks and sell U.S. dollars, you must receive the
marks in a German bank and deliver dollars to a U.S.
bank. Forward transactions settle on any date fur-
ther in the future than spot settlement. Most portfo-
lio managers trade spot in order to facilitate the
purchase and sale of foreign securities; they use the
forward market for hedging operations.
Foreign exchange is largely a banker's game.
Trading is conducted 24 hours a day, every day
except weekends and holidays. At the start of the
day, the first market to open is New Zealand, then
Australia, Tokyo, Hong Kong, and Singapore. Next
to open are the central European markets, Zurich,
Frankfurt, and Paris. London is by fat the largest
center for foreign exchange dealing, and New York
is the second largest, but foreign exchange trades in
about a hundred other, smaller money centers. You
can trade virtually any currency in any of the major
money centers, but liquidity is greatest during the
New York morning because London and New York
overlap, timewise. Dealing is done by telephone
and by a private computer network, called "direct
dealing," that Reuters operates. The foreign ex-
change market is fast, and orders of any size can be
Table 1. Annual Spot Exchange Rates on the British Pound Sterling, Gennan Deutsche Mark, and Japanese Yen
and Rates of Return to a Dollar-Based Investor
British Pound German Mark Japanese Yen
Exchange Rate of Exchange Rate of Exchange Rate of
Date Rate Return Rate Return Rate Return
December 31, 1973 2.32 2.70 280.11
December 31, 1974 2.35 1.13% 2.41 12.27% 301.20 -7.00%
December 31, 1975 2.02 -13.90 2.62 -8.09 304.88 -1.20
December 31, 1976 1.70 -15.86 2.36 11.02 293.26 3.96
December 31, 1977 1.91 12.10 2.11 11.92 240.38 21.99
December 31, 1978 2.04 6.80 1.82 15.60 194.55 23.56
December 31, 1979 2.23 9.25 1.73 5.71 239.81 -18.87
December 31, 1980 2.38 7.10 1.96 -11.94 202.43 18.47
December 31,1981 1.91 -19.90 2.25 -12.73 219.78 -7.89
December 31, 1982 1.62 -15.24 2.38 -5.65 234.74 --{'.37
December 31, 1983 1.45 -10.35 2.72 -12.56 231.48 1.41
December 31, 1984 1.16 -20.17 3.15 -13.69 251.26 -7.87
December 31, 1985 1.45 24.74 2.45 28.94 200.40 25.38
December 31, 1986 1.48 2.46 1.95 25.50 158.48 26.45
December 31, 1987 1.88 26.94 1.57 24.14 121.07 30.90
December 31, 1988 1.81 -3.75 1.78 -11.56 125.00 -3.15
December 31, 1989 1.61 -10.86 1.69 4.98 143.68 -13.00
December 31, 1990 1.93 19.62 1.50 12.96 135.69 5.89
December 31, 1991 1.87 -3.24 1.52 -1.59 124.69 8.82
December 31, 1992 1.52 -18.85 1.62 -5.98 124.84 -D.12
Average Return -1.16 3.64 5.33
Standard deviation 14.72 13.80 14.54
Average absolute value 12.75 12.47 12.23
Source: Swiss Bank Corporation.
lars at 120.00 yen and buy dollars at 120.05 yen.
Forward foreign exchange is quoted in terms of
points that are added or subtracted from spot levels.
Each settlement date has its own points, and you can
choose any future date that is not a bank holiday.
The resultant "outright" rate is the exchange rate at
which you can trade foreign exchange for settlement
beyond spot delivery. Like spot, forward points are
quoted as two-way, bid and offer prices.
Why do the forward points exist? The answer is
that the market prevents a riskless conversion to a
higher interest rate currency. In other words, if
deutsche mark interest rates are higher than dollar
rates, you cannot convert dollars to marks and then
hedge back in the forward market at the same price.
The equalizer-or perhaps the spoiler-is the for-
ward points. In formal theory, this relationship is
called the covered interest parity theorem.
The simplest form of hedging of international
portfolios is done in the forward market. For exam-
ple, a manager who is holding Japanese equities and
takes the view that the yen is going to fall against the
dollar might sell yen forward for three months
against dollars. As the three-month settlement date
approaches, provided his view on the yen has not
changed, the manager could roll the forward contract
to avoid having to deliver yen and receive dollars.
Of course, the accrued gains or losses on the forward
'91 '92 '87 '83 '79 '75
60 L.--__---'-__---' -'---__---L '----..J
'71
80
100
Source: FINEX.
120
140
160
180,---------------------
done. It is far and away the largest capital market.
By ancient convention, most currencies are
quoted "European style," meaning units of foreign
currency per U.s. dollar (e.g., 120 yen per dollar or
1.65 marks per dollar), but the British pound, the
Australian dollar, the New Zealand dollar, and the
ECU are quoted"American style," meaning dollar
per unit of currency. Spot foreign exchange is quoted
as a two-way price. Dollar-yen quoted by a dealer
at 120.00-120.05 means that the investor can sell dol-
Figure 1. FINEX U.S. Dollar Index
93
hedge would have to be settled with the counterparty
bank. Note that no cash exchanges hands until the
settlement date. For this reason, investors must have
obtained credit approval from their dealing bank
before they can do forward contracts.
Hedging can also be done using currency futures
contracts. Currency futures are listed on the Chicago
Mercantile Exchange's (CME's) International Mone-
tary Market with duplicate listing on the Singapore
International Monetary Exchange. Trading in these
contracts is active, but it is considerably less than the
enormous volume each day in the cash, interbank
market. The CME's clearinghouse sets initial margin
requirements and pays or collects variation margin
each day, based on exchange rate movements.
Options on Foreign Exchange
The currency option market is the world's largest
option market. This market is mainly interbank and,
therefore, is an over-the-counter market. In the inter-
bank market, currency options trade in parallel with
spot and forward foreign exchange. Virtually all of
this trading is in standard, European options, mean-
ing no exercise until expiration. Parenthetically, the
fastest growing portion of the market is trading in
exotic options, especially knock-outs or barrier op-
tions.
There are two listed markets for currency op-
tions. The Philadelphia Stock Exchange lists options
on actual foreign currency, and the CME lists options
on its own currency futures contracts. Currency fu-
tures options deliver a currency futures contract
upon exercise.
Most use the Garman and
Kohlhagen adaptation of the Black-Scholes
3
model
for standard European currency options. This ap-
proach lets the foreign currency interest rate stand in
for the continuous dividend in Merton's version of
the model.
4
All of the favorite "greeks" are there,
delta (sensitivity to the spot exchange rate), gamma
(sensitivity of delta to the spot exchange rate), theta
(sensitivity to the passage of time), and tau (the sen-
sitivity to changes in implied volatility). In the case
of currency futures options, most people use Black's
model as a first approximation.
5
Futures options are
2Mark B. Garman and Steven W. Kohlhagen, "Foreign
CurrencyOptionValues," Journal ofInternational Money and Finance
(December 1983):231-37.
3Fischer Black and Myron Scholes, "The Pricing of Options
and Corporate Liabilities," Journal of Political Economy (May/June
1973):637-59.
4Robert C. Merton, "Theory of Rational Option Pricing," Bell
Journal of Economics and Management Science (Spring 1973):141-83.
94
American inexercise convention, so ifmore precision
is needed, binomial models are used.
Applications
As I mentioned, the U.s. dollar has been in a secular
downturn since 1985. I predict that some day, this
will end, and when it does, dollar-based investors
will develop a tremendous interest in currency hedg-
ing. So I will now turn to some practical applications
of exchange rate risk management.
ASimple Forward Hedge
If you wanted to hedge the exposure to a single
foreign currency, you could sell that currency for-
ward against your home-base currency. When only
two business days remain to settlement, you must
roll the position to establish a new forward hedge,
provided that you still want the hedge. This proce-
dure sounds simple enough, but there are two possi-
ble complications.
The first complication comes from the forward
points. Currently, the U.s. dollar has the lowest
interest rate among the major currencies. The inter-
est parity theorem tells you that the interest rate
spread must be impounded in the forward points.
When you sell forward a "discount" currency (mean-
ing a currency with a higher interest rate than your
own), you pay forward points. For example, Table
2 displays the MSCI EAFE index exposures as of
December 31,1992. Currently, the weighted average
interest spread of this index is more than 300 basis
points. Hedging the EAFE currency exposure can be
expensive when home-country interest rates are low.
Of course, the opposite is true in hedging a premium
currency (one that has a lower interest rate than the
home-country currency). In this case, the forward
points are paid to you because you are foregoing a
higher home-currency interest rate by investing
abroad.
The second complication is that the hedging pro-
gram will produce realized currency gains and losses
whereas the underlying portfolio will produce equal
but opposite accrued currency losses and gains. Re-
cently, a major pension fund abandoned its currency
hedging program because it had to stop its losses as
the dollar fell against European currencies. In real-
ity, this situation was now a form of "money illu-
sion," because true economic losses were not possi-
ble-they were hedged, after all. Perhaps what both-
ered the fund managers was the obvious, up-front
realized losses in the currency hedge program, in
spite of the fact that the accrued currency gains on
5Fischer Black, "The Pricing of Commodities Contracts,"
Journal ofFinancial Economics (January/March 1976):167-79.
Table 2. Currency Exposure of MSCI EAFE Index as
of December 31, 1992
------ - - - - - ' - - - - - - - - - - - - ~
the underlying equity securities portfolio were
equally large. I wonder if the managers would have
considered the program a "success" if the dollar had
risen. If it had gone the other way, would they have
counted the hedging profits without considering the
foreign exchange losses that had accrued against the
equity portfolio?
Basket Hedges
In the case of a multicurrency exposure, you first
must make a decision as to whether you want to
manage the currency risk of the benchmark index or
of the actual portfolio. In practice, active equity man-
agers usually have some index such as EAFE or the
FT-Actuaries index as their benchmark. If managers
hedge the benchmark, they are implicitly saying that
they are responsible only for currency risk arising
from its deviations from the index exposure, not for
the currency risk of the benchmark index itself. If a
manager takes an exposure of 10 percent in Hong
Kong, whose weight in the benchmark index is ap-
proximately 3 percent, then he or she is at risk for the
7 percent exposure to the Hong Kong dollar. If the
manager subsequently were to reduce Hong Kong's
exposure in the equity portfolio to zero, he or she
would be responsible for taking an implied short
position (minus 3 percent) in the Hong Kong dollar.
On the other hand, the manager may decide to hedge
whatever actual exposure is assumed. In this case,
the manager has a free hand to pick country
weightings without heed to currency risk.
If a portfolio has a great number of currency
exposures, hedging each one separately will not be
Country
Japan
United Kingdom
Germany
France
Switzerland
Netherlands
Hong Kong
Australia
Spain
Italy
Sweden
Singapore
Belgium
Denmark
Austria
Norway
New Zealand
Finland
Total
Source: Swiss Bank Corporation.
Exposure
43.16%
19.72
6.60
6.73
4.44
3.33
3.01
2.84
1.96
1.93
1.41
1.47
1.27
0.69
0.49
0.36
0.33
0.26
100.00
economical. This problem is why practitioners often
resort to using basket hedges-direct applications of
Markowitz portfolio theory. The first step is to mea-
sure the currency exposure that is to be managed.
Next, use a quadratic optimizer to solve for an opti-
mal basket of currencies, as shown in the first column
of Table 3 for the EAFE index. Because most inter-
national portfolios are usually top-heavy in the four
or five largest equity markets, a small number of
currencies can be used effectively to hedge the whole
exposure. In Table 3, the yen accounts for 46 percent
of the basket, the pound for about 22 percent, the
mark for about 17 percent, and the franc for about 9
percent. More currencies could have been added to
the basket to reduce the projected tracking error, but
the transaction costs of executing and maintaining
the basket would have risen.
The forecast tracking error in this example is only
83 basis points a year, an amount that would proba-
bly be acceptable to most investors. This is an ex ante
forecast, however. The ex post tracking error-the
actual error-might be quite different. One reason
the forecast of tracking error could be biased is that
the variance-covariance matrix of currency returns,
the key ingredient fed into the optimizer, could be
wrong. Another problem is that one or more of the
currencies not included in the basket might exhibit
erratic behavior. If this is a concern, you could make
a hybrid basket hedge. First, make one-off matched
hedges for the problematic currencies and then pro-
ceed to construct a basket hedge for the remaining
currency exposures. The procedure then is just a
repeat of the forward hedge: Hedge each basket cur-
rency by selling a series of rolling forward contracts.
Multicurrency Option Hedges
The currency basket approach can also be used
to implement an option hedge. A simple strategy is
to buy put options on the currencies in the basket.
This strategy would provide downside protection
but allow for the possibility of true upside apprecia-
tion were the value of the foreign currencies to rise.
The investor must choose where to strike the put
options, a decision that determines the degree of
downside protection. Complete protection would
be achieved with a floor of 100 percent. If the floor is
set at 95 percent, the largest possible loss to currency
depreciation is 5 percent for one year. Naturally, the
higher the floor, the more expensive will be the bas-
ket of puts. This choice is not different from setting
the deductible loss limit on a casualty insurance pol-
icy.
The total cost of the basket of at-the-money puts
shown in Table 3 is 4.83 percent of the underlying
portfolio. Some of this cost is forward points. The
95
Table 3. O ~ Y e a r At-the-Money Option Basket Hedge
Optimal U.s. Strike
Currency Foreign Dollar asa
Basket
a
Implied Interest Interest Percent
Currency (Weight) Volatility Rate Rate of Spot
Japan 46.13% 9.10% 3.31% 100.00%
United
Kingdom 21.73 13.80 5.62 100.00
Germany 17.42 12.80 7.45 100.00
France 9.15 12.80 9.57 100.00
Unhedged
(US$) 5.57 3.68% 100.00
Total/
average 100.00 11.22 4.92 100.00
Source: Swiss Bank Corporation.
aExpected annual tracking error of 0.83 percent.
Cost as a
Percent of
Spot
3.33%
6.24
6.86
8.14
4.83
weighted average foreign interest rate is4.92 percent,
whereas the US. dollar rate is only 3.68 percent,
making the spread 1.24 percent. Thus, the true op-
tion cost is only 3.59 percent for an at-the-money
hedge. This cost is a function of the maturity of the
option and the prevailing levels of implied volatili-
ties.
At this point, you would have achieved a form
of currency protection. The basket of put options
would protect you against declines in foreign curren-
cies and, in the opposite case, you would profit from
a rise in their value. The story does not have to end
here, however. Over time, you can get a "refund" of
sorts on part or all of the cost of the basket by moni-
toring the cross-rates. When cross-rates, such as ster-
ling-mark and mark-yen, move significantly, then
some cheaper option basket will deliver the same,
identical downside protection. This effect is attribut-
able to the gamma of the options. When a cross-rate
moves, one dollar exchange rate rises and the other
falls. The delta on the put on the rising exchange rate
must go up, and the delta on the put on the falling
96
exchange rate must fall. The trick is to understand
that, in absolute value terms, the delta that rises
changes by more than the delta that falls, this being
a consequence of the convexity, or gamma, of the put
option. Therefore, the basket can be rearranged, and
some options can be liquidated without loss of pro-
tection.
Another potential cost saver is to attempt to
replicate the put options rather than buy them. Op-
tion replication functions by buying and selling for-
ward contracts in amounts that correspond to the
delta of the target option. Dynamic replication can
be very profitable, but it is a full-time job. A lot of
things can go wrong because exchange rates have a
habit of not moving for a long period of time but then
suddenly lurching in one direction or another. So
unless you are willing to sit in front of a lot of screens
and have associates doing the same thing in the other
two trading time zones, I question whether you will
have a good experience. Probably you will do better
just buying the options.
Question and Answer Session
David F. DeRosa
Question: Do the forward for-
eign exchange markets have any
regular anomalies such as the for-
ward rate being skewed resulting
in positive arbitrage?
DeRosa: I have heard this argu-
ment before, and some fairly sub-
stantial people have offered em-
pirical evidence to this effect. My
problem is that the only argu-
ments that have been set forth to
explain such a phenomenon are
based on irrational investor be-
havior. My early training as an
economist makes me suspicious
but I will keep an open mind on
the issue.
Question: Does the forward
rate represent interest variance or
anticipation of future spot prices?
DeRosa: I take your phrase "in-
terest variance" to mean the inter-
est rate spread between the two
currencies. Interest rate parity
tells us that the forward exchange
rate is solely a function of the dif-
ference between the interest rates
on the two currencies. Consider
the following transaction, which
consists of three parts: (1) Sell the
base currency, and buy foreign
currency spot; (2) invest in a for-
eign currency bond; (3) sell the fu-
ture value of the foreign currency
bond forward to arrive back in
the base currency. The net return
must be equal to the interest rate
that prevails in the base currency,
or riskless arbitrage would be
profitable. This no-arbitrage con-
dition forces the forward rate to
deviate from the spot rate by the
interest rate differential between
the two currencies.
The second concept you men-
tioned is more difficult. Some the-
oreticians have advanced the no-
tion that the forward exchange
rate is the market's anticipation
of the future spot exchange rate.
There is a similar idea in the liter-
ature on the term structure of in-
terest rates, namely that the for-
ward interest rate is an estimate
of future spot interest rates. How-
ever fine this may be on theoreti-
cal grounds, it has little empirical
support, at least in the case of ex-
change rates.
Question: How would you in-
corporate into your model the un-
derlying foreign earnings in U.s.
dollars of a Japanese stock in
your portfolio?
DeRosa: I would ignore them.
All that matters to the overlay
manager is whether the share
price is denominated in dollars or
yen. This is not to say that the
currency risk of the company, in
its balance sheet and income state-
ment, is not a matter of concern
to the investment manager of the
underlying portfolio.
Question: Are you able to calcu-
late the cross-currency implied
volatility for a portfolio of curren-
cies?
DeRosa: Cross-rate options are
readily quoted in the market, so
you can know their levels of im-
plied volatilities. Otherwise, you
might look into some of the work
that has been done on GARCH
models.
97
Using Derivatives to Manage Currency Risk
Murali Ramaswami
Vice President
Salomon Brothers, Inc.
Options can be used to exploit the presence of trends in currency returns and to control
risk. Suitable strategies include rolling simple options and quantity-adjusting options,
a dynamic, formula-based hedging strategy. Using synthetic options should produce
incremental returns.
Derivatives can be used to manage currencies as well
as the traditional assets. This presentation provides
some results of my study on active currency manage-
ment. In that study, I examined the historical perfor-
mance of seven major currency markets with the
objective of understanding their distributional char-
acteristics. I also examined the success of predictive
models of the currency markets.
1
Characteristics of Currency Returns
On a return-to-risk basis, equities would have been a
superior asset class to currencies during the 1980s.
Table 1 shows annual returns on seven currencies for
the 1980-91 period. The "Fxlndx" column is a com-
posite index I created of returns on these seven cur-
rencies. For this study period, FxIndx had a positive
mean return of 2.25 percent, compared with 12.65
percent for the equity market; the risk levels of the two
are virtually the same. The difference in return sug-
gests the possibilityof uncompensated risk in owning
currency assets as opposed to domestic equities.
The distributional characteristics of weekly for-
eign exchange returns are shown in Table 2. The
positive kurtosis values indicate a fat tail to this
distribution-that is, the center is flatter than that of
a normal curve-which implies trends. The degree
of kurtosis declines as the return interval increases
from daily to weekly. During this study period, most
individual currencies exhibited even stronger trends
in their returns than the index. The series also show
evidence of positive skewness, which indicates that
IThis presentation is based on the Research Foundation
monograph: Murali Ramaswami, Active Currency Management
(Charlottesville, Va.: The Research Foundation of the Institute of
Chartered Financial Analysts, 1993).
98
large increases in returns are more probable than
large declines. Again, the means are nonzero, but the
returns are statistically insignificant.
Figure 1 shows how the empirical distribution of
weekly currency returns stacks up against the nor-
mal distribution and the distribution of equity mar-
ket returns. The figure makes apparent some of the
conclusions implicit in Table 2. The fatter tail for
currencies means we should expect certain trends in
the currency market relative to a normal distribution
and relative to the equity market distribution.
The distributional characteristics of the returns
to the seven currencies suggest the existence of
trends. Analysis of the autocorrelations in currency
returns data can be used to confirm or negate the
existence of such trends. Autocorrelation coeffi-
cients for daily currency returns are presented in
Table 3. During the study period, the deutsche
mark, French franc, and Swiss franc showed signifi-
cant one-day lagged negative autocorrelations, or
reversals. Unlike the other European currencies, the
pound showed no significant autocorrelations, nor
did the currency index. The S&P 500 showed a sig-
nificant positive correlation on a daily basis lagged
one period. It had reversals for Lag 2, Lag 3, and Lag
4, however.
The horizons over which significant trends or
return reversals occur can be estimated through vari-
ance ratio tests. The variance ratio is:
VR(K) = Var(Rh /K
Var(Rt) ,
where
k-l
Rf = L Rr-i, and
;=0
Rt = Return in month t.
Table 1. Annual Returns to Currencies and Market Indexes, 1 ~ 9 1
Year DM FFr SFr A$ C$ FxIndx
a
S&P500
a
1980 16.24% -13.52% 7.93% -11.92% -11.24% 5.68% -2.20% 2.91% 25.77%
1981 -7.57 -12.47 -22.34 -22.63 -D.24 -3.81 0.75 -11.65 -9.73
1982 ---6.29 -5.97 -16.71 -16.95 -11.63 -13.89 -3.60 -10.88 14.76
1983 0.77 -14.37 -11.77 -21.96 -8.84 -8.72 -1.27 -7.25 17.27
1984 -8.02 -14.11 -21.61 -13.84 -16.57 -8.63 -5.98 -11.07 1.40
1985 22.61 25.36 22.04 24.48 21.93 -19.10 -5.62 21.85 26.33
1986 22.98 23.62 2.16 15.64 24.22 -2.61 1.35 16.91 14.62
1987 27.04 20.21 23.67 18.48 24.30 8.20 5.78 27.05 2.03
1988 -2.17 -11.12 -2.75 -11.92 -15.72 17.89 8.94 -3.971 2.40
1989 -14.85 4.53 -11.71 4.47 -3.35 -9.05 2.74 -9.48 27.25
1990 5.90 12.03 17.59 12.47 19.02 -2.20 -D.17 9.26 ---6.56
1991 8.28 -1.49 -3.16 -1.88 ---6.03 -1.74 0.17 3.30 26.31
Mean 5.41 1.06 -1.39 -2.13 1.32 -3.16 0.07 2.25 12.65
Std.dev. 14.06 15.57 16.34 16.70 16.26 10.08 4.37 13.73 13.14
Max. 27.04 25.36 23.67 24.48 24.30 17.89 8.94 27.05 27.25
Min. -14.85 -14.37 -22.34 -22.63 -16.57 -19.10 -5.98 -11.65 -9.73
Source: Interactive Data Corp.
aFxIndx is a composite equity-markets-weighted currency index. Data for the indexes cover the 1980-91 period.
Key: Japanese yen
British pound
DM = German deutsche mark
FFr French franc
SFr Swiss franc
A$ Australian dollar
C$ Canadian dollar
Table 2. Distributional Characteristics of Weekly Foreign Exchange Returns, 1978-91
Characteristic DM FFr SFr A$ C$ FxIndx S&P500
Observations 736 736 736 736 736 736 735 624 624
Mean percent 0.086 0.040 -D.005 -0.018 0.053 -D.057 -D.007 0.026 0.207
Standard
deviation 1.529 1.578 1.595 1.586 1.783 1.311 0.608 1.305 2.201
Annual std. dev. 11.027 711.377 11.500 11.440 12.858 9.456 4.386 9.412 15.873
Skewness
a
0.533 0.114 0.054 0.185 0.180 -1.910 0.096 0.316 -D.526
Kurtosis
a
1.372 0.766 1.880 2.407 0.622 12.344 2.310 0.774 3.073
t-statistic
for mean 1.526 0.681 -0.083 -D.315 0.806 -1.183 -D.307 0.499 2.346**
Source: Murali Ramaswami, Active Currency Management (Charlottesville, Va.: The Research Foundation of the Institute of Chartered
Financial Analysts, 1993).
aThe standard errors for skewness and kurtosis for 1978 to 1991 are 0.09 and 0.181, respectively; for 1978 to 1984, they are 0.128 and 0.255,
respectively; for 1985 to 1991, they are 0.128 and 0.256, respectively.
** Significant at the 99 percent level.
Table 3. Autocorrelation Coefficients, Daily Returns, 1 9 7 ~ 9 1
Lags DM FFr SFr A$ C$ FxIndx S&P500
Lag 1 0.009 -D.002 -D.113* -D.082* -D.077* 0.001 -D.057 0.005 0.055*
Lag 2 0.014 -0.004 0.013 0.007 0.016 -D.032 -D.012 -D.008 -D.040*
Lag 3 0.046* 0.010 0.040* 0.040* 0.011 -D.006 0.022 0.046 -D.Q28
Lag 4 0.008 -D.002 0.002 0.025 0.017 -D.009 0.013 -D.005 -D.044*
LagS 0.004 0.015 -D.022 -D.026 -D.004 0.Q11 -D.003 0.021 0.053*
Q-statistic
a
95.612 20.048 100.00 100.0 99.99 78.304 95.97 87.37 100.00
Source: Ramaswami, Active Currency Management.
aThe level of confidence for the Box-Pierce Q-statistic of order 10 (or 10 lags).
*Significant at the 95 percent level.
99
30
Normalized Weekly Returns
Source: Ramaswami, Active Currency Management.
A survey of the literature on forecasting currency
returns is very discouraging; nobody concludes that
Poterba
2
and others found in the context of the equity
market.
Using overlapping periods in the variance ratio
analysis may make the standard statistical tests of
significance inappropriate. Therefore, for each cur-
rency and for each of the two indexes, Monte Carlo
tests were used to assess the significance of these
variance ratios; in these tests, the weekly returns
were rearranged randomly and a new set of ratios
was computed. One thousand trials were performed
for each currency and for each holding period.
Figure 2 illustrates the bootstrap probabilities of
the variance ratios for the deutsche mark, the yen,
and the pound at various investment horizons. The
bootstrap probabilities are similar to the levels of
significance in hypothesis testing except that in this
case, the smaller the number the greater the signifi-
cance. For example, at a six-month horizon, the
probability that the variance ratio for the deutsche
mark would have exceeded 1.39 (its variance ratio in
Table 4) is about 7.3 percent. Clearly, the variance
ratio was unlikely to have been obtained through
sheer chance. Beyond 18 months, for deutsche
marks, the bootstrap probabilities are even smaller
and the variance ratios are statistically significant. In
contrast, the bootstrap probability for the deutsche
mark at 3 months is relatively large and the variance
ratio is not statistically significant. The probability
of obtaining the variance ratios I got was almost 22
percent, so I cannot place too much confidence in the
3-month ratio. Beyond a 6- to 8-month investment
horizon for all currencies, the variance ratios-even
under this bootstrap probability scheme-become
statistically significant.
2James Poterba and Lawrence Summers, "Mean Reversion in
Stock Prices: Evidence and Implications," Journal of Financial
Economics (October 1988):27-59.
Forecasting Expected Returns
20 o
OL-_..L-.='<1::::.L_-..L_----'__
-4.0 -3.0 -2.0 -1.0
25
FxIndx
S&P500
Normal
5
The test is designed to determine whether the
series is a random walk over a period of time. For
six-month returns, for example, the variance of a six-
month buy-and-hold return series is computed and
then divided by the variance of the monthly returns.
If this ratio is 1, the series is a randomwalk. If the ratio
is larger than 1, the six-month return has a larger
variance than the one-month return, which would be
indicative of trends. If the market has reversals,
rather than trends, the ratio should be less than 1.
Variance ratios estimated for various holding
periods for eachof the sevencurrencies are presented
in Table 4. The fact that all the variance ratios for all
of the currencies except the Canadian dollar are
greater than 1 suggests trends exist in weekly re-
turns. The S&P 500 returns had trends up to a 12-
month investment horizon, and then they revert.
The variance ratios for all of the currencies and for
the currency index were highly significant. These
findings are not contrary to what Summers and
20 /.
h
/:
.9
I :
"5
15 I
:9
,
.b
Ii if>
10
Q
J
;;
Figure 1. Distribution of Weekly Returns, Currency
Index and S&P500, 1978-91 Period
Table 4. Variance Ratios, Weekly Returns, Period
Holding
C$ FxIndx S&P500 Period DM FFr SFr A$
3 months 1.16 1.12 1.07 1.06 1.11 1.16 0.84 1.12 1.08
6 months 1.56 1.35 1.39 1.45 1.35 1.16 0.87 1.46 1.16
9 months 1.49 1.28 1.41 1.48 1.33 1.02 0.80 1.42 1.11
12 months 1.70 1.42 1.75 1.83 1.48 1.15 1.00 1.76 1.13
18 months 1.88 1.43 2.19 2.21 1.71 1.12 1.23 2.12 0.99
24 months 1.98 1.45 2.39 2.42 1.82 1.21 1.47 2.36 0.74
36 months 2.46 1.48 2.46 2.53 1.77 1.18 1.88 2.74 0.56
Source: Ramaswami, Active Currency Management.
100
Figure 2. Bootstrap Probabilities for Variance Ratios, Weekly Returns, 1978-91
36 24 18 12 9 6 3
25 . - - - - - - - - - - - - - - - - - - - - - - - - - - - - ~
E
~
.g 20
P:
0-
~
2) 15
o
o
I:Q
"0
Q)
~ 10
I:Q
6
~
1j 5
l'::
.$
~
o
Investment Horizon (months)
W;j Yen
Pound
D Deutsche mark
Source: Ramaswami, Active Currency Management.
Source: Ramaswami, Active Currency Management.
Figure 3. Cumulative Daily Return before and after
Significant Positive Event
Days Surrounding Event
--FxIndx
- - - S&P500
15 11 7
-
3 o
/-----/---....._/
!
I
I
I
-3 -11 -7
return was 1.63 percent for the currency index and
2.79 percent for the equity index. If they both had
been perfectly normal distributions, the event day
average excess return would have been 2 percent for
both indexes.
Market efficiency implies that the cumulative
excess return is insignificantly different from zero up
to the event day and thereafter stabilizes at a higher
(or lower) level of cumulative excess return. By this
criterion, the currency index return is remarkably
efficient, but for the equity market, a dramatic in-
crease of two standard deviations in market returns
appears to be preceded by equally dramatic cumula-
tive declines in the prior 15 days. In other words, for
3.-----------,----------------,
~
E: 2
B 1
~
gs 0 r----.-. ~ _:--
~ -1 ''_ ~
w "-
Q) -2 ,
:> ,
. ~ -3 "-
]-4
::l
U -5 L-._..-L.._--'__--l.-__~ ' _ _ _ . . . . L _ _ ___I._ ____'
-15
Reactions to Unusual Events
Although currency returns may be unforecast-
able in a statistical sense, determinations could be
made as to whether reactions to unusual events
(large positive or negative currency moves) were
predictable and whether the adjustments from such
events lasted long enough to formulate profitable
investment strategies. An "unusual event" is de-
fined as a daily return that is two standard deviations
above or below the mean daily return. The question
is whether the markets are efficient before and after
an unusual event. If after a significant rise (fall) the
currency continues to rise (fall) or revert over several
days, then profitable trading rules canbe formulated.
Figure 3 shows the average cumulative excess
return (i.e., more than two standard deviations in
excess of the meandaily return) accrued from 15days
prior to the event day (Day 0) and continued for 15
days after the event. The mean event day excess
3Robert E. Cumby and John Huizinga, "The Predictability of
Real Exchange Rate Changes in the Short and Long Run," National
Bureau of Economic Research Working Paper 3468 (1990).
currency returns are forecastable. These studies look
at various forecasting models that use macroeco-
nomic variables-GNP data, production data, infla-
tion, money flows, current account, and so forth-
and conclude that currency returns are difficult to
forecast. One exception is a paper by Cumby and
Huizinga.
3
This study shows that one can forecast
real exchange returns.
101
the equity market, although large increases in daily
returns are caused by reversals from a prior period's
declines, in the currency market, large market moves
appear to be unanticipated.
Figure 4 illustrates what happened during a neg-
ative two-standard-deviation event. As in a large
increase in daily return event, the currency market
appears efficient in that the large negative return
event day is unanticipated and the daily excess re-
turns after the event are almost zero. The equity
market appeared to anticipate the impending large
decline in daily return about six days before the event
day but did not fully assimilate the information. The
individual currencies, as before, were efficient with
respect to event days with large declines in returns.
The declining market is not as inefficient as the rising
market.
Figure 4. Cumulative Daily Return before and after
Significant Negative Event
~
0
~ " ' C
'\
I::: -1
.... "-
E "
Q) -2
"- -
\
U)
-3
U)
\ Q)
u
-4
I
><
u:I
\
Q)
-5
:>
\
'.c
OJ
-6 "-
........... -./''' ........ _'\.
"5
/' "-
S
-7
" - ~
;:l
U
-15 -11 -7 -3 0 3 7 11 15
Days Surrounding Event
--Fxlndx
- - - S&P500
Source: Ramaswami, Active Currency Management,
The currency market would not have provided
an opportunity to forecast this significant event. Cu-
mulative excess return was hovering around zero,
and that is what it should be with an efficient market.
For the stock market, once the return had gone down
by two standard deviations, it stayed stable. The
equity market showed some predictability for about
10 days before the event, and then it stabilized.
Existence of Linear and Nonlinear Trends
Although the currency markets appear to be ef-
ficient with respect to unusual event days, the daily
returns exhibited trends and reversals. The excess
return analysis is in the context of linear dependence.
It considers only two points at a time: today's return
and some previous day's return. The two sets of
returns are linearly uncorrelated, meaning today's
return is uncorrelated with yesterday's return or
102
with the day before yesterday's return. Perhaps mul-
tiple correlation is appropriate-today's return is
jointly correlated with yesterday'S and day before
yesterday'S return.
Some new tests of nonlinear dependency have
been devised to find out whether returns are corre-
lated two or more at a time. One is called a BDS
statistic after its developers, Brock, Dechert, and
Scheinkman.
4
In a linear analysis, you are looking at
the distance between two returns-today's return
and yesterday's return-and counting the number of
times these two returns are less than a proximity
distance,I. If you have three or four returns, you look
ata vectorofreturns-today's, yesterday'S, two days
before today, and three days before today. You take
every four-dimensional vector of returns over the
study period (here, 1978-91) and measure the dis-
tance between this vector and the next vector and ask
how many times these four-dimensional vectors
were apart by the proximity distance, 1. That is all
this statistic is, but it is a very effective statistic. It has
been applied by BDS most recently in studying var-
ious markets.
Table 5 presents the BDS statistics for the seven
currencies and the two indexes using daily returns.
For two dimensions and a proximity distance of 0.5,
the currency returns are statistically significant, but
for the S&P SOD, they are not. The equity returns
begin to be significant in higher dimensions, but the
currency returns are significant in all dimensions.
The BDS statistic suggests the prevalence of linear
dependencies in the currency returns but only non-
linear dependencies in the equity market. The im-
plication is that, although profitable trading rules
based on autocorrelations in daily returns between
two points in time may be feasible in the currency
markets, such rules may not be viable in the equity
markets.
ATrading Rule for Currencies
The chaotic nature of equity market returns has been
cause for concern. As shown in Table 5, S&P 500
returns are not forecastable because they are chaotic
at the higher dimensions. Currency returns are non-
linear but may look like a random walk; unlike the
equity market, they are forecastable in a nonlinear
sense. The S&P market shows complex dependen-
cies, but at much higher dimensions, making it diffi-
cult to forecast for a period that is meaningful for
investment management.
4For a description of the BDS statistic, please see Chapter 2 in
William A. Brock, David A. Hsieh, and Blake LeBaron, Nonlinear
Dynamics, Chaos, and Instability (Cambridge, Mass.: The MIT Press,
1991).
Table 5. Brock, Dechert, SCheinkman Test, Daily Currency Returns, 1978-91 Period
Dimensions Proximity
Distance OM FFr SFr A$ C$ Fxlndx S&P500
2 0.5 11.89 10.17 7.93 9.54 6.04 24.06 15.58 2.13 0.85
3 0.5 14.82 12.68 10.15 13.69 7.70 30.69 18.57 3.35 1.13
5 0.5 23.21 16.83 15.18 23.86 11.11 44.59 25.31 6.87 2.59
7 0.5 37.73 25.45 20.92 41.48 16.29 73.67 39.23 9.78 3.82
2 1.0 10.83 10.19 10.03 8.31 6.97 22.76 13.52 2.43 2.78
3 1.0 13.07 12.35 11.00 11.28 8.35 26.61 15.79 3.48 3.43
5 1.0 17.67 15.30 12.97 17.08 11.38 32.24 19.61 6.83 4.82
7 1.0 22.87 18.78 14.78 23.05 14.75 38.92 23.72 9.52 6.11
Source: Ramaswami, Active Currency Management.
Note: The BDS statistic has a standard normal distribution, so those numbers larger than 1.96 are statistically significant.
Because trends and reversals are present in the
currency markets but no simple linear currency fore-
casting model exists, trading rules were formulated
in a nonlinear form. The simple filter rules are char-
acterized as "technical" trading or "momentum"
trading but are essentially nonlinear decision rules
expressed as functions of currency returns. The trad-
ing rule is: Buy when the dollar value of the foreign
currency rises Rpercent above its previous local low;
sell when it falls F percent from its previous local
high. Further, we set Rand F equal and did not
optimize to improve the trading rule performance.
The results of applying this rule are shown in
Table 6. I assumed round-trip transaction costs of 12
basis points. As expected, the presence of nonlinear
dependence in currency returns resulted in profit-
able trading rules. Among the currencies, the yen
showed the most gains and the four major curren-
cies-the yen, pound, mark, and Swiss franc-
yielded significant excess returns from filter rule
strategies. Interestingly, one cannot make any
money on the S&P 500 using this trading rule. The
S&P 500 returns are all negative and significantly so
as the filter increases.
This trading rule is nothing but a synthetic rep-
lication of an option. I buy the asset as it goes up and
sell as it comes down. I find incremental return
currencies, both on individual and composite levels,
and options should be benefitted by such a prevail-
ing trend in the market. I just took advantage of that.
Option-Based Strategies
The statistical distributions of currency returns sug-
gest the use of convex payoff strategies (options) to
exploit the presence of trends in currency returns.
Also, the simultaneous presence of reversals in daily
returns argues for the use of appropriate "filters" in
the synthetic implementation of options. The diffi-
culty in forecasting currency returns and exploiting
the nonlinear dependency in returns through the use
of filters also suggests the use of options to control
risk.
Table 7 presents simulated results (currency re-
turns) of hedging a multicurrency equity portfolio
comprising countries in the FT-EUROPAC index
with a multicurrency quantity-adjusting put option.
These returns are net of transaction costs and are
simulated using daily historical data. The average
unhedged return was 1.47 percent, and the fully
hedged return was about the same, but compare the
standard deviations. Leaving the currencies un-
hedged produces the same return with much more
risk.
The rolling simple option involves buying indi-
vidual options to hedge an underlying asset. I as-
sume that I have equity investments in 20 countries
and want to hedge the currencies generated by this
Table 6. Net Profitability of Filter Rule Strategy, Daily Returns, 1978-91 Period
Filter Rule OM FFr SFr A$ C$ Fxlndx S&P500
0.5% 0.88** 0.13** --0.86 -1.29 --0.27** 0.02 --0.63 3.05** -2.58
1.0 3.67** 2.53** 3.29** 0.58 2.60** --0.00 -0.28 2.04** -1.28
2.0 2.35** 2.59** 1.79 1.80 1.72 0.95 -0.29 1.82 --4.02
3.0 2.16** 0.77 0.45 0.46 0.58 -0.75 0.56 2.18 --{;.38**
5.0 3.07** 0.76 0.32 1.93 2.15 0.39 0.96 1.13 --{;.53**
8.0 1.93 1.90 1.50 1.47 2.56 -0.50 0.27 1.19 -9.52**
10.0 0.47 1.49 1.62 0.48 1.74 0.18 0.43 1.05 -8.59**
Source: Ramaswami, Active Currency Management.
** Significant at the 99 percent level.
103
Conclusion
rency returns that benefit the dynamic trading strat-
egy used to create the option synthetically.
From the distributional characteristics of currency
returns, I found that, unlike equitymarkets, currency
markets are efficient. Nevertheless, statistically,
mechanistic trading rules can be profitable. The cur-
rencies have a long-run zero return, but they have
profitable incremental returns over the shorter term.
I also found that currency returns are nonlinear. Be-
cause the patterns of currency returns are convex,
option-based strategies are appropriate. When done
synthetically, the investor should end up with a layer
of incremental return coming from the active man-
agement aspect of synthetically implementing an op-
tion.
Table 7. Return Results for Currency Protection Strategies, 1980-91
Rolling Quantity-Adjusting
Year Unhedged Fully Hedged Simple Option Option
1980 2.91% 2.27% 3.84% --{).37%
1981 -11.66 4.70 -0.72 1.43
1982 -10.88 2.23 -3.52 -1.65
1983 -7.25 1.60 -5.53 -2.68
1984 -11.07 2.66 -3.15 -0.80
1985 22.30 0.21 16.21 14.87
1986 16.91 0.59 10.29 11.86
1987 27.05 1.84 19.55 22.13
1988 -3.97 2.68 -4.20 -2.77
1989 -9.48 2.03 -3.07 -3.11
1990 9.26 --{).24 6.14 7.16
1991 3.30 -3.18 -1.32 -0.52
Annualized mean 1.47 1.43 2.57 3.51
Standard deviation 10.84 0.58 6.89 6.16
Source: Ramaswami, Active Currency Management.
equity movement. Depending on the underlying
asset moves, each month I buy or sell off the basket
of individual put options. I should have suspected
the rolling simple option strategy would do better _
thanthe fully hedged or unhedged strategies because
of the trending nature of currency returns. That is
what happens; the mean return is 2.57 percent, but
with more risk than the fully hedged strategy.
The quantity-adjusting option-an equity-
linked foreign currency put option-strategy is a
dynamically implemented, formula-based hedging
strategy involving the purchase and sale of currency
forwards. The frequency of the transaction is gov-
erned by the filter rules adopted to adjust the actual
hedge amount to the formula-recommended hedge
quantity. This layer of incremental return is attribut-
able to: (1) the lack of speculative capital for holding
open currency positions over the medium and long
terms, and (2) the nonlinear dependencies in cur-
104
Question and Answer session
Murali Ramaswami
Question: Given the paradox
of efficient, yet profitable, trad-
ing strategies, should hedged or
unhedged benchmark indexes
be used to judge an investment
manager's performance?
Ramaswami: Because trends
are embedded in currency re-
turns and options seem to be
the favorite instrument, 50/50
hedged/unhedged benchmarks
would be the appropriate
choice. Even if you buy an at-
the-money option, you are start-
ing at 50 percent hedged and 50
percent unhedged. The issue of
currency exposure has gained
in importance given the re-
search by Fama and French and
subsequently the claims by
their peers that we must use the
universal! global beta instead of
the domestic beta. In that con-
text, any domestic holding must
also have some foreign expo-
sure. This is also the outcome
of Fischer Black's universal
hedging research.
Question: Could you pre-
cisely define your trading rule
and how the filter works? Be-
cause the S&P 500 returns were
significantly negative under
this filter rule strategy, can you
systematically make profits by
doing exactly the opposite strat-
egy when the filter level is
reached?
Ramaswami: The filter rule is
mechanistic. It says buy the cur-
rency when its dollar value is R
percent above its most recent
low and sell that currency when
its dollar value is the same R
percent below its most recent
high. You buy as the currency
value begins to go up and sell
as the currency value begins to
come down, using a filter that
varies. The process is very sim-
ilar to dynamic or synthetic im-
plementation of an option. A
synthetic implementation of op-
tions must also have an element
of incremental return that
comes through such a mechanis-
tic trading rule.
If you did exactly the oppo-
site of the rule I just stated for
the S&P returns, you would
probably find significant incre-
mental returns for the trading
strategy, but the filters are far
wider than 0.5 percent. You
lose statistically significant re-
turns for the S&P when the fil-
ters are 3, 5, 8, and 10 percent.
That is an indication of rever-
sals, which we have already
seen in the S&P returns.
Question: Did you include
transaction costs in your re-
search?
Ramaswami: For currencies,
the results include transaction
costs of 12 basis points. For the
S&P, they do not, so one,should
add at least 75 basis points
round-trip to the cost.
Question: Please describe
your model for forecasting vari-
ances.
Ramaswami: In the currency
market, high volatility is fol-
lowed by high volatility for a
few days, then by a period of
stable returns, and so on. I
modeled the volatility of these
currency returns. The details of
this model can be found in my
Research Foundation mono-
graph, Active Currency Manage-
ment. I found that nonlinear de-
pendencies exist in currency re-
turns. This nonlinearity could
come through the volatility
rather than through the mean of
the returns, so I try to forecast
volatility rather than the returns
directly. I then use these fore-
casted volatilities to set limits
on the expected return and fore-
cast a mean return for the cur-
rency. Whenever the band wid-
ens, the market should provide
a higher risk premium for hold-
ing that currency; when the
band narrows, expect a decline
in the premium normally ex-
pected of that currency. I can
compare such variations and
risk premiums of cross-curren-
cies and make cross-currency
bets. This use of the forecast is
different from the usual use.
You can also do a direct fore-
cast of the volatility and trade
in options based on the fore-
casts.
Question: What happens to
your event analysis when you
exclude October 19877
Ramaswami: If I dropped the
month of October, I would
probably see a slightly different
result, but I do not believe the
result would change in any
major way because the weight
of October 1987 in this analysis
is only lout of 61 for the cur-
rency and lout of 89 for the eq-
uity index.
105
Using Nonstandard (Exotic) Derivatives in
Managing Portfolio Risk
Eric S. Reiner
Senior \flee President
LelandO'Brien Rubinstein Associates, Inc.
Exotic options are more flexible and more complex than the plain vanilla options. They
are available on an aTe basis in an increasing variety of forms, each tailored to a specific
use and objective. Understanding exotic options helps investors understand better how
to use standard options.
Exotic options fall into four classes: Single-asset/sin-
gle-tenor, single-asset/multiple-tenor, multiple-
asset/ single-tenor, and multiple-asset/ multiple-
tenor options.
components of a foreign investment separately. Fi-
nally, I will provide an overview as to how to put
these instruments together in the big picture.
Single-Asset / Single-Tenor Contracts
The single-asset/ single-tenor option is compar-
atively simple to analyze. It involves only one un-
derlying asset, a stock or foreign currency, or for
fixed-income instruments, one particular bond. Fur-
ther, the focus is on only one date-a single date of
expiration-and only one underlying asset price
over the life of the option. The standard vanilla call
and put are simple examples of that.
These options can be combined into portfolios
with various structures. One structure, for example,
might combine calls and puts with different strike
prices and perhaps different numbers of calls and
puts. You could also use something more funda-
mental than a standard option-binary options,
which are options with two discreet payoffs. Mark
Rubinstein and I discussed binary options and the
idea of hedging a contingent liability in an article in
Risk (October 1991).
As an example of the use of binary options,
suppose the S&P 500 mayor may not finish above
450 in three months' time. A standard call option
with a strike price of 450 would have a payoff that
The word "exotic" used to describe certain deriva-
tives immediately brings certain pictures to mind.
Although they may be interesting and stimulating
pictures, they do not correspond with the rational or
wise investment management view. Somehow pic- --------.--------------
tures of bikinis with Hawaiian prints do not help us Taxonomy of Exotic Options
understand how to manage portfolios or control
risks. Those working with exotic options are trying
to change that view. I amfond of the term "nonstan-
dard options," which means options with different
characteristics from standard options, the plain va-
nilla calls and puts. Other people like to use the term
"second-generation derivative products." Although
often used narrowly, "evolving term-structured
products" is perhaps the description that should be
used.
Exotic options help us understand somewhat
better how to use standard options. They help us
focus on key risk factors, distinctions among differ-
ent types of options, and different uses. Each of these
options can be obtained from any reasonably large
dealer; they are not imaginary.
This presentation will provide a taxonomy of
exotic options and then describe how these options
are priced and used, focusing on a few commonly
used path-dependent options. Average-rate options
are the ones people have good reason to use, not
necessarily in a portfolio management context but in
the context of a corporate treasurer trying to hedge a
foreign currency risk. I will also discuss multiasset
contracts, basket options, and derivatives that are
constructed to help manage the equity and currency
Note: Mr. Reiner is now with UBS Securities Ltd.
106
increases steadily once the S&P passes 450. A binary
option would have an all-or-nothing payoff pattern.
If the S&P 500 finishes above 450, the option pays off,
but the payoff does not change as the level of the S&P
500 varies above 450. If someone has a particular
contingent liability, he or she might choose to hedge
that contingent liability with a binary all-or-nothing
option.
Single-Asset / Multiple-Tenor Contracts
Single-asset/ multiple-tenor contracts, also
known as path-dependent contracts, are options
with a single underlying asset, but the contract has
values over many different dates. Examples include
compound options, extremum-based options, and
average-rate options.
A compound option is an option on another
option. An investor who wants some exposure to the
S&P 500 could buy an option to get that exposure.
Suppose, however, that the investor does not want
to pay the premium for that option. Instead, he or
she wants to lock in the premium for a call option but
delay making the decision for some time. This can
be done by buying an option on an option. This
strategy is path dependent because the payoff de-
pends on two dates-the date when you must choose
whether to buy the underlying option, and the expi-
ration date of the underlying option. Those factors
are the simplest path dependents. Two tenors are
involved.
For extremum-based options, what counts is not
just the final asset value but also the maximum or
minimum price of the underlying asset over the
whole life of the contract. One example of this type
of option is a barrier option. The most basic barrier
option, a down-and-out call, is like a standard call.
If, however, the underlying asset price crosses below
some lower barrier during the life of the contract,
then the down-and-out aspect goes away. Returning
to the example of the S&P 500 three-month call
struck at 450, you might put a knockout barrier at 400
on that option. If the S&P 500 ever falls below 400
during the next three months, the call option disap-
pears. These down-and-out and up-and-out options
come in many different varieties.
Another example of an extremum-based option
is the lookback option. A lookback call is like a
standard at-the-money call option, but if the under-
lying asset price ever goes down, the strike resets as
the price moves downward. If the asset price comes
back up, however, the strike price does not, so it stays
at the lowest level. Suppose you start with the S&P
at 430. If the S&P goes to 429, the strike price will be
reset to 429. If the S&P then goes to 435, the strike
price stays at 429. This option lets the investor buy
the S&P at its lowest value during the next three
months.
Lookback options are somewhat less practical
for purposes of risk management, in part because
these options have huge premiums. The premium
on a new lookback call is about twice that of a stan-
dard call option at the money; if the premium on a
one-year call option is $7, the premiumon a lookback
call option is about $13 to $14. Would you be willing
to pay someone who claims to have perfect market
timing ability and is always able to buy at the lowest
price 13 percent of the underlying value? The
lookback also comes in a put version in which the
strike is reset upward. The put strike is the maxi-
mum price the underlying asset reaches during the
life of the contract.
The third type of extremum-based options is bi-
nary barrier options. In some even stranger versions
of binary options, the contingent liability is based on
whether one strikes a particular barrier level.
For the average-rate option, the payoff depends
on the average value of the underlying asset over the
life of the contract. The value does not depend on
every tick of the underlying asset's price, however.
You might choose some sampling dates, say daily
close or once a month, as the basis for the contract's
payoff. This type is an innovation in exotic options
with some practical value, and so these options are
very popular.
Suppose you have a set of risky cash flows dur-
ing the future-receipts in pounds or yen, for exam-
ple. Say, General Motors is selling cars in Japan and
expects $100 million yen monthly. If the company is
concerned about the yen going down against the
dollar, it can choose some option-based strategy to
protect it on the downside. If it is concerned that the
yen will fall and the value of its receipts will decline,
it may want to buy a strip of options, one every
month to protect the yen level, and strike them at the
forward price for each month.
The problem is that the company is not looking
at the whole integrated picture of its cash flows.
From a rational viewpoint of managing the down-
side exposure, it should go over the relevant account-
ing period, lump the cash flows together, and protect
the level at which it exchanges the set of cash flows
rather than each cash flow individually. An average
rate option can do this.
Multiple-Asset / Single-Tenor Contracts
Multiple-asset/ single-tenor contracts are be-
coming popular, particularly in managing portfolios
or international exposures. Two types of two-asset
options can be built to deal with two individual
exposures: linked equity-eurrency hedges and out-
107
performance spread options.
Managing foreign investments is different from
managing a diversified portfolio of domestic assets:
the foreign investments entail two risk factors that
potentially can be controlled. The first is the under-
lying assets priced in their own currencies. For the
FTSE, it is the FTSE's price in pounds. Ifdollar-based
investors are investing in U.K. assets, the FTSE's
value in pounds is not the only factor that matters;
there is also the translation rate from pounds to
dollars. Investors must be concerned withbothcom-
ponents, and linked equity-eurrency hedge options
allow you to do that.
Another interesting structure to deal with dual
risks is outperformance hedge options. For example,
suppose a manager is using the S&P 500 Index as a
benchmark for a set of clients. The consultants say
the clients must have international exposure, al-
though they are comfortable with the S&Pas a bench-
mark. They also say diversification is great, but they
are not sure about the future of international mar-
kets. They would like a structure that would help
their clients invest in the Japanese market, for exam-
ple, and provide protection on the downside if the
Japanese market stops its upward ride and tanks.
An outperformance option structure speaks to
that requirement. The structure needed is one that
will payoff the better of the Nikkei or the S&P 500,
or the difference only if the Nikkei outperforms the
S&P 500. The currency could be allowed to float, or
the manager could control the currency. From an
options viewpoint, this structure is a zero-sumgame,
except for the premium of course.
Outperformance options can be expanded to be-
come multi-asset options. In this case, the structure
is known as a Best-of-Nstructure. Obviously, a con-
tract that is structured to return the best of multiple
markets will be more costly. This approach can also
be extended to a basket of assets, in which the option
is structured to hedge the aggregate piece.
Multiple-Asset / Multiple-Tenor Contracts
This type of structure protects against, for exam-
ple, the downside of a collection of different cash
flows in many different currencies. This is the new-
est type of exotic or nonstandard derivative, and it is
developing very quickly.
Exotic Option Pricing
The three questions to ask when pricing an exotic
structure are How is a particular option defined?
How is it used? What are some of the risk character-
istics of its structure?
The Black-Scholes model is the most frequently
108
used option pricing model. This model can be stated
mathematically in the form:
ds
- = (lJ.s [S,t] - ds [tJ) dt + as (t) dz
s
,
s
where Il can be path dependent and Zs(t) - N(O,t).
This model assumes continuous, frictionless
trading (Le., no jumps, lags, or transaction costs);
price movements are homoscedastic; and interest
rates, dividend yields, and volatility are known in
advance. It is not quite true to say the Black-Scholes
framework is based on lognormal asset prices, but
the standard deviation of returns over a short period
of time should be proportional to the volatility of the
underlying asset price. These assumptions allow
one to assume a lognormal distribution and to as-
sume that the gross rate of return on the underlying
asset is equal to the risk-free rate of return, although
the rate of return can be adjusted for risky coun-
terparties.
Call Options
The Black-Scholes model can be used to price a
standard (European) call with a strike price of K at
time T using the following formula:
C=Se-
dT
N(xl) - Ke-
rT
N(XI - crff),
where
_ log(SiK) + (r - d)T aff
Xl - crff + 2
Because of the symmetry principle, the value of
the put can be obtained from the Black-Scholes for-
mula just as easily by negating this whole expression
and the arguments of the normal integrals-that is,
make the whole formula negative and then make the
arguments of the normal integrals negative.
The pricing of an American option is different,
however, because you have to consider the possibility
that the option might be exercised. This is a path-de-
pendent situation in which a binomial model would
be useful. If a portfolio of other securities also has
options in it, analyzing the risk characteristics is cru-
cial. The risk characteristics are measured as follows:
Delta is the ratio of the change in an option's price
for a given change in the underlying asset or futures
price. Delta approaches 1 when the underlying asset
price is high and approaches zero when the underly-
ing asset price is low.
Gamma is the ratio of the change in the option's
delta for a given change in the underlying asset or
futures price. Gamma relates to how much of the
underlying asset must be traded to keep a hedge on.
As an option approaches expiration, the gamma of
an at-the-money option is quite high, indicating that
the hedger may have to do a lot of trading to keep the
hedge on.
Theta is the negative of the ratio of the change of
an option price to a change in expiration date. Theta
measures how fast the time value goes away. People
managing a portfolio of options need to pay attention
to theta.
Vega is the ratio of a change in an option price to
a change in the volatility of the underlying security.
It represents volatility risk. The volatility risk is
greater farther away from expiration, and it decays
as expiration approaches.
Rho is the ratio of the change in an option price
to a change in interest rates. Rho measures the
option's interest rate risk. It is more or less a duration
measure. Options with high values of Rho have the
greatest exposure to fluctuations and interest rates.
Collars
Simple options can be combined in a variety of
ways to form exotic or more complex options. The
simplest portfolio of options is a collar. The payoff
for a collar is:
max(ST - Kl,O) - max(ST - K2,O) with Kl < Kz .
In words, the strategy is to be long an option with a
strike of Kl and short an option with a strike K2,
which in effect caps the payoff at expiration. The
pricing is much easier for a European option than for
an American option because the payoff is simply the
sumor difference of the value of the component parts
(in this case the two options). This is known as the
linearity principle. Pricing an American option is
more complicated, because the rational exercise pol-
icy depends on what is in the whole portfolio at that
particular time.
Equity-Linked Bonds
Equity-linked bonds are an example of how op-
tions can be used to add an equity component to a
bond structure. The payoff for an equity-linked
bond is:
LC(ti) +M + U. max(ST - K,O) .
I
The linearity principle applies here, so the payoff
structure can be decomposed to determine the value
of the strategy. Each piece is valued individually,
and then they are summed.
Path-Dependent Options
An option is path dependent if its payoff is mea-
sured in terms of where the asset price path has been
between the beginning (t =0) and the expiration (t:::
nof the contract. The three types of measures most
commonly focused on are the maximum price, min-
imum price, and average price. Standard European
options are not path dependent. For standard Euro-
pean options, the only consideration is the asset's
value at expiration. American options are path de-
pendent, although not in the sense that a particular
path leads to a known value. Rather, the dependence
is based on the model of rational exercise policy. The
model used for American option valuation and exer-
cise drives the path dependence.
Many exotic options are path dependent.
Lookbacks, for example, depend on the maximum
and minimumprices, because the strike is readjusted
based on the path of the stock's price. Barrier options
are also path dependent, because their value de-
pends on whether the asset hits the maximum or the
minimum level.
Compound Options
Compound options are options to buy or sell an
option with maturity date T and strike price K at time
t for strike k. There are four basic types of compound
options.
A call on a call allows you to delay the decision
about buying a call by paying a small amount of
premium now and effectively locking in a price for
the call in the future. The payout for a call on.a call
is max(Ct - k,O).
A put on a call allows you to hold a call and save
the ability to unload it at some future date, locking in
a price for doing that. The payout for a put on a call
is max(k - Ct,O).
The other two are a call on a put, with a payout
of max(Pt - k,O), and a put on a put, with a payout of
max(k - Pt,O).
Chooser options have similar structures. These
options allow the owner to choose at time t the better
of two options (normally, a put and a call) with
possibly different times to maturity and strike prices.
This option essentially allows the owner to lock in
the price of a call or put option or, for that matter, any
type of option. With a chooser option, the seller of
the option gives the buyer the right to decide, say
three months from now, whether he or she wants to
receive a put or a call. It is like a straddle except the
decision point is delayed by, in this case, three
months. With a standard straddle, at expiration,
either the call pays off or the put pays off. With a
chooser option, the decision is made at some point
before the expiration of the option, which lowers the
premium somewhat. This device is very useful to
those who expect information flow from the markets
in the near future and would like to make the deci-
109
sion whether to take a call or put after the informa-
tion arrives.
Average-Rate Options
The first average-rate options were done on oil
out of Tokyo; hence they are often called Asian op-
tions. In a portfolio management context, average
rate options do not have much applicability, because
most equity investors are not concerned with the
timing of cash flows. These investors have an invest-
ment horizon and are concerned with asymmetries
and risk exposures at that particular time horizon. In
a corporate context, however, when companies want
to hedge some aspect of risky cash flows, variation
in time-averaged prices is often a more appropriate
measure of risk than variance of the stock price.
Average rate options might be used to hedge fluctu-
ations in commodity and raw material prices or in
exchange rates. Accordingly, options based on aver-
ages may be more attractive than standard options to
some potential hedgers.
Average-rate options are attractive for several
reasons. First, lumping exposure together makes
sense from an accounting and management view-
point. Another advantage is that the effective vola-
tility of an average-rate option is lower than a stan-
dard European option with the same expiration date,
and the premiums are typically half as large.
These options benefit from the fact that the vol-
atility of a portfolio is lower than the volatility of the
individual components. In fact, hedging each cash
flow individually creates overinsurance. Unless all
of the prices are perfectly correlated, some will move
up and some will move down, so some offsetting of
risks will occur. By creating a portfolio of risk expo-
sures, protection on that portfolio will be somewhat
less expensive than protection of the individual cash
flows lumped together. In other words, an option on
a portfolio is cheaper than the corresponding portfo-
lio of options. Compared with a basket of options,
the premium for an average-rate option is about 10
percent lower.
The two basic types of average-rate options are
the average-price optionand the average-strike-price
option.
An average-price option is like a standard option
except that the underlying asset is replaced by the
average price over time. The payoff is max(AT- K,O),
where AT is the arithmetic average of asset prices
sampled between the beginningand expirationof the
option. The average is typically an arithmetic aver-
age, so it is like a weighted portfolio, which makes
this product harder to analyze quantitatively. Geo-
metric averages turn out to be easier because the
geometric average of lognormal variables is also log-
110
normal. The sum of lognormal variables is not log-
normal, however. In a corresponding version of
these options, the payoff does not have a fixed strike,
but rather the average price is acting as a strike for a
standard option.
The average-strike-price option, for which the
payoff is max(5T - AT,O) is not as useful as many
others. The only practical situation in which to use
these is, if traditionally the investor had settled at the
end of an accounting period at the spot price, at that
settlement date, this type of average-rate option
might be used to gain better control of the average as
opposed to the spot price.
Nothing about the use of average-rate options
requires use of the same underlying nominal amount
all the time. The nominal amount might be 100 mil-
lion yen this month, 250 million yen next month,
maybe skip a month, and then another 100 million
yen the next month. The average in the average-rate
option does not have to be constant.
Basket Options
Investments can be combined into baskets rather
than dealing with them on an individual basis. Con-
sider an investor with diversified foreign holdings or
one who has cash flows in multiple currencies.
Should that investor insure each cash flow individu-
ally or protect the aggregate? For derivative strate-
gies, it does not matter. The payoff for a call on a
basket is the weighted sum of the returns on the
different assets in the basket minus the strike price or
zero. Absent perfect correlation between the assets'
returns, a portfolio of assets provides some diversifi-
cation and lower volatility than the sumof the pieces.
Similarly, the effective volatility of the basket option
will be smaller than the weighted sum of the volatil-
ities of options on the individual components.
Basket options look cheaper than the portfolio of
options, but there is no free lunch. First, you may be
paying less in premium, but you are not getting the
same payoff. This is a stochastic dominance issue. In
some situations, the individual options would pay
off but the basket option will not. The options on the
assets whose prices go up, for example, would have
paid off, whereas the option on the basket might not
have paid off if it contained a mix of assets whose
prices went up and some whose prices went down.
Basket options make sense for investors whose objec-
tive is benchmark protection of the whole portfolio.
Second, basket options are subject to what is
known as an innovation premium, which is what a
dealer charges because you do not knowhow to price
the structure and he or she does. Also, trying to
hedge with basket options can be complex. The vol-
atility of the different basket components varies, and
the correlations are highly unstable. If market pa-
rameters shift in one country or another, the hedge
can get into trouble. The dealer will explain how
hard it is to hedge the structure and, therefore, why
he should charge a lot of money for his services.
Conclusion
Exotic derivatives can be used in many ways. One
caneither look at exposures to a single asset over time
or at a single exposure at one point in time or at
multiple exposures and multiple time frames.
One advantage of risk aggregation is diversifica-
tion; one disadvantage is innovation costs. Another
disadvantage is implementation features. We do not
always know how the market parameters interact
when different exposures are lumped together. Ex-
Exhibit 1. Family Tree of Binary Options
Binary Options
I
posure to risk is easier to look at in an individual
context than in a whole portfolio. Looking at portfo-
lio risks as a whole is crucial, however. Of course, if
you are sensitive to market parameters in different
ways, you will face some complexity in hedging.
Maybe every risk should be broken into tiny
pieces and reaggregated differently. Mark Rubin-
stein and I thought about how we might do this and
came up with Exhibit 1, which illustrates a family
tree of different structures that can be built out of
binary options. This example is one set of ways to
break up exposure. Breaking up exposures and re-
aggregating themis very complicatedand confusing.
Nevertheless, the derivatives industry is moving in
this direction-looking at all of the different expo-
sures, breaking themapart, analyzing them, and put-
ting them back together in a structure that makes
sense for protection or exposure.
Path-independent
I
(Payout at expiry)
I
Payout at hit
(1-4)
I
Path-dependent
I
Payout at expiry
(5-28)
I
(Strike only) (Barrier only) Barrier anIy
(5-12)
,.---_1_-----,
Barrier and strike
(13-28)
I
(In) In
(5-8)
Out
(9-12)
In
13-20
Out
(21-28)
(1,3) (2,4) (5,6) (7,8) (9,10) (ll,12) (13,15,17,19) (14,16,18,20) (21,23,25,27) (22,24,26,28)
Cash Asset Cash
or 0 or 0 at hit
nn
Call Put Call Put (l)
Asset Cash Asset Cash Asset Cash Asset Cash Asset Cash Asset Cash Asset Cash Asset Cash Asset Cash Asset
at hit at hit at hit or 0 or 0 or 0 orO or 0 orO orO orO orO orO orO orO or 0 orO orO orO
I
(2) (3) (4) (5) (6) (7) (8) (9) (10) (ll) (12) (13) (17) (l5) (19) ,L, (2]) (25) (23) (27)
0" C,,'''''' C,," ""'C,"",
Portfolios
I "9=P"' I
Cap Cap Supershare
Can Put
Down- Down- Up- Up- Down- Down- Up- Down-
and- and- and- and- and- and- and-
in- in- in- in- out- out- out- out-
Call Put Call Put Call Put Call Put
Standard Standard
Call Put
Source: Eric S. Reiner
111
Question and Answer Session
Eric S. Reiner
Question: The modified Black-
Scholes model does not take into
account dividends for early expi-
ration or early exercise. How can
you adapt the binomial method
to take these factors into account,
using as an example the quanto
option, which is an equity option
on a foreign equity? What about
N-way underlying quantos?
Reiner: For a quanto option, we
are concerned about currency
risk. No matter what it is, the
payoff from this option is trans-
lated into another currency at a
preagreed upon exchange rate,
typically the forward rate.
For example, I am interested in
gaining exposure to the English
market. I buy a call option on the
FTSE. Because the pound has
been unstable, I am concerned
that when that option pays off in
pounds, those pounds may be
worth less. I will agree with my
dealer that whatever the FTSE op-
tion pays will be converted over
to dollars at the rate of, say, $1.50
per pound. If the pound goes
down, I do not lose. If the pound
goes up, however, I lose out on
its appreciation.
Question: How does that strat-
egy differ from an ordinary cur-
rency option?
Reiner: The strategy differs be-
cause the notional principal un-
derlying the currency option
changes. The underlying equity
exposure is driving the boat, and
112
the currency forward's notional
principal amount changes with
the option payoff.
The binomial formula must
take into account two risk fac-
tors-eurrency and equity-with
the quanto option. The first good
model was a multiasset version of
the Cox-Ross-Rubinstein model.
A better way to solve the problem
is to collapse the two risks into
one dimension. If the underlying
asset is actually a basket, and if it
is a European option, it can be
valued like a standard basket op-
tion with some tweaks. This is
not possible for an American ver-
sion, however.
Question: What do you think of
the pricing and usefulness of the
Citicorp average price S&P 500
deposit notes?
Reiner: I have not analyzed
them, so I cannot make a judg-
ment. They are basically average-
rate contracts. The key feature on
the average component is that it
is somewhat different because it
is not actually focusing on the av-
erage as much as it is just making
sure that one particular point in
time is not gamed, which is one
nice feature of the average-rate
option. Some contracts have aver-
aging over a short period to make
sure someone is not putting the
squeeze on by gaming up the
market at expiration to create a
huge liability at that point.
Question: Will that contract be
hard to price?
Reiner: I believe it does not
have an American exercise fea-
ture, which means pricing it
should take a good quant an after-
noon.
Question: Given the state of
knowledge and technology, peo-
ple figure out pretty quickly how
to value these options. How long
does an average innovation pre-
miumlast?
Reiner: It is amazing how fast
an innovation premium goes
away. Its life span depends on
how quickly the dealers are able
to figure out how to hedge
against some of the risk compo-
nents. The life span with some of
the path-dependent options on
liquid underlying assets that do
not show a lot of jumpiness is as
soon as someone comes up with
it and shows it to clients.
Some structures are harder
than others to put a fair value on.
A classic case is the Best-of-N
structures on foreign equity that
has a lot of correlation exposure.
The correlation exposure is a con-
cern, but there is no clean way to
hedge it. In currency markets,
correlation is very easy to hedge,
but in equity markets, it is not.
So these options have a large pre-
mium that represents not only
the newness of the product but
also the fact that correlation can
go haywire.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ~ _ . " . _ - - -
The Comparison of Strategies Using
Derivatives
Maarten L. Nederlof
Principal
TSA Capital Management
The extent to which derivatives are used in a portfolio (and the types of derivatives that
are used) affect the portfolio's risk-return profile. Options can be particularly effective
because of their asymmetrical nature. The choice of which option strategy to use depends
on the manager's market expectations and objectives.
Many investors have a hard time understanding
how to compare the assets in their portfolios, much
less how to use derivatives. The traditional way to
make these comparisons is known as modern port-
folio theory (MPT), and most people use the capital
asset pricing model to value assets using this frame-
work. The basic concept of MPT is that an asset's risk
premiumis determined by its contribution to the risk
of the investor's overall portfolio-it is the compen-
sation for a marginal unit of risk in a portfolio. This
additional risk is what should be considered as op-
posed to the total risk of the individual asset.
Risk-Return Profiles
The two metrics for evaluating an asset using MPT
are expected return and estimated risk. Deriva-
tives-particularly options, which are asymmetric
investments-eomplicate the comparison of risk and
return. Nevertheless, the same approach can be
taken to evaluate a derivative, especially if it is not a
stand-alone investment but part of a hedge or a larger
portfolio.
Expected returns are typicallybased on the mean
of a return distribution, which has been determined
using historical data. Some firms, for example, use a
20-year average risk premium, assuming that the
past 20 years are representative of the risk premium
on stocks. The assumptions required of a historical
sample can create several problems, and investors
must decide whether those assumptions are good.
Alternatively, expected returns can be estimated
using a subjective forecast.
Risk is most commonly represented by the stan-
dard deviation of a return distribution. The standard
deviation is typically calculated using a historical
sample and is most useful to describe a symmetric,
normal distribution. Gary Gastineau discusses how
to evaluate an investment from a probabilistic per-
spective.
I
I am more of an empiricist and prefer to
consider how a return distribution or probability of
outcomes fits with what an investor wants, as op-
posed to looking at the volatility by itself.
Figure 1 plots in standard risk-return space the
performance of seven basic asset classes during the
past 20 years. The up-sloping tendency indicates an
increasing risk premium in moving from low-risk
instruments to high-risk instruments. For example,
an investor is compensated for going into long-term
corporate bonds from T-bills. Similarly, stocks have
both higher return and higher risk than bonds. Ulti-
mately, small-capitalization stocks provide the high-
est returns over this time frame to compensate for the
increased risk an investor must bear.
For simplicity, people assume that returns to
stocks are normally distributed in the short run. In
reality, this assumption is not precisely true because
stock prices do not go below zero (although stock
returns can be negative, as illustrated by the straight
payoff line). Figure 2 shows a typical distribution
and a payoff diagram for a traditional stock index
investment. According to this return distribution,
the expected mean return is 13.3 percent and the
standard deviation is 17.5 percent. These assump-
tions lead to a 22 percent chance that the return will
be negative over a one-year horizon.
Figure 3 illustrates a typical distribution and
lSee Mr. Gastineau's presentation, pp. 54-f,1.
113
20
oL-._..l-_-L_----l__..L-_--l.-_--'_---::'::--::'
o 4 8 12 16 20 24 28 30
Standard Deviation (%)
Figure 1. Traditional Framework:
Historical Performance of Seven Asset
Classes, 1972-92 Period
S&P500
stock/bond investment. The distribution was esti-
mated using Monte Carlo simulation and the as-
sumption that the correlation between stocks and
bonds is 43 percent. Apparently, the concept of di-
versification works, even empirically. The mean of
this portfolio is 11.6 percent, and the standard devi-
ation is 12.5 percent, but the risk of losing money
dropped to 18 percent from more than 20 percent on
each of these assets separately. The payoff line is still
straight, especially if done in the context that the
portfolio can go up or down. The mean return is
about 200 basis points more than the bond-only
yield, but the increase in risk is marginal (0.2 percent-
age points).
Small-Cap Stock
Long-Term
Corporate .
Bond... High-YIeld Corporate Bond
Intermediate Long-Term Government Bond
Government
Bond
16 r-
c
.2 12 f-
Q)
p:::
c
2l 8 r-
::"2 T-Bill
4
Source: Ibbotson Associates.
Figure 2. Typical Probability Distribution and Payoff
Diagramfor aTraditional Stock Investment
Figure 4. Typical Probability Distribution and Payoff
Diagram for aSO/SO Stock/Bond
Investment
-40 0 13.3%
Stock Index Return
Source: TSA Capital Management.
60
C
h
;l
r
OJ
;.:::
p:::
0 li
li
0
'"
i1
.J
'"
0
.J
"E
h
0
P-.
h
0
P-.
P-.
18%
-30 0 11.6% 50
Portfolio Outcome
Source: TSA Capital Management.
Adding certain derivatives to existing holdings is not
like adding simple investments together. The asym-
metric outcome nature of options tends to alter the
portfolio in a quite useful manner, and this is
illustrated using a protective put strategy and a cov-
ered call strategy.
Protective put strategy. With a protective put
strategy, the investor buys an asset and buys a put on
that asset to eliminate some of the downside risk.
The investor can decide where he or she wants to
strike the put in order to gain the desired protection.
The return distribution and payoff diagram are
shown in Figure 5. The payoff diagram for the un-
derlying investment-in this case, a stock index in-
vestment-is still linear, and it has a defined stan-
dard deviation of 17.5 percent.
When a protective put strategy is added, the
payoff line is "kinked" and the distribution is no
longer symmetric. When 50 percent of the portfolio
is covered with puts struck at the money, the proba-
bility mass starts to get squashed to the right; half of
the portfolio will get stopped out at the put strike. As
50 o 9.6%
Bond Index Return
-30
payoff diagram for a bond index. This mean is 9.6
percent and the standard deviation is 12.3 percent.
Both the mean and standard deviation are lower than
those for stocks, which is consistent with Figure 1.
The probability of losing money, however, is 23 per-
cent in a year.
Combining the two asset classes into one portfo-
lio improves the risk-return profile of the portfolio.
Figure 4 shows the typical distribution for a 50/50
Source: TSA Capital Management.
Figure 3. Typical Probability Distribution and Payoff Adding Derivatives
Diagram for aTraditional Bond Investment
114
Figure 5. Probability Distribution and Payoff
Diagram for a Protective Put
Figure 6. Probability Distribution and Payoff
Diagramfor aCovered call
[
!-.-'--------------
/ ~ - - - - - - - -
/;'/1
// I
.-"../' I
.' ":,? I
/ .r
/".7./' .. ;;lJ ...
. ,? .'> I
// ;.------ I
..7 .. ':"---
80 60 40 -20 0 20
Return (%)
-60 -40 80 60 40 -20 0 20
Return (%)
-60 -40
Underlying Security
(left- and right-hand scales)
50% Protective Put
(right-hand scale)
100% Protective Put
(left- and right-hand scales)
100% Protective Put Net
of Premium (left-hand scale)
Underlying Security
(left- and nght-hand scales)
50% Covered Call
(right-hand scale)
100% Covered Call
(left- and right-hand scales)
100% Covered Call Net
of Premium (left-hand scale)
Source: TSA Capital Management.
Source: TSA Capital Management.
the coverage is increased to 100 percent of the port-
folio, the probability of a return less than the pre-
mium has been eliminated. Investors must pay for
this privilege, however. In fact, the return declines
by the amount of the option premium, so it is slightly
lower than it would be if the option were free. The
standard deviation also declines significantly be-
cause one side of the distribution has been clipped
off. The probability of principal loss is about the
same as before-24.5 percent-but, in the 100 percent
strategy, the probability of performing below what
was paid for the put is now zero.
The give-up in return for a protective put strat-
egy is not much more than the cost of the option. In
reward, the downside probability is significantly re-
duced and the standard deviation is lower.
Covered call strategy. With a covered call
strategy, an investor writes calls on stocks already
owned. This strategy generates premiums in the
short term but sacrifices the stock if the price moves
through the exercise price. The strike price usually
is set near the expected price. Figure 6 shows the
distribution and payoff for this strategy, assuming
the call is sold at 10 percent out of the money.
The payoff for a covered call portfolio is
"kinked," and the distribution is not symmetric. The
portfolio will perform up to a certain point (10 per-
cent out of the money, in this case) and then be
stopped out from any further upside participation-
completely for a 100 percent covered call strategy
and for half of the portfolio for a 50 percent covered
call strategy. The probability mass is shifted toward
the call strike. Both the expected return and the
standard deviation decline. The standard deviation
is about 7.3 percent, a 10 percent reduction relative
to the underlying asset distribution. In this case, the
probability of principal loss is 16 percent.
A covered call strategy results in a reduction in
return similar to that of the protective put, even
though a premium is earned. The strategy does not
provide significant reduction in downside probabil-
ity, but the standard deviation is visibly reduced
(indicating that standard deviation may not be the
appropriate measure of risk for this strategy).
The degree to which derivatives are used in port-
folios affects the risk-return trade-off of the portfolio.
Figure 7 plots the risk-return trade-offs for a series
of portfolios with call and put options. As the in-
Figure 7. Risk-Retum Trade-offs for Portfolios with
Various Portions of call and Put Options
c
....
.75
.2 15
(J)
0::
"0
~
u
(J)
0... 8
><
'-'-l
10 16.2 17.5
Standard Deviation
-- Covered Call Strategy
- - - Protective Put Strategy
Source: TSA Capital Management.
115
Figure 8. Probability Distribution for aCovered call
and Protective Put
Underlying Security
(right-hand scale)
50% Coverage
(right-hand scale)
100% Coverage
(left- and right-hand scales)
100% Coverage Covered Call
and Protective Put
_______ _
Source: TSA Capital Management.
80 60 40
.c
----------_ .. _ .. ;.:::
"8
P-.
-20 0 20
Return (%)
.. '
-60 -40
vestor writes covered calls on this portfolio (first 25
percent, then 50 percent, 75 percent, and finally the
entire portfolio), the risk is reduced because the size
of the probability distribution is decreasing. The
protective put strategies are similarly decreasing the
size of that distribution, and the standard deviation
is lower, adding to the stability of the portfolio. The
covered call has a considerably larger effect onreduc-
ing the standard deviation because of the premium
earned.
As these figures show, the standard deviation
measure of observable risk is a problem when deriv-
atives are used. Covered calls, for example, do not
preclude losing money just because they are earning
a premium and reduce the standard deviation. De-
spite the lower standard deviations, covered call
strategies are by no means a safer investment. This
strategy limits the upside but looks like a better risk
reducer than a strategy such as a protective put,
which limits the downside. This exercise shows that
standard deviation is a poor risk descriptor.
--------_.. _------------
Option Strategies
Given the deficiencies of MPT when derivatives are
involved, how do you decide if an option strategy is
appropriate for a particular portfolio? Just as pro-
grammers try to talk away a bug and turn it into a
feature, maybe the problems in judging options as an
asset can be used to our benefit. Two possible "bugs"
are that asymmetry leads to distorted risk measures
and that asymmetry can lead to nonparticipationand
missed opportunities. On the other hand, asymme-
try can also truncate possible outcomes and allow
selective participation.
Adding traditional assets to a portfolio is like
white, unfocused light. Options can be used as a
financial lens or prism to focus on a particular por-
tion of participation. Derivatives are versatile and
can be used in various combinations to customize a
product and lock in a defined portion of the partici-
pation in an investment. Figure 8, for example,
shows the probability distribution for a strategy that
combines the protective put and covered call, which
is the equivalent of buying an asset, buying the put,
and selling the call. As the options are added, the
distribution narrows. At 100 percent coverage, the
distribution focuses on a very distinctive piece of
participation, The potential series of outcomes is
reduced to a defined range of interest.
The follmving examples illustrate the versatility
available through the use of options.
Example 1. Suppose a traditional equity in-
vestor toward the end of the year is concerned about
giving back some gains. Maybe the investor can
tolerate no more than 10 percent downside risk. One
solution would be to buy a protective put 10 percent
out of the money. For a premium, this truncates
downside participation 10 percent below where the
portfolio would participate without the put. For
some investors, however, the cost of the premium is
unacceptable.
Another solution is to use a zero-premium par-
ticipating forward. This strategy involves buying
out-of-the-money puts and selling in-the-money
calls at the same price. Because the calls are in the
money, their premiums are higher than the premi-
ums paid on the puts. With this strategy, you give
up some upside participation to assure downside
protection, with no outlay of premium. By no means
is this free, however, and that should be emphasized;
the payoff still looks like a long call, but the slope in
participation on the upside is not as steep as that of
the underlying investment, and thus this strategy
will underperform in a market rally.
A third possible solution, called a range forward
(or collar, zero-premium risk reversal, or kinky for-
ward),is shown in Figure 9. In this strategy, you buy
out-of-the-money puts and sell out-of-the-money
calls, generally in the same proportion as the puts but
at such a strike price that the premium earned pays
off the cost of the puts. You give up all the upside
beyond the call strike price, participating fully be-
tween the strike prices, and eliminate the downside
below the put strike price. This strategy is similar in
probability to the strategy in Figure 8. The payoff is
within the 10 percent downside limit but can still
return up to 14 percent. If your risk premium as-
116
Figure 9. Payoff Pattern for aRange Forward
/'
/'
/'
/'
/'
/'
/'
j
o 0
J
-10 +14.8
S&P 500 Return (%)
Source: TSA Capital Management.
sumptions are not that high, this could be a desirable
strategy.
Example 2. Suppose a global investor is bull-
ish on the German stock market but bearish on the
deutsche mark. One solution is to buy a basket of
German stocks. Pick through some German newspa-
pers to see what is worth buying, take that position,
sell the deutsche mark forward to dollars, and buy a
simple hedge to eliminate the currency exposure.
This solution is not very satisfactory, however, if
German stocks decline. A forward hedge is expen-
sive. Even selecting German stocks is difficult for
U.S. investors.
Figure 10. Payoff Pattern for a call Option on the
DAX and aSynthetic Forward Short
Strategy
/'
/'
/'
/'
/'
/'
/'
/'
J /'/'/'/'
:; 0 f----------/'--'---/' .
U /'
X /'
i3 /'/'
/'
/'
/'
/'
/'
/'
/'
/'
DAX Return
--- DMK
--Short Call
...... Long Put
Source: TSA Capital Management.
Another solution, which is illustrated in Figure
10, would be to buy a call option on the DAX, and
then use options to create a synthetic forward short
position by buying a deutsche mark put and selling
a deutsche mark call, both with the same strikes and
time to expiration. This strategy provides a synthetic
hedge for currency exposure, but it suffers from two
problems. First, it is expensive, because you must
pay the forward points. The interest rates are differ-
ent, and the market is smart enough to know how to
prevent arbitrage. Second, this strategy does not
hedge the full exposure, because as soon as the equity
market moves, the amount that requires hedging
changes.
In a third solution, the dealer takes the currency
risk and rolls it into one investment giving the same
participation in a call on the DAX but denominated
in dollars. Of course, you will pay extra to have that
currency risk removed. One of the benefits in pack-
aging these options together is that the DAX and
deutsche mark do not move in lockstep. They are
uncorrelated, so placing these options in a portfolio
usually reduces its overall volatility. This alternative
is often cheaper than buying the individual options.
Example 3. Suppose a global investor wants
to invest in the French market and that French rates
are currently high (11.1 percent). Forward curves are
implyinga 325-basis-point drop in six months, which
would have an adverse effect on the French franc.
Most people believe it unlikely that rates will decline
that much, however. A possible solution, shown in
Figure 11, is to buy an agency-issued, dollar-denom-
inated note with an interest rate structure embedded
in it. It is a one-year maturity instrument paying a
fixed coupon, and the principal is linked to the
Figure 11. Payoff Pattern for an Agency-lssued,
Dollar-Denominated Note with an
Embedded Interest Rate Structure
~
c
....
.a 7.25-
~
'iiJ
;S
-7.25 f--------/
6.5 8.4 11.1
Level of PIBOR (%)
Source: Goldman, Sachs &Co.
117
PIBOR (Paris Interbank Offered Rate), which has a
cap and a floor. When you buy this instrument, you
are betting that the forward rate is wrong in predict-
ing the way rates will go. If you are correct and rates
do not drop below 8.4 percent, you collect the 7.25
percent return. This strategy has potential risks,
however, and knowing what they are is important.
If you are wrong, and rates decline to 6.5 percent, you
will lose 7.25 percent of your principal. Also, you
may be assuming the credit quality of the issuer of
the note.
Example 4. Suppose an investor thinks ge-
neric competition will adversely affect the earnings
of brand-sensitive consumer nondurable stocks. A
way to benefit from this information would be to
short a basket of consumer nondurable stocks, such
as Philip Morris, Colgate, or Proctor & Gamble, to
capture any decline. If the stock market rallies
sharply, however, this strategy would be disastrous.
An alternative approach might be to short the
basket of stocks and buy S&P500 futures to eliminate
the beta of the portfolio. In this case, if the market
rallies, the portfolio participates because of the
hedge. The problem here is that the futures hedge
needs to be managed. As the stock market moves in
price, someone must watch and adjust the number of
futures contracts needed. Although trading firms
usually have traders to perform this function, pen-
sion funds usually do not want to hire money man-
agers for that task after paying brokers for the premi-
ums for some of these options.
Another strategy is to buy puts on each stock in
the group. You do not need a futures hedge because
you are not as worried about the market anymore.
You have a limited upside exposure on the shorts,
and the maximum loss in the event of a market rally
is the cost of the options. You have a defined range
of participation. The problem is that the cash outlay
is huge. Most Wall Street strategists know that just
buying puts on a portfolio is not a good way to hedge,
because any cost in put premium will eclipse any
returns or benefits they could have made.
118
A fourth possible solution is to buy a put on the
basket of stocks. This approach usually lowers over-
all volatility and has a lower premium outlay. The
maximum loss in the case of a market rally is the cost
of that one-basket option.
Summary
These examples illustrate the number of strate-
gies investors are starting to carry out on a regular
basis. You can now go to broker / dealers and ask
themfor"quanto" options (hedged of currency risk),
"diff" options (an option on the difference in return
between two instruments), and "best of" options
(which payout the best of a collection of specified
alternatives). If you want to capture a particular
idea, you can request an option on a specific list of
stocks. Using derivatives is a great way to focus on
a specific scenario and take advantage of it.
Conclusion
Options are attractive investment instruments for
several reasons. First, they are focusable instru-
ments. They can be used to lock in a particular
scenario. They offer no guarantees, but they do offer
a specific focusing of potential participation and lim-
itation of risk. Very often in implementing a specific
idea, using options is the most efficient way, because
any attempt to use some of the underlying assets
becomes complicated.
Options are customizable. Customizable op-
tions generally trade through the OTCmarket, which
may entail liquidity and credit risk problems. Many
of these risks must be considered in deciding to use
derivatives.
Ultimately, these investments must be under-
stood. They must be considered in a framework that
accurately represents to potential investors how they
will act under all possible scenarios, and this frame-
work must also include how they will combine with
other instruments already held.
Question and Answer Session
Maarten L. Nederlof
Question: We know one of the
reasons option-based strategies
do not fall on the risk-return line
is asymmetry. Another possible
reason, however, is that pricing
of options is based on a lognor-
mal versus normal return distri-
bution. How does lognormality
affect your strategies?
Nederlof: The assumption of
lognormality is more pertinent to
price space than return space (be-
cause the price of a stock, for ex-
ample, cannot be zero, but the re-
turn can be). The lognormal as-
sumption for price movements
and the normal assumption for
stock returns are not necessarily
in conflict. In these strategies,
you get largely the same results
regardless of the distribution as-
sumption, whether you are con-
sidering prices in a lognormal
framework or returns in the nor-
mal framework.
Question: Will the zero-pre-
mium participating forward al-
ways be dominated by some
other strategy? For example, if
you think prices will fall, you
should go into cash. If you think
prices will rise, you should leave
your position unhedged. If you
think prices will be somewhere in
the middle of the strike range, do
a collar.
Nederlof: I agree. My discus-
sion was merely an illustrative
progression from a protective put
to a zero-premium participating
forward and then to a collar.
Question: You reviewed the
covered call and the protective
put, arguing that a strategy limit-
ing the upside gain is a better risk
reducer than one limiting the
downside risk. Please explain
how you came to that conclusion.
Nederlof: The point I tried to
make is that you can come to
that type of conclusion if you
use standard deviation blindly
as an indicator of risk. Because
you are earning a premium in
the covered call strategy, the
strategy tends to give a higher
return for a given amount of
risk. At the same time, that
premium now in your account,
along with the total returns
generated by your market
participation, gives a lower
overall standard deviation
even though the downside
risk could be just as large in
terms of the market move-
ment. The danger in looking
at these strategies in a standard
deviation framework is that
you will buy the idea that a
risk reducer is a risk reducer even
if it only limits upside participa-
tion.
119
Managing Derivatives: The Plan Sponsor's
View
Matthew R. SmITh
Senior Portfolio Manager
Amoco Corporation
Despite some "start-up" problems, using derivatives has several benefits for pension
funds. Amoco, for example, uses them to create market-neutral portfolios and synthetic
equity funds, to adjust portfolio duration, and to control its multimanager portfolio.
We mainly use derivatives in four applications. The
major one is in market-neutral portfolios. We have
hired outside managers who rank stocks from, say, 1
to 1,000 according to how much they like them. They
buy the top quartile and short the bottom quartile-
those they think will be underperformers. Then,
they optimize the characteristics of the two to reduce
tracking error. The beauty of the market-neutral
approach is that it does not depend onwhat the stock
market does. Through the use of futures or swaps, a
market-neutral portfolio can be turned into anything
you want, which is evidence of the flexibility of this
approach. Not everyone believes in this type of strat-
egy, however. Whereas an optimist will say, "I will
get twice the alpha because I will get the alpha from
the outperformers and the alpha from shorting the
underperformers," a cynic will say, "If they cannot
pick winners, why should they be able to pick losers?
So they have twice the risk."
The second application for derivatives is called
synthetic equity funds, which uses aggressive cash
handle its monthly outflows using a protective put
strategy, because we are concerned about the down-
side in any particular 12-monthperiod. The Founda-
tion wants to have the cash flows necessary to write
the checks each month to those that depend on them.
Using exchange-traded puts, we replicate a strip of
monthly put options on the portfolio to protect the
projected cash flows 12 months in advance, because
we currently do not have direct access to an aTe
market. The options replication is done in-house,
using off-the-shelf software.
Derivatives Applications
The size of Amoco's U.S. pension fund is about $2.2
billion. Our plan was larger until we laid off about
8,500 employees and they all wanted lump-sumpay-
ments. Table 1 shows the asset mix for the fund. Our _
policy reflects a fairly traditional asset mix except
that our total equity exposure is fairly high for a U.S.
pension fund. The only major asset allocation bet is
underweighting international equities, which has
been a sound decision for the past couple of years.
So far in 1993, however, avoiding international equi-
ties may not have been such a good bet. People are
still cautious about the international market, perhaps
because of Japan's problems.
Three-fourths of Amoco's pension fund is man-
aged externally, and many of these managers are
fairly traditional. They do fundamental stock pick-
ing, country allocating, sector bets, and duration
bets. For the 25 percent managed internally, we use
an S&P 500 benchmark and a blend of fundamental
analysis and derivatives. Our external managers
give us daily reports on all of their positions, and we
look at performance daily. This practice is unusual;
most plans look at performance two months after the
last quarter.
We also manage a portfolio for the Amoco Foun-
dation, which is a charitable organization used to
further Amoco community service objectives. We
The Amoco pension fund uses derivative strategies
to achieve some of its objectives. I will use Amoco to
illustrate howa pension fund can use derivatives and
to share what our experience with derivatives has
been.
Fund Management
120
Derivatives are new to the Amoco pension fund and
to the corporation itself. Much of our experience has
been in trying to introduce derivatives into a setting
in which they are totally alien. With a pension fund,
you must consider not only yourself and the people
you work with but also the trustees. Most major
trustees are sophisticated in handling futures, op-
tions, and derivative securities, but the people on
your specific account may not know what you are
talking about. You must spend a lot of time educat-
ing them. Our experience with derivatives falls into
four areas.
Administrative challenges. Using derivatives
creates several administrative challenges-specific-
ally, in the areas of reporting, accounting, and tradi-
tional portfolio management systems. Although tra-
dition?l systems record such measures as profit and
loss in an appropriate manner, they are not good for
measuring risk exposures. For example, some port-
folio management systems mark the futures basically
to zero, so the portfolio seems to have a lot of cash
and no equity. The challenge is to find a system that
will factor in a portfolio's risk.
Dealing with outside auditors is another prob-
lem. Many of our auditors are used to auditing only
long-position transactions. In the market-neutral
portfolios, half the positions are short positions. The
auditors cannot audit short positions, futures trans-
actions, or many of the structured notes we have, so
age the cashcomponent in a fairly aggressive manner
using mortgage-backed securities or callable corpo-
rate bonds that have already been called. A fourth
application is currency hedging.
Structurednotes canbe used to avoid restrictions
on the use of derivatives. For example, we currently
do not have access to OTe products, but we can buy
notes from certain issuers. When these notes involve
structured coupons, we canbuy themand get around
the guidelines. This approach actually increases
credit risk and reduces return, but it does provide
additional investment flexibility.
Another application of futures in a typical pen-
sion fund setting is to gain control of the aggregate
portfolio, which is what we are trying to manage.
Our external managers may be making decisions
they think are optimal for their own slice of the
portfolio, but our mandate is to manage the aggre-
gate optimally, which is sometimes a problem. For
example, when we get the reports on the managers'
trades every morning, we may find that one manager
is selling a stock and another manager is buying it.
For us, two commissions just walked out the door.
Policy
55%
15
15
5
5
--.2-
100%
65%
8
15
5
4
-----..L
100%
Current Asset Class
Equities
Domestic
International
Fixed-income
Limited partnerships
Real estate
Cash
Total
Table 1. Amoco U.S. Pension Asset Mix
Source: Amoco Corporation.
management plus futures. We are trying to beat the
S&P 500 by 200 basis points. The traditional way to
add value is to pick stocks or sectors, but we use
derivatives and fixed-income-type approaches to
achieve this objective. One strategy is to look for
investments that will beat LIBOR by, say, 200 basis
points. Then, put a future on top of that and get the
S&P plus 200 basis points. For example, if you think
u.s. interest rates will be flat, stable, or trending
down, instead of holding a LIBOR-based instrument,
you could hold, say, two-year Treasuries to get the
extra yield. If rates fall, you get capital appreciation.
Even if rates rise a little, the investment is somewhat ----------------------
self-hedged because the two-year Treasuries will roll Experience with Derivatives
down the yield curve. If the yield curve is unusually
steep, that should be in your favor. In the current
environment, this method is a fairly low-risk way to
add value in a different sense than is usually meant
and also to use derivative products.
Other fixed-income instruments we have used in
our portfolio include some of the interest floaters.
We have used a GoldmanSachs product-aninverse
floater related to average rates in the G-7 countries.
This is a simple way to bet on lower rates on a global
basis and add value to a u.s. equity portfolio. Also
beginning to look attractive is emerging market debt.
We have done some hedged Mexican cetes (govern-
ment bills) that yield LIBOR plus 175 basis points.
The risk is that Mexico will impose currency controls
in the next couple of months.
Our portfolio is like a synthetic index fund, but
it should not be thought of as passive, because it
incorporates views and active strategies involving
risk. The value added in these techniques is coming
not so much from the use of derivatives but from the
view that portfolio managers are taking. The deriv-
atives provide the flexibility to add value in this
manner as opposed to traditional ways.
A third application of derivatives is to adjust the
duration of fixed-income portfolios. Futures are
used for this purpose because they are cheaper than
cash instruments. Many managers, instead of hold-
ing outright Treasuries, hold futures and then man-
121
increasingly large parts of our portfolio cannot be
audited. The challenge is to educate the auditors
about what we are doing and how derivatives work.
A common statement is that futures are cheap
and liquid. That is not true when a custodian is
involved, because you have to include brokers' com-
missions, market impact, and incidentals such as
National Futures Association fees. Portfolios are
charged for each cash flow out of the custodian ac-
count. If you plan to hold futures and roll them
quarterly, that means extra costs. Suppose that the
market goes down for half of the days. On those
days-or half of the time-the money will go out of
the account, and you incur a $10 or $25 charge each
time that happens. When all these charges are added
up, futures are not necessarily inexpensive.
Benchmark problems. Another challenge is
finding an appropriate benchmark. A fund manager
must receive permission to use options in the fund.
Later, the pension committee asks the manager to
prove the options added value. This is virtually
impossible. Separated from the underlying asset, the
profit and loss on an option position is essentially
meaningless. The two cannot be separated. You
cannot separate a covered call position from a stock
position and say you made or lost money on the call,
because the performance of the option is tied to how
the underlying stock did. We spent a lot of time
trying to develop some heuristics as to what portfolio
managers would have done had they not had access
to options. To show that we added value, we built a
big table comparing the returns on portfolios using
our various derivatives strategies with returns on a
portfolio using none of these strategies.
Education. Education is very important.
One cannot assume people know about derivatives,
even those who attend conferences or hear it from
brokers. For example, I know someone who swore
up and down he knew all about options, had at-
tended classes at the Merc, and had seen all the
hockey-stick diagrams. We did a covered call in the
portfolio, and the stock went down a dollar, but the
option only went down half a point. He told us
something was wrong; we had sold the wrong op-
tion, because it was down only 50 cents when the
stock was down $1. We realized he had seen the
hockey-stick diagrams and believed that was the
way they trade. He did not understand all the inter-
mediate steps. He thought if a call was in the money,
then it traded dollar for dollar; if it was out of the
money, then the price change was zero. It is impossi-
ble to underestimate what biases or ignorances at-
tach to even basic products.
This lack of understanding is also true of senior
management. I have talked with many corporate
122
treasury departments about how they use deriva-
tives for corporate risk management, interest rates,
currencies, and pension funds. Almost universally,
senior management is not aware of how these prod-
ucts work, but they trust the people using them to do
a good job-although some places will not trust you
that much.
Image. The fourth problem is the image de-
rivative products have. For example, a Wall Street
Journal article said "new derivative products are sur-
prisingly complex." The article discusses so-called
derivative products put together with financial and
mathematical alchemy. On the same page, however,
was an article entitled, "What price CMOs? Funds
have no idea how to price CMOs." Nobody says we
must look at the CMO market, which is probably a
huge market. Nobody says we must put circuit
breakers in the CMO market. Why do they treat
derivatives differently?
The other image that worries senior manage-
ment is traders who run amok. They have seen
stories about traders who trade $1 billion, put the
tickets in the drawer, andleave them there for several
months. Meanwhile, the company loses millions of
dollars. That problem is difficult to get around. No
matter what controls or guidelines are in place, some-
one will always get around them. It goes back to the
trust issue, which is a very difficult one to determine.
Other image problems involve OTC versus ex-
change-traded instruments. Some people say ex-
change-traded futures are deep in liquidity, but
many are not. The S&P 500 is deep in liquidity, but
try to get a quote on the Russell 2000 future; it will be
a one-up market and probably two points wide.
There is an illusion that pension funds do only large
trades; $2.2 billion is a lot of money, but we have
many ideas for strategies we would like to do in
much smaller sizes than most people want to do in
the OTC market. We seem to be too big for some
exchange-traded instruments but too small for the
OTCmarket.
Similarly, people believe that the OTC market
provides a lot of customization. In our experience,
that is not true. We had a case on the corporate side
in which we were trying to get a price on a put option
on what was basically a blend of gold and copper.
Nobody wanted to quote us a price. Instead, we got
evasive answers: "We do not want to price this," "We
are not selling that this week," or "Our lawyers do
not want us to talk about it because it might be
construed as an offering." We ended up calling
somebody in Paris to get a price. Some of the capa-
bilities of the OTC market are hyped up, and al-
though some of the products offered look good, they
are probably expensive and not something we would
want to use.
On the corporate side, the Arkansas-Best case,
which involved the tax treatment of some corporate
hedging activities, will inhibit corporate use of deriv-
atives, because the accounting and tax implications
are a major factor in a corporation's decision as to
how it will hedge. Multinationals have different tax
status in different countries. They like to play one tax
code off against another. In a case in Indiana, the
members of a grain co-op sued their board for not
hedging, because they lost a lot of money. They won
their suit and were upheld in the state court of ap-
peals. This verdict, if carried through, could have a
phenomenal impact on the investment industry. The
day may come when it will be considered imprudent
not to use derivatives.
Conclusion
Going forward, we expect to expand our use of syn-
thetic funds and get into more of the swap markets,
especially international. International swaps help
avoid much of the cost associated with ordinary cash
investments, making swaps or derivative-based
strategies appealing. The other implication is that
the use of derivatives leads to an asset classless view
of the world. All we are looking for is value added;
we do not care where it comes from. An equity
manager who can beat the S&P can swap the S&P
return and get a bond return or an international
return and essentially create whatever portfolio with
whatever value added is expected.
Aside from being an enemy, risk can be a friend;
an example would be collar-type strategies. As a
portfolio manager, my goal is to beat the S&P by 200
basis points. It does not do me any good to beat it by
300-400basis points, but it would do me a lot of harm
to underperform. I would be happy to sell off some
of that return to somebody else and buy downside
protection. The only way this can be done is through
derivative markets, because the pricing of the op-
tions in a collar strategy comes from the volatility of
the underlying asset. To access that price, you must
have access to derivatives; buying the downside pro-
tection you want is also a derivatives transaction.
123
Question and Answer Session
Matthew R. Smith
Question: Given the perfor-
mance measurement problems,
how have you convinced your
board to let you use derivatives?
Smith: Two strategies helped in
the transition: the market-neutral
strategy and the synthetic fund,
or aggressive cash management
with futures, because we could
clearly identify the benchmark as
124
the S&P 500. Identifying the
value added is easy with that
type of strategy. We have not
convinced them as to what value
the options strategies added.
Question: Do your external
managers use derivatives? If so,
do you provide guidelines on
how derivatives are to be used?
Smith: Mostly they use them
for duration adjustments and
currency hedging. We do not
provide them with any formal
guidelines except that, because
we are a pension fund, they can-
not be leveraged, whatever
"leverage" means when you
are using derivatives.
Self-Evaluation Examination
l. Luskin argues that many market participants 4. Which of the following is not a means for re-
incorrectly view derivatives as a separate asset ducing the credit exposure on an interest rate
class. Which of the following best summarizes swap agreement?
his reason for making this statement? a. Collateralization agreements.
a. Derivatives are traded in separate mar- b. Special-purpose vehicles.
kets from equity and debt, with different c. Mark-to-market contracting.
pricing conventions. d. Varying notional principal.
b. Derivatives allow for the opportunity to
unbundle the risk embedded in tradi-
5. If a manager who currently has a position in
tionaI securities, a choice most investors
real estate assets wishes to convert a portion of
do not like to have.
these holdings into a fixed-rate government
c. Derivatives can be structured for both do-
and corporate bond portfolio, he or she could:
mestic and international applications,
a. Enter into two bond index swaps in
while stock and bond investments have
which he or she receives the index rate
only a single dimension.
and pays LIBOR.
d. Both the derivatives industry and traders
b. Enter into two bond index swaps in
in that industry tend to be younger than
which he or she receives LIBOR and pays
the typical derivatives customer.
the index rate.
c. Enter two swaps, one paying the return
2. What best explains why brokers were able to
on a mortgage security index and receiv-
offer clients what appeared to be extraordinary
ing LIBOR and the other receiving the
returns on swap deals linked to foreign equity
return on a bond index and paying
indexes in the early days of that market?
LIBOR.
a. Foreign equity markets are generally less
d. Enter into two swaps, one receiving the
efficient than u.s. equity markets.
return on a mortgage security index and
b. Brokers had the incentive to "give deals
paying LIBOR and the other paying the
away" in order to start aneventually prof-
return on a bond index and receiving
itable business.
LIBOR.
c. Brokers were willing to prorate their sav-
ings from strategies they adopted to avoid
6. The swap structure with an interest rate sensi-
dividend income taxes.
tivity most closely resembling that of a mort-
d. The returns were not really extraordinary;
gage-linked security is:
clients were paying in the form of hidden
a. An arrears swap.
fees and commissions.
b. An index-amortization swap.
c. A "diff" swap.
3. An exchange-traded interest rate call option
d. A yield-curve swap.
with a strike price of 93.50 and premium of 27
7. If an insurance firm has $1 million in fixed-rate
basis points would provide an effective interest
rate floor of:
assets with a duration of 10 years and $0.8
a. 6.23 percent.
million in liabilities with a duration of 8 years,
b. 6.27 percent.
what hedge is needed to eliminate the
c. 6.50 percent.
company's interest rate exposure if the contract
d. 6.77 percent.
it is using has a face value of $0.1 million and a
duration of 4.5 years?
a. Short 8 contracts.
b. Long 8 contracts.
c. Short 18 contracts.
d. Long 18 contracts.
126
8. According to McMillan, which of the following 12. What percentage of the world's equity market
is not a component of the modern asset/liabil- capitalization (as measured by the FT-Actuar-
ity problem for an insurance company? ies World Index) is covered by index futures
a. Designing an asset/liability portfolio that
contracts?
hedges the company's short straddle posi-
a. 35 percent.
tion.
b. 55 percent.
b. Designing an accounting system that is
c. 75 percent.
capable of speeding up collection of re-
d. 95 percent.
ceivables and slowing down payments
13. Suppose that a portfolio manager expects stock
owed by the firm.
c. Designing and implementing a r a h ~ - c r e d -
prices to rise in the near future but is concerned
iting strategy to optimize the value of the
about downside risk. Accordingly, she wants
company's rate reset option. to reduce her exposure to the S&P 500 Index by
d. Designing marketable products that re- using an index-derivative-based hedging strat-
flect the cost of embedded call and put egy. Which of the following is likely to be the
options. least successful in accomplishing her goals?
a. Buying an index put option and selling an
9. Suppose the effective duration of a 30-year non-
index call option.
callable bond is presently 11 years. If a five-
b. Buying an index put option and selling
year call option with a strike price at par value
another index put option that is farther
is attached to this instrument, how would the
out of the money.
duration of the overall position be changed?
c. Shorting an index futures contract on a
a. It would be reduced because of the posi-
portion of the stock that is held.
tive convexity effect of the embedded call.
d. Buying an index put option.
b. It would be increased because of the pos-
itive convexity effect of the embedded 14. While the total swap market has outstanding
call. notional principal of about $4 trillion, equity
c. It would be increased because of the neg-
swaps contribute only about $40 billion of this
ative convexity effect of the embedded
amount. What factor best explains the histori-
call.
cally small growth of the equity swap market?
d. It would be reduced because of the nega-
a. Equity swaps are generally difficult to
tive convexity effect of the embedded call.
price in an efficient market.
b. Unlike interest rate swaps, equity swaps
10. A tactical (as distinct from a strategic) decision
entail credit risk.
using derivatives is one that:
c. Until recently, Internal Revenue Service
a. Involves an investor's "big picture" of the
rulings inhibited tax-exempt investors
financial landscape.
from using this product.
b. Requires an investor to act as a quasi-mar-
d. Market regulations prohibit dealers from
ketmaker.
c. Involves an application attempting to ex-
quoting both interest rate and equity
ploit a specific market condition.
swaps to the same clients.
d. Always involves an attempt to arbitrage
some aspect of the investment market.
11. Why is the standard deviation an inappropriate
way to measure the risk of an option or an
option-linked position?
a. Because it is a symmetric measure.
b. Because it involves squared deviations.
c. Because it is not what Markowitz recom-
mended.
d. Because it ignores"downside" risk.
127
15. Suppose that a global equity manager starts 18. If the interest rate in Country A is higher than
with a portfolio having the following composi- that in Country B for default-free securities
tion: 50 percent U.s. stocks, 35 percent Japanese with the same maturity, the currency in Coun-
stocks, and 15 percent U.S. cash equivalents. If
try A will:
he then takes a long position in the Nikkei 225
a. Trade at a forward discount to the cur-
Index representing 15percent of the entire port-
rency in Country B.
folio, what will be the resulting allocation?
b. Trade at par value to the currency in
a. 50 percent U.S. stocks, 50 percent cur-
Country B.
rency-exposed Japanese stocks.
c. Trade at a forward premium to the cur-
b. 50 percent U.s. stocks, 35 percent cur-
rency in Country B.
rency-exposed Japanese stocks, 15 per-
d. Stand in any of the relationships to the
cent currency-hedged Japanese stocks.
currency in Country B, depending on rel-
e. 35 percent U.S. stocks, 50 percent cur-
ative trade imbalances.
rency-exposed Japanese stocks, 15 per-
19. Which of the following indicates the presence
cent U.s. cash equivalents.
d. 35 percent U.s. stocks, 35 percent cur-
of a trend (rather than a reversal or random
rency-exposed Japanese stocks, 15 per-
movement) in a time series of currency returns?
cent currency-hedged Japanese stocks, 15
a. A variance ratio that is significantly less
percent U.s. cash equivalents.
than 1.
b. A variance ratio that is significantly
16. In a global equityportfolio, which of the follow-
greater than 1.
ing would not be a way to protect a specific
c. A variance ratio that is not significantly
country exposure against downside risk?
different from 1.
a. Selling a call option and buying a put
d. A variance ratio that is not significantly
option on that country's stock index.
different from zero.
b. Buying an out-of-the-money put option
on that country's stock index.
20. Ramaswami argues that active currency man-
c. Buying a call option and selling a put
agement can be profitable because:
spread on that country's stock index.
a. The currency market is less efficient than
d. Sellinga put option witha lowstrike price
the stock market.
and buying a put option with a high strike
b. The U.s. dollar has been relatively weak
price on that country's stock index.
on world markets during the past decade,
so shorting the dollar should always be
17. The advantage of hedging an investment port-
successful.
folio that has exposures to multiple foreign ex-
e. Currency returns must go up when the
change rates with a currency basket hedge is
value of the underlying asset portfolio
that:
goes down.
a. A forward contract on the EAFE index is
d. The trends in currency returns, although
always the best one to use.
nonlinear, can be exploited by specially
b. A certain amount of currency exposure
designed filter trading rules.
should always be unhedged because for-
21. In the class of nonstandard ("exotic") options,
eign exchange risk is a zero-sum game.
e. Derivative products will never be avail-
single-asset/ multiple-tenor contracts are those
able for all the currencies in a portfolio.
that involve only one underlying asset but con-
d. It will be cheaper than hedgingeach of the
sider values at several different dates. Select
exposures independently.
from the following list the contract that does
not fit this definition.
a. Outperformance options.
b. Barrier options.
c. Compound options.
d. Average-rate optiolils.
128
22. Consider two different option contracts on the
same underlying asset. The options are alike in
all respects except that the terminal return to
Contract A is based on the asset's price at the
expiration date while the terminal return to
Contract Bis based on an average of the asset's
prices during the holding period. In this case,
Contract A:
a. Will be less expensive than Contract B
because the latter is an exotic derivative.
b. Will be more expensive than Contract B
because its volatility will be greater.
c. Will have the same price as Contract B
because they involve the same asset over
the same amount of time.
d. May be either more or less expensive than
Contract B, depending on the composi-
tion of the investor's overall portfolio.
23. Nederlof argues that even though a covered
call position might generate a higher expected
return and lower standard deviation than a
protective put position on the same stock, the
former is actually riskier than the latter be-
cause:
a. Covered calls have an asymmetric return
distribution while protective puts do not.
b. The standard deviation is not an appro-
priate way of comparing two strategies
that generate different expected returns.
c. Buying rather than selling an option al-
ways reduces price volatility in any port-
folio.
d. Protective puts limit downside losses, and
covered calls limit upside gain potential.
24. Select from the following choices the only de-
rivative-based strategy that could be designed
to give an equity manager protection against
downside losses, some variability in potential
returns, and no up-front cost.
a. Acquiring a protective put option.
b. Going long in a forward contract de-
signed around a portfolio identical to the
one the manager holds.
c. Buying an out-of-the-money put and sell-
ing an out-of-the-money call.
d. Writing a covered call option and rein-
vesting the proceeds ina risk-free security.
25. Froma plansponsor's perspective, whichof the
following best describes the administrative
challenge of implementing a derivative-based
trading program?
a. Selecting an appropriate benchmark for a
portfolio that includes options.
b. Establishing accounting systems capable
of valuing derivative strategies fairly.
c. Educating both the plan trustees and the
firm's senior management as to what de-
rivatives are and how they can be used.
d. Overcoming the image problems associ-
ated with trading in derivative markets.
129
Self-Evaluation Answers
l. b. By offering the opporhrnity to unbundle 11. a. Gastineau documents that option-based
the risk of an underlying position, deriva- investments generate inherently asym-
tives force investors to be aware of that metric return distribution patterns, mak-
risk. Investors can ignore the risk by treat- ing a symmetric statistic such as the stan-
ing the derivative as a separate position. dard deviation an inappropriate measure
2. Luskin notes that tax-exempt U.s. holders
of uncertainty.
c.
of foreign securities are subject to divi-
12. d. Table 2 in Hill's presentation shows that
dend withholding tax, so dealers who can
countries having either actual or syn-
find a way to avoid those taxes them-
thetic stock index futures contracts com-
selves could pass a portion of the profit
prise 95.3 percent of the FT-A World
along through the swap.
Index.
3. a. Kawaller argues that the effective floor for
13. a. As Hill explains in her Case Study A, the
an interest rate call option can be calcu-
zero-premium collar would limit the
lated as (100 - strike price - option pre-
manager's upside potential too much to
mium).
be acceptable.
4. d. In Brown's discussion of swap-related
14. c. In his second presentation, Gastineau
credit risk, only varying notional princi-
notes that, until July 1992, the IRS treated
pal is not mentioned as a way to reduce
receipts from equity swaps as unrelated
this exposure.
business income, which is fully taxable.
5. c. This transformation is illustrated in Ex-
15. b. Clarke's Exhibit 1 shows that, in the ab-
hibit 2 of Kopprasch's presentation.
sence of a separate currency position, the
6. b. Kopprasch notes that an index-amortiza-
Nikkei futures contract will convert the
tion swap is typically structured to have
cash equivalents into a position with Jap-
the same prepayment dynamics as a
anese equity exposure but denominated
mortgage when interest rates change but
in dollars
it removes the exposure to other prepay-
16. c. This position is exactly the opposite of
ment rationales.
what Clarke calls an option collar and
7. a. According to the formula and analysis would actually enhance the portfolio's
McMillan provides, this firm is exposed downside country exposure.
to rising interest rates and would require
17. d. Citing the work of Markowitz, DeRosa's
a short position of eight futures contracts
to offset this risk.
point is that some of the myriad currency
exposures will naturallyoffset each other,
8. b. Although such anaccounting systemmay making it most economical to hedge the
be desirable for other reasons, it is not one entire foreign exchange risk as a portfolio.
of the components of a modern asset/lia-
18. As DeRosa explains, trading based on the
bility management programdiscussed by
a.
McMillan.
covered interest parity theorem will en-
sure that the currency of the high-interest-
9. d. McAdams provides a detailed example in
rate country trades at a forward discount
his presentation of how, from an
to equalize returns and prevent arbitrage
investor's perspective, the effective short
opportunities.
position in the embedded call option
19. b. In Ramaswami's study, currency return
shortens the duration of the bond.
Gastineau's first presentation stresses
trends were indicated by variance ratios
10. c.
that exceeded 1; reversals required those
that tactical decisions are those that have
ratios to be less than l.
a specific purpose such as dividend cap-
ture or yield enhancement.
130
20. d. Ramaswami demonstrates that currency 23. d. Both strategies produce asymmetric re-
return trends are convex in nature and turn distributions, but only the protective
can be exploited by option-based trading put truncates downside risk, which
strategies. makes a covered call position much more
21. a. Reiner notes that outperformance options
susceptible to losses.
ultimately involve a choice amongseveral 24. c. Nederlof gives an example of such a
asset classes rather than just one. hedge, which is called a range-forward
22. b. In his description of path-dependent op-
position.
tion valuation, Reiner explains that the 25. b. Although all of these responses represent
price-averaging process reduces both the legitimate concerns, only the accounting
volatility and price of a standard contract. system problem falls in the category of
what Smith calls an administrative
challenge.
131