Professional Documents
Culture Documents
Contents
Editorial 2
Professor Dr Tan Hui Boon
How sticky is the adjustment of retail interest rates in Malaysia? 3
Empirical evidence from commercial banks and nance companies
Dr Muhamed Zulkhibri Abdul Majid
Market orientation, cultural intelligence and loan performance: 22
A study in Malaysian retail banking
Dr Voon Boo Ho & Carolin Ann
An empirical assessment of product-level home lending interest rates 30
Evidence from National Australia Bank
Dr Harry M. Karamujic
Impact of oil price shocks on government revenue and expenditure: 51
Evidence for Malaysia
Dr Tan Juat Hong
Bankers Journal Malaysia
The Journal of the Institute of Bankers Malaysia
Issue No. 134 June 2010
KDN PP 3781/03/2011(026549)
ISN 0126-9534
2
Editorial
Financial intermediation
and economic growth
The global economic recovery strengthened further during the rst quarter of 2010, led by the robust growth
in the East Asian emerging economies. Malaysia, among others, has recorded a double digit growth of
10.1% in that quarter. Unfortunately, the Euro-zone turmoil triggered by the sovereign debt crisis has cast
doubts on the sustainability of the recovery.
Nevertheless, recent reports by Bank Negara Malaysia (BNM) and the Malaysian Institute of Economic
Research (MIER) show that our countrys economic growth remains rmly on track supported by strong
private domestic consumption and investment as well as public investment spending. In addition, a strong,
resilient and efcient nancial system that raises the level of investment and savings, and enables efcient
allocation of funds is also crucial for the economy. It is vital to safeguard nancial stability which ensures
the process of intermediation to continue uninterrupted during the period of recovery. BNM has taken
measures to foster a sound and efcient banking system that responds effectively to external shocks and
the changing needs of domestic markets.
This issue of the Bankers Journal Malaysia features a few studies relating to nancial intermediaries. The
rst paper, by Dr Muhamed Zulkhibri Abdul Majid, investigates the interest-rate pass-through from money
market rates to various retail lending and deposit rates in Malaysia. The empirical evidence provided by the
study indicates that the vast majority of retail lending rates pass-through is less than complete, and the speed
of adjustment differs across various retail rates. Commercial banks, as large nancial institutions, adjust their
rates to uctuations in the market rates faster than their smaller counterparts (nancial companies). These
smaller nancial institutions seemed to be more dependent on bank credit and compete less rigorously
in the market.
Dr Voon Boo Ho and Carolin Ann, in the second paper, discuss the relationship between market orientation
and loan performance of Malaysian retail banks. Their results show that market orientation affects the
loan performance of retail banking. This relationship is positively moderated by the cultural intelligence
of credit employees. Within the three principal dimensions of market orientation, customer orientation is
found to play a relatively more vital role than competitor orientation and inter-functional coordination in
inuencing loan performance.
The third paper, by Dr Harry M Karamujic, seeks to examine the movement of Australian product-level home
lending interest rates using a structural time series modeling approach. His results conrm the presence
of cyclicality in the interest rates. Besides, the rates also exhibit signicant seasonality. Nonetheless, the
majority of the observed statistically signicant seasonal factors cannot be attributed to any of the three
seasonal effects, namely the spring, autumn and end of the nancial year effects.
Last, but not least, the paper by Dr Tan Juat Hong determines the short-run impact of fuel price uctuations
on the Malaysian scal budget through two approaches, the impulse response functions and variance
decomposition. The empirical estimations show that both government revenue and expenditure are affected
signicantly by world fuel price volatility. A positive fuel price shock contributes a greater variability to
government expenditure, whereas a negative fuel price shock exerts a greater impact on government
revenue. However, such impacts are only short-run phenomena.
Professor Dr Tan Hui Boon
Nottingham University Business School
The University of Nottingham Malaysia Campus
3
How sticky is the adjustment of retail interest
rates in Malaysia? Empirical evidence from
commercial banks and nance companies
Dr Muhamed Zulkhibri Abdul Majid *
Senior Economist, Department of Monetary and Financial Policy, Bank Negara Malaysia
ABSTRACT This paper examines the interest rate pass-through from money market rates to various
retail lending and deposit rates for nancial institutions in Malaysia. The evidence shows that the vast
majority of retail lending rates pass-through is less than complete, while the speed of adjustment varies
across administered interest rates. Adjustment in lending rates tended to be more sluggish than that of
deposit rates. Finance companies are quicker in adjusting their deposit rates than the commercial banks,
but are slower in adjusting their loan rates. Empirical analysis also shows that interest rate adjustment
is asymmetric and faster in a period of monetary easing rather than in one of monetary tightening. The
evidence suggests the importance of nancial institutions in the transmission of monetary policy, reect-
ing a situation where the adjustment processes are not uniform across different types of institutions and
instruments.
THE PASS-THROUGH FROM policy rate and many market rates to retail interest rates has
attracted enormous attention in the light of competition in the banking sector over the past
few years. The pass-through has been discussed, in particular, in the context of monetary
policy since the transmission of money market rates to bank retail rates is an important
element in the monetary transmission process. The speed and symmetry of price movements
in response to changes in market conditions and macroeconomic shocks affect economic
efciency through misallocation of costs when prices are not in equilibrium. Ideally, changes
in the policy rate should be completely passed through to money market and retail rates
over a reasonably short horizon. Thus, the effectiveness of monetary policy depends on the
existence of interest rate pass-through, both the degree and speed of pass-through of short-
term money market rates to retail rates (Bredin et al, 2001).
The motivation for this study comes from the fact that, despite the importance of the micro
aspect of the monetary transmission mechanism, only a few in-depth micro studies have
been performed in this area of research in emerging market economies. Even though the
empirical literature on interest rate transmission presents diverse and sometimes conicting
estimates, discussing the methodological and specication related issues may contribute
to the understanding of these differences. In order to appraise the relationship between
monetary policy and the role of nancial institutions, it is essential to examine the whole
chain of transmission from the policy rate via the money market rates to the bank retail
rates. In addition, understanding the interest rate pass-through in the nancial market can
shed some insight into the underlying mechanism of monetary transmission and may provide
valuable complementary evidence explaining the credit channel.
This paper attempts to provide empirical evidence on interest rate pass-through in Malaysia
along with nancial market reform and market liberalisation by addressing four important
questions. Firstly, to what extent are changes in the short-term money market rates passed
through to various banks retail rates and the speed at which changes in these money
market rates are transmitted to the various retail interest rates. Secondly, how does the
institutional distinction between banks and non-banks inuence the speed of the interest
rate pass-through? Thirdly, how do the nancial market reforms over the years impact on
the underlying process of monetary transmission? Finally, unlike other studies of developed
nancial markets, this paper also explores the reasons why banks in relatively well-developed
* I would like to thank Prof P Mizen, Prof A Guariglia, Prof A Mullineux and Prof S Bougheas for their valuable
comments. This paper also benetted from comments made by seminar participants from Bank Negara Malaysia
and the University of Nottingham. The views expressed in this working paper are those of the author and do not
necessarily represent those of the institution(s).
4
nancial markets may face discontinuous
changes to ofcial rates.
This paper is set out as follows: Section
II gives an overview of the interest rates
structure and development in Malaysia.
Section III reviews related studies in the
main literature with respect to interest
rate pass-through. Section IV outlines the
approaches to the methodology. Section V
discusses the choice of data and estimation
results. Finally, Section VI provides
concluding remarks.
II. A BRIEF OVERVIEW OF THE INTEREST
RATE POLICY IN MALAYSIA
Malaysias interest rate regime experienced
its rst signicant shift towards liberalisation
in the late 1970s with the deposit and prime
lending rates of commercial banks being
freed from administrative control. Prior to
1983, market rates were still guided by the
interest rates of the lead banks, with Bank
Negara Malaysia (BNM) indirectly controlling
the general level of interest rates via banks
cost of funds. The Base Lending Rate (BLR)
framework was introduced in 1983 such
that banks were allowed to quote rates
based on their BLR, which constituted the
cost of funds. Since 1983, Malaysia has seen
several changes to the BLR computation,
with the aim of ensuring a more responsive
lending rate to changes in monetary policy.
In 1987, the BLR followed a two-lead bank
system in which each banks BLR was not
allowed to deviate by more than a certain
percentage from those of the two lead banks.
This framework was superseded by a more
liberalised BLR framework in 1991. This
new framework tied the BLR directly to a
banks cost of funds. From November 1995,
the formula for computing the BLR was
linked directly to the daily average 3-month
interbank rate. However, the transmission
effect was too slow, due mainly to its
averaging feature, which meant that any
changes to the interbank rates would have
a long lagged effect. In 1998, to address
the lagged transmission effect, the 3-month
intervention rate was used as the anchor
rate for the computation of the BLR. This
was essentially to reduce the lag effect of
changes in the policy rate. Figure 7.1 shows
the behaviour of interest rates in various
BLR frameworks.
1
Although the policy rate was not explicitly
disclosed by the central bank in the earlier
periods of interest rate liberalisation, the
3-month interbank money market rate is the
rate that has generally been accepted as the
central banks signalling rate. Bank Negara
Malaysia would conduct its money market
operations to inuence this interbank rate,
which in turn is linked to the BLR. At the same
time, the changes to the chosen interbank
rate would be transmitted across the yield
Source: Bank Negara Malaysia
1 See Appendix for details of the evolution of the interest rate regime in Malaysia.
Figure 1: Interest Rate Regimes
0
2
4
6
8
10
12
14
16
J
a
n
-
8
3
J
a
n
-
8
4
J
a
n
-
8
5
J
a
n
-
8
6
J
a
n
-
8
7
J
a
n
-
8
8
J
a
n
-
8
9
J
a
n
-
9
0
J
a
n
-
9
1
J
a
n
-
9
2
J
a
n
-
9
3
J
a
n
-
9
4
J
a
n
-
9
5
J
a
n
-
9
6
J
a
n
-
9
7
J
a
n
-
9
8
J
a
n
-
9
9
J
a
n
-
0
0
J
a
n
-
0
1
J
a
n
-
0
2
J
a
n
-
0
3
J
a
n
-
0
4
J
a
n
-
0
5
ALR
BLR
3-Month Interbank
Declared BLR +
Freely
Determined
Margin
0.5% + BLR of
Two Lead Bank
Freely
Determined Rate
Based on Cost of
Fund
BLR using 3-
month
Interbank
BLR using 3-month
Intervation
BLR using
3 month
Interbank
5
curve to the interbank rates of shorter and
longer maturities, which would inuence
the banks deposit rates and cost of funds.
Since the introduction of the intervention
rate in 1998, the BLR formula is linked
directly to BNMs 3-month intervention
rate. In April 2004, BNM announced further
changes in its monetary policy stance based
on the market mechanism through its
overnight policy rate, replacing the 3-month
intervention rate.
Table 1: Asset and Liability Structure of Commercial Banks and Finance Companies in Malaysia
as at December 2005 (in RM Million)
Commercial Banks Finance Companies
RM million % RM million %
Demand Deposits 78,059.2
14.17
- -
Saving Deposits 57,241.5
10.39
2,916.2 6.58
Fixed Deposits 290,025.9
52.64
35,690.9 80.56
Up to 1 month 91,626.1
16.63
6,250.5 14.11
> 1 2 months 9,255.1
1.68
917.5 2.07
> 2 3 months 34,131.2
6.20
3,854.2 8.70
> 3 6 months 24,222.1
4.40
3,734.4 8.43
> 6 9 months 6,002.7
1.09
1,163.0 2.62
> 9 12 months 105,900.1
19.22
12,819.2 28.93
More than 12-months 18,888.6
3.43
6,952.2 15.69
Other Deposits 64,551.8
11.72
4,548.6 10.27
Repurchase Agreements 42,935.0
7.79
168.9 0.38
Negotiable Instruments of Deposits 34,813.5
6.32
4,840.8 10.93
Total Deposits 550,929.5 72.37
1
44,305.7 64.75
1
Amount Due to 48,410.1 1,080.7
Other Liabilities and Bankers Acceptances 99,298.3 18,107.3
Capital and Reserves 62,287.0 4,927.4
Total Liabilities & Shareholder Funds 761,254.8 68,421.1
Total Loans 447,453.3 58.77
2
56,089.9 81.98
2
Hire Purchase 30,789.2 6.88 26,950.1 48.05
Housing 91,211.8 20.38 5,186.6 9.25
Other Loans and Advances 325,452.3 72.73 23,953.2 42.71
Amount Due from other nancial institutions 130,540.4 4,071.5
Securities 74,096.4 3,227.8
Deposits Placed and Reverse Repos 34,061.1 2,754.4
Negotiable Instruments of Deposit 6,554.3 440.8
Cash 3,737.0 79.7
Fixed Assets & Other Assets 64,812.3 2,951.5
Total Assets 761,254.8 68,421.1
1 Computed based on total deposits divided by total liabilities and shareholders funds.
2 Computed based on total loans divided by total assets.
Source: Bank Negara Malaysia
The retail rate setting can be classied by
three major types of nancial institutions
namely commercial banks, nance
companies and merchant/investment
banks. Table 1 provides a comparison
of the asset and liability structure of the
commercial banks and nance companies
(merchant/investment banks were excluded
due to their small contribution to the banking
systems overall lending and deposit-taking
activities). In terms of size, commercial banks
6
On a similar line of research, Crespo-
Cuaresma et al (2004) and Tieman (2004)
study interest rate pass-through in Central
European countries. They nd that there
are signicant differences across market
interest rates and countries concerning
long-run elasticity of market interest rates
to changes in the key policy rate, but that
the pass-through is more pronounced in
recent years. Espinosa-Vega and Rebucci
(2003) compare Chile and several developed
countries and they nd that the size of pass-
through in Chile is smaller than the pass-
through in developed countries. Scholnick
(1996) examines the rigidity of commercial
bank interest rates, using evidence from
Singapore and Malaysia. The results
conclude that in both countries, the mean
adjustment lag is shorter when the deposit
rate is above its equilibrium than when it
is below its equilibrium. However, the study
only covers the earlier part of the interest
rate liberalisation period.
The second group of studies concentrates
their analysis in country-specic cases.
Most of the studies are concentrated in
developed countries. For instance, studies
by Winker (1999) and Weth (2002) for
Germany, Manzano and Galmez (1996) for
Spain, Bredin et al (2001) for Ireland, De
Bondt (2005) for the euro area, Neumark and
Sharpe (1992) for the USA, Lim (2001) for
Australia, Hofmann and Mizen (2004) for the
UK and Gambacorta (2005) for Italy. These
studies reveal the evidence of structural
differences in the interest rate pass-through
for specic countries and across different
types of nancial institutions. From these
studies, in general, the speed at which bank
lending and deposit rates adjust to changes
in market rates among others is related
to the institutions size, its renancing
conditions and the extent of its business
with non-banks.
Specically, in the case of developed
countries, Cottarelli et al (1995) explore why
the transmission of monetary policy rate is so
slow in Italy. They nd that the high degree
of stickiness is explained by the constraints
to competition in the nancial system.
Moazzami (1999) conrms that interest
rate stickiness in the US is higher than in
Canada during the 1970s and 1980s. The
US lending rate rigidity, measured by the
impact multiplier however has decreased in
recent years when compared to the rst half
of the 1990s. Hofmann and Mizen (2004)
use the Monti-Klein model to explain the
differences in pass-through on 13 deposit
and mortgage rates of banks and building
are much larger than nance companies.
Their major sources of funds are deposits
from individuals and institutions, which
account for more than 72% of total funding,
whereas deposits account for only 65% of
the nance companies total funding. On the
other hand, the nance companies major
use of funds are loans, which account for
82% of their total assets while commercial
banks have more diversied portfolios of
assets with loans accounting for only 59%
of total assets.
III. REVIEW OF RELATED LITERATURE
In the empirical literature, there are two
types of studies on interest rate pass-
through, those that analyse interest rate
transmission using cross-country data and
those that give evidence using time-series
data for specic countries. The rst group
of research computes impacts and long-run
effects for different countries and relates its
ndings with differences in the countrys
nancial structures and macroeconomic
environments. The second group, based
generally on country case studies, examines
the differences in the pass-through arising
from different types of interest rates for
various nancial institutions operating in
the country-specic nancial system.
Cross-country studies for the euro area
and OECD countries from the mid-1990s
broadly show that changes in the ofcial or
money market rates are not fully reected in
short-term bank lending rates to enterprises
after 3 months, but that the pass-through
is higher for longer-term rates (Bank for
International Settlements, 1994; Cottarelli
& Kourelis, 1994; Borio & Fritz 1995).
Recent cross-country studies by Mojon
(2000), Kleimeier & Sander (2002), Donnay
& Degryse (2001), Toolsema et al (2001),
Sorensen and Werner (2006), and Kwapil
& Schaler (2006) also nd short-term
sluggishness and a large heterogeneity in
the pass-through of market rates to bank
rates between euro area countries.
Egert et al (2008) analysed the size of the
interest rate pass-through in ve Central
and Eastern European countries the
Czech Republic, Hungary, Poland, Slovakia
and Slovenia. The results conrm earlier
ndings in the literature that the pass-
through is generally very low for overnight
deposit rates, but becomes substantially
higher for short- to long-term deposit rates.
Corporate lending rates are much more
responsive to changes in the policy rate than
deposit or household loan rates.
7
societies in the UK. They nd that deposit
rates adjust fully in the long run whereas
mortgage rates do not. On the other hand,
in the case of emerging markets, there is
limited research literature which include,
among others, Berstein and Fuentes (2003)
for Chile, Humala (2007) for Argentina,
Alfaro et al (2002) for Chile, Charoenseang
and Manakit (2007) for Thailand and
Horvath et al (2004) for Hungary. Studies
in emerging markets show that the pass-
through of interest rates is generally lower
than in the developed economy.
The study on pass-through using micro data
with the panel method is also very limited.
De Graeve et al (2007) study the interest
rate pass-through for Belgium. They nd a
similar conclusion that the pass-through is
relatively slow and less than complete with
little evidence of asymmetric adjustment.
Mueller-Spahn (2008) studies the pass-
through from money and capital market
rates to bank retail rates in Germany,
taking into consideration the heterogeneity
in the nancial market. The results suggest
that pass-through in both short- and long-
run interest rates is incomplete and also
show the evidence of heterogeneity across
retail products and banks. Aydin (2007)
examines the speed and rate of adjustment
of lending rates to monetary policy rate in
Turkey for corporate, housing, cash and
automobile loans using bank-level micro
data. The estimation results reveal that the
pass-through is higher for all types of loans
but incomplete even in the long-run. The
results show that while corporate loans are
not sensitive to changes in the policy rate,
housing, cash and automobile loan rates
are.
IV. THE MODEL OF INTEREST RATE PASS-
THROUGH
A starting point for looking at the interest
rate pass-through is the mark-up between
retail rates and the money market rate. The
interest rate differential achieved at the
end of an adjustment process is called the
equilibrium mark-up, approximated by the
average loan mark-up. In a world of perfect
competition with complete information,
prices equal marginal costs and the
derivative of prices with respect to marginal
costs equals one. However, this derivative
typically becomes less than one when
the perfect competition and information
assumptions are relaxed. Applying this
idea to the banks price setting, results in
the following marginal cost pricing model
equation (Rousseas, 1985).
c o o + + =
P R
i i
1 0
[1]
where i
R
is the price the bank sets which
represents the endogenous administered
(lending or deposit) rate, i
P
is the marginal
cost price approximated by the comparable
market interest rate corresponding to
the interbank rate, which is assumed
to be exogenous. The underlying idea is
that market interest rates are the most
appropriate marginal cost prices because
of their accurate reection of the marginal
funding costs banks face, c is the disturbance
term and o
0
and o
1
are model parameters.
As discussed in Rousseas (1985), o
0
measures the constant mark-up and o
1
the
degree of pass-through in the long term.
The coefcient o
1
depends on the demand
elasticity of deposits and loans with respect
to retail interest rates. If the demand for
deposits and loans is not fully elastic,
parameter o
1
is expected to be less than one.
A demand function that is not fully elastic
and imperfect competition may lead to a less
than full pass-through. If o
1
< 1, switching
costs, information asymmetries and other
market imperfections are considered to
have caused an imperfect pass-through.
Retail bank interest rates in less competitive
or oligopolistic segments of the retail bank
market adjust incompletely and only with
delay, while bank interest rates set in a fully
competitive environment respond quickly
and completely (Laudadio, 1987).
Hannan and Berger (1991) and Cottarelli
and Kourelis (1994), argue that the
stickiness of bank lending rates as a result
of market power can also be explained by
different features of the nancial structure,
i.e. competition among banks and non-
nancial intermediaries, the degree of
capitalisation, the volatility of interest rates
and restricted entry into the banking sector
by regulatory agencies. Market power and
an inelastic demand for retail bank products
may also result from the existence of
switching costs and asymmetric information
costs. Switching costs may arise when bank
customers take into account the cost of
acquiring information and the search and
administrative costs incurred in switching
from one bank to another. On the other
hand, asymmetric information costs may
arise when there is adverse selection and
moral hazard (Stiglitz & Weiss, 1981) with
regards the setting of lending rates by
banks. If o
1
> 1, banks are considered to not
ration credit supply, but to increase lending
8
rates to compensate for higher risks (De
Bondt, 2002). Deposit demand is expected
to be relatively elastic with respect to the
banks deposits rate when close substitutes
for deposits exist. Therefore, the long-run
pass-through is complete when o
1
is equal
to one.
In the presence of xed adjustment costs
retail bank interest rates will adjust to
changes in market interest rates only if those
adjustment processes are lower than the
costs of maintaining a non-equilibrium bank
rate (Hannan & Berger, 1991). Consequently,
it is important that a time dimension is
explicitly considered in the adjustment
process of retail interest rates to changes in
market rates. The degree of market power
and asymmetric information costs are likely
to have long-term effects, while switching
costs are expected to play a role, particularly
in the short-term adjustment process. The
greater the elasticity of demand for deposits
and loans, the higher the costs of keeping
retail bank interest rates out of equilibrium.
In analysing the interest rate pass-through,
the empirical literature usually relies on
the generalisation of the following bivariate
error-correction model (Winker, 1999;
Heffernan, 1997; Scholnick, 1996):
[2]
where i
R
and i
P
are the retail rates and market
rates respectively and | stands for the long-
run pass-through. The main advantage
of this empirical approach is that it takes
into account both short-run dynamics and
the possibility of a cointegration or long-
run equilibrium relationship between bank
retail rates and market rates.
The empirical literature also looks at the
issue of asymmetry in the pass-through.
One can hypothesize that the short-term
dynamics are different depending on
whether the monetary policy rate increases
or decreases. Following Scholnick (1996)
and Egert et al (2007) the asymmetric
interest rate pass-through specication can
be written as:
( )
t t t
P
t
R
t
i u c o c o o o + + + A + = A
1 3 1 2 1 1 0
1
[3]
where the asymmetric mean lags (AML) are
dened as:
[4]
[5]
A simple test for symmetry is then given by
the Wald-test with a _
2
(1) distribution for the
following restrictions: H
0
: o
2
= o
3
. A rejection
of the null hypothesis indicates that there is
asymmetry in the speed of adjustment and/
or in short-term dynamics depending on the
direction of the change in the policy rate.
V. DATA CHOICES AND ESTIMATION
RESULTS
The monthly series of interest rate data
for both the Malaysian commercial banks
and nance companies are sourced from
the Monthly Statistical Bulletin of Bank
Negara Malaysia. The sampling period,
which is from January 1983 to December
2005, covers a time span of 23 years.
Even though the overnight rate has been
the policy rate for Malaysia since 2004,
the changes in monetary policy stance are
proxied by changes in the interbank rates
(the Malaysian 3-month interbank rate) due
to the longer time series. The retail lending
rates are the average quoted base lending
rates (BLR) and the average lending rates
(ALR) for both commercial and nance
companies while the retail deposit rates are
the savings deposit rate, 3-month, 6-month,
9-month and 12-month xed deposit rates.
Table 2 provides the descriptive statistics for
the sample data used in the analysis. The
deposit rates for both the banks and nance
companies are positively related with
maturity, indicating an upward sloping yield
curve. Moreover, nance companies tend to
offer higher deposit rates and charge higher
lending rates than the commercial banks.
This is consistent with the fact that nance
companies tend to rely heavily on deposits
as a source of funding and are perceived to
have higher default risk than commercial
banks. As there is no deposit insurance
scheme in Malaysia, nance companies,
which are much smaller than commercial
banks, have to pay a premium in order to
attract deposits.
2
We test for unit roots in money market and
retail interest rates in our sample using the
Augmented Dickey-Fuller (ADF) test (Dickey
& Fuller, 1979) and the Phillips-Perron (PP)
test (Phillips & Perron, 1988) to establish
2 Perbadanan Insurans Deposit Malaysia (PIDM) was established in September 2005 with the enactment of the PIDM
Act. PIDM is an independent statutory body that provides deposit insurance protection for depositors in the event
that a member institution is unable to repay its obligations.
9
T
a
b
l
e
2
:
D
e
s
c
r
i
p
t
i
v
e
S
t
a
t
i
s
t
i
c
s
o
f
t
h
e
I
n
t
e
r
b
a
n
k
R
a
t
e
s
a
n
d
R
e
t
a
i
l
I
n
t
e
r
e
s
t
R
a
t
e
s
C
o
m
m
e
r
c
i
a
l
B
a
n
k
s
I
n
t
e
r
b
a
n
k
R
a
t
e
s
L
e
n
d
i
n
g
R
a
t
e
s
D
e
p
o
s
i
t
R
a
t
e
s
O
v
e
r
n
i
g
h
t
3
-
m
o
n
t
h
A
v
e
r
a
g
e
R
a
t
e
B
a
s
e
R
a
t
e
S
a
v
i
n
g
3
-
m
o
n
t
h
6
-
m
o
n
t
h
9
-
m
o
n
t
h
1
2
-
m
o
n
t
h
M
e
a
n
5
.
6
3
6
.
3
9
9
.
4
9
8
.
5
5
6
.
1
9
6
.
1
9
6
.
3
5
6
.
4
6
6
.
6
7
M
e
d
i
a
n
5
.
9
4
6
.
8
5
9
.
3
9
8
.
2
1
6
.
4
6
6
.
6
7
6
.
7
4
6
.
7
8
6
.
8
9
M
a
x
i
m
u
m
1
3
.
5
2
1
3
.
1
2
1
3
.
5
4
6
.
7
4
6
.
7
8
6
.
8
9
1
0
.
5
0
1
0
.
5
0
1
0
.
8
8
M
i
n
i
m
u
m
1
.
2
4
2
.
6
3
6
.
4
6
1
0
.
5
0
1
0
.
5
0
1
0
.
8
8
3
.
0
0
3
.
2
8
3
.
7
9
S
t
d
.
D
e
v
2
.
3
1
2
.
4
0
1
.
6
4
3
.
0
0
3
.
2
8
3
.
7
9
2
.
2
4
2
.
1
9
2
.
1
0
F
i
n
a
n
c
e
C
o
m
p
a
n
i
e
s
L
e
n
d
i
n
g
R
a
t
e
s
D
e
p
o
s
i
t
R
a
t
e
s
B
a
s
e
R
a
t
e
A
v
e
r
a
g
e
R
a
t
e
S
a
v
i
n
g
3
-
m
o
n
t
h
6
-
m
o
n
t
h
9
-
m
o
n
t
h
1
2
-
m
o
n
t
h
M
e
a
n
1
2
.
2
4
9
.
7
3
5
.
2
9
6
.
5
1
6
.
6
5
6
.
7
8
7
.
0
0
M
e
d
i
a
n
1
2
.
0
0
9
.
2
5
4
.
8
4
6
.
9
2
7
.
0
0
7
.
1
2
7
.
1
5
M
a
x
i
m
u
m
1
6
.
4
4
1
4
.
7
0
9
.
0
0
1
3
.
5
0
1
.
5
0
1
1
.
5
0
1
2
.
0
0
M
i
n
i
m
u
m
9
.
7
5
7
.
4
5
2
.
6
5
2
.
5
0
3
.
2
1
3
.
3
4
3
.
9
7
S
t
d
.
D
e
v
1
.
2
0
1
.
6
6
1
.
6
4
2
.
4
2
2
.
3
6
2
.
2
8
2
.
2
1
10
the degree of data integration. Furthermore,
as argued by Ng and Perron (2001) that
many unit root tests suffer from low power
and size distortions, we also apply the four
tests developed by Ng and Perron (2001)
(NgP tests) with improved size and power for
the unit root tests. Tables 3 and 4 provide
the results of both unit root tests. The ADF
and the PP tests suggest that the series are
I(1) and these results are conrmed by the
four NgP tests. As money market rates and
retail interest rates are I(1), we proceed with
testing for cointegration based on Johansen
(1988) between market and retail rates.
The results of the cointegration tests are
reported in Table 5. To test for robustness,
the Johansens eigenvalues and trace tests
are carried out. The results of the Johansen
tests show that all the retail rates are
cointegrated with the benchmark market rate
in favour of at least one cointegrating vector.
The results of the cointegration tests show
that there is a statistically signicant long-
run relationship between the benchmark
market rate and the administered interest
rates for both commercial banks and
nance companies. The estimated long-
run coefcients relationships are reported
in Table 6. As expected, the mark-up for
lending rates is much higher than that of
deposit rates. On average, the loan rate
mark-up is about 5% whereas for deposit
rates, it is about 1% on average. As for the
degree of pass-through, the results show
that the pass-through for both lending
and deposit rates is not complete. The
observation of incomplete pass-through
reects the fact that consumers face less
competitive pricing, consistent with the
model of Rosen (2002).
In the case of lending rates, it is about 50%
pass-through. The reason for the relatively
lower pass-through rate on both lending
and deposit rates is probably because the
market is generally not competitive. For
deposit rates, the pass-through rate (which
is in the 80% to 90% range) is higher than
that of lending rates. This is probably due
to the switching costs in the loan market as
compared with that for the deposit market.
As a result, it is easier for customers to
switch banks when deposit rates are not
competitive than when the lending rates are
not competitive. In addition, the relatively
lower degree of pass-through for lending
rates may be due to the adverse selection and
moral hazard reasons cited in the Stiglitz-
Weiss (1981) asymmetric information model.
The long-term adjustment results also
show that the pass-through on nance
companies deposits is higher than that
of commercial banks. This suggests that
nance companies tend to rely more on
deposits as a primary source of funding
given that their size is relatively smaller
and thus riskier than that of commercial
banks. Finance companies have to compete
more aggressively for deposits not only by
offering higher deposit rates, but also by
readjusting their rates more frequently to
changes in the market rates. Furthermore,
the savings rates for both nance companies
and commercial banks show a considerable
stickiness compared to the xed deposit
rates. The pass-through rate on nance
companies and commercial banks savings
deposit rates are 48% and 39% respectively,
while the average pass-through rate of their
xed deposits are relatively higher at 79%
and 72% respectively. The results indicate
that adjustments to market conditions are
much slower and competitive pressure
seems to be lower for the savings rate than
the xed deposit rate.
As the retail rates are found to be cointegrated
(at least one cointegrating vector) with the
benchmark market rate, the proper short-
run dynamics is given by the single equation
error-correction model (ECM). The results of
symmetric ECM are reported in Table 6. All
the estimates of are found to be negative
and statistically signicant as expected.
The results show that retail rates are mean
reverting to long-run equilibrium. In other
words, the rates will adjust upwards when
they are below the equilibrium level and
adjust downwards when the series are above
their equilibrium levels.
The results in Table 6 also show that the
short-run adjustment speed differs across
nancial products and nancial institutions.
On average, the short-run adjustment speed
for deposit rates is faster than that for lending
rates. In the case of commercial banks,
their adjustment mean lag (AML) for xed
deposit rates is about 4.7 months, but the
AML for lending rates is about 10.9 months.
On average, the AML for nance companies
xed deposit rates is about 3.7 months,
while the AML for lending rates is about
14.1 months. This nding is consistent with
the ndings made by Hofmann and Mizen
(2004) for the UK and Sorensen and Werner
(2006) for the euro area. The likely reasons
for this result, as pointed out earlier, may be
due to the relatively higher switching costs
and the potential adverse selection and
moral hazard problems in the loan market.
11
T
a
b
l
e
3
:
U
n
i
t
R
o
o
t
T
e
s
t
s
(
i
n
L
e
v
e
l
)
A
u
g
m
e
n
t
e
d
D
i
c
k
e
y
-
F
u
l
l
e
r
P
h
i
l
l
i
p
s
-
P
e
r
r
o
n
(
P
P
)
N
g
-
P
e
r
r
o
n
(
N
g
P
)
t
t
t
Z
t
Z
t
t
M
Z
a
M
Z
t
M
S
B
M
P
T
C
o
m
m
e
r
c
i
a
l
B
a
n
k
s
3
M
I
B
-
2
.
1
9
7
[
1
]
-
2
.
7
2
7
[
1
]
-
2
.
2
0
1
-
2
.
7
4
9
-
0
.
7
2
6
-
0
.
3
6
8
0
.
2
8
9
*
1
5
.
5
5
5
*
B
L
R
-
1
.
2
6
2
[
0
]
-
2
.
2
4
0
[
0
]
-
1
.
6
1
9
-
2
.
6
9
3
0
.
3
8
8
0
.
3
1
7
0
.
8
1
8
*
4
3
.
4
4
1
*
A
L
R
-
1
.
7
4
8
[
1
]
-
2
.
0
8
5
[
1
]
-
1
.
7
9
5
-
2
.
1
5
1
-
0
.
0
0
1
-
0
.
0
0
1
0
.
5
6
0
*
2
2
.
3
4
4
*
S
A
V
-
1
.
2
5
1
[
1
]
-
1
.
9
4
7
[
1
]
-
1
.
3
1
1
-
2
.
1
7
5
0
.
8
3
8
0
.
7
1
9
0
.
8
5
7
*
5
1
.
6
5
5
*
F
D
3
M
-
1
.
8
9
1
[
1
]
-
2
.
1
5
7
[
1
]
-
1
.
9
8
6
-
2
.
2
7
8
-
1
.
5
6
4
-
0
.
6
6
9
0
.
4
2
2
*
1
1
.
9
5
8
*
F
D
6
M
-
1
.
7
4
8
[
1
]
-
2
.
0
6
2
[
1
]
-
1
.
9
0
3
-
2
.
2
6
9
-
0
.
9
1
9
-
0
.
4
4
3
0
.
4
8
2
*
1
5
.
3
8
5
*
F
D
9
M
-
1
.
7
5
7
[
1
]
-
2
.
1
1
0
[
1
]
-
1
.
8
9
8
-
2
.
3
0
6
-
0
.
8
8
5
-
0
.
4
2
3
0
.
4
7
8
*
1
5
.
3
6
8
*
F
D
1
2
M
-
2
.
0
3
5
[
1
]
-
2
.
3
0
5
[
1
]
-
2
.
1
6
8
-
2
.
5
0
0
-
0
.
9
1
7
-
0
.
4
5
0
0
.
4
9
1
*
1
5
.
7
2
4
*
F
i
n
a
n
c
e
C
o
m
p
a
n
i
e
s
B
L
R
-
2
.
2
4
9
[
1
]
-
2
.
4
4
0
[
1
]
-
2
.
2
9
3
-
2
.
5
4
2
-
0
.
9
1
3
-
0
.
4
4
4
0
.
4
8
7
*
1
5
.
5
8
3
*
A
L
R
-
1
.
1
2
5
[
0
]
-
2
.
2
6
2
[
0
]
-
1
.
1
2
5
-
2
.
1
7
7
1
.
0
5
3
1
.
0
2
3
0
.
9
7
2
*
6
7
.
1
3
4
*
S
A
V
-
1
.
0
2
2
[
3
]
-
2
.
0
4
1
[
3
]
-
0
.
9
5
3
-
2
.
0
1
4
1
.
0
5
7
0
.
9
1
4
0
.
8
6
4
*
5
4
.
6
4
8
*
F
D
3
M
-
1
.
9
1
0
[
1
]
-
2
.
3
1
4
[
1
]
-
1
.
9
5
7
-
2
.
3
7
7
-
2
.
4
5
1
-
0
.
9
0
3
0
.
3
6
8
*
8
.
9
8
4
*
F
D
6
M
-
1
.
8
1
2
[
1
]
-
2
.
2
6
7
[
1
]
-
1
.
8
2
4
-
2
.
2
8
5
-
1
.
8
2
9
-
0
.
7
2
9
0
.
4
0
0
*
1
0
.
7
5
1
*
F
D
9
M
-
1
.
7
7
8
[
1
]
-
2
.
2
8
7
[
1
]
-
1
.
8
3
8
-
2
.
3
6
9
-
1
.
4
1
5
-
0
.
6
0
0
0
.
4
2
4
*
1
2
.
2
4
3
*
F
D
1
2
M
-
1
.
9
6
8
[
1
]
-
2
.
4
1
1
[
1
]
-
1
.
9
8
2
-
2
.
4
3
9
-
1
.
2
9
5
-
0
.
5
7
7
0
.
4
4
5
*
1
3
.
1
1
8
*
N
o
t
e
:
*
a
n
d
*
*
d
e
n
o
t
e
r
e
j
e
c
t
i
o
n
a
t
t
h
e
1
%
a
n
d
5
%
l
e
v
e
l
s
r
e
s
p
e
c
t
i
v
e
l
y
.
N
u
m
b
e
r
s
i
n
[
]
i
n
d
i
c
a
t
e
t
h
e
l
a
g
l
e
n
g
t
h
s
.
1
%
c
r
i
t
i
c
a
l
v
a
l
u
e
s
f
o
r
t
(
a
l
l
o
w
s
f
o
r
a
d
r
i
f
t
t
e
r
m
)
a
n
d
t
t
(
a
l
l
o
w
s
f
o
r
a
d
r
i
f
t
a
n
d
d
e
t
e
r
m
i
n
i
s
t
i
c
t
r
e
n
d
)
a
r
e
-
3
.
4
9
a
n
d
-
4
.
0
4
r
e
s
p
e
c
t
i
v
e
l
y
w
h
i
c
h
a
r
e
o
b
t
a
i
n
e
d
f
r
o
m
M
a
c
K
i
n
n
o
n
(
1
9
9
1
)
f
o
r
t
h
e
A
D
F
a
n
d
P
P
w
h
i
l
e
t
h
e
c
r
i
t
i
c
a
l
v
a
l
u
e
f
o
r
t
h
e
N
g
P
t
e
s
t
i
s
o
b
t
a
i
n
e
d
f
r
o
m
N
g
a
n
d
P
e
r
r
o
n
(
2
0
0
1
)
.
T
h
e
N
g
P
t
e
s
t
s
w
e
r
e
b
a
s
e
d
o
n
e
q
u
a
t
i
o
n
s
i
n
c
l
u
d
i
n
g
a
c
o
n
s
t
a
n
t
,
w
h
i
l
e
w
e
d
o
n
o
t
r
e
p
o
r
t
t
h
e
N
g
P
t
e
s
t
b
a
s
e
d
o
n
b
o
t
h
c
o
n
s
t
a
n
t
a
n
d
t
r
e
n
d
.
T
h
e
d
e
t
a
i
l
s
o
f
t
h
e
r
e
s
u
l
t
s
c
a
n
b
e
o
b
t
a
i
n
e
d
f
r
o
m
t
h
e
a
u
t
h
o
r
s
.
F
u
r
t
h
e
r
m
o
r
e
,
w
e
u
s
e
t
h
e
M
o
d
i
e
d
A
I
C
t
o
s
e
l
e
c
t
t
h
e
a
d
e
q
u
a
t
e
l
a
g
l
e
n
g
t
h
.
12
T
a
b
l
e
4
:
U
n
i
t
R
o
o
t
T
e
s
t
(
i
n
F
i
r
s
t
D
i
f
f
e
r
e
n
c
e
)
A
u
g
m
e
n
t
e
d
D
i
c
k
e
y
-
F
u
l
l
e
r
P
h
i
l
l
i
p
s
-
P
e
r
r
o
n
(
P
P
)
N
g
-
P
e
r
r
o
n
(
N
g
P
)
t
t
t
Z
t
Z
t
t
M
Z
a
M
Z
t
M
S
B
M
P
T
C
o
m
m
e
r
c
i
a
l
B
a
n
k
s
3
M
I
B
-
1
3
.
1
7
2
[
0
]
*
-
1
3
.
1
4
7
[
0
]
*
-
1
3
.
2
8
5
*
-
1
3
.
2
6
1
*
-
1
2
2
.
4
9
4
*
-
7
.
8
2
5
*
0
.
0
6
3
0
.
2
0
0
B
L
R
-
9
.
7
8
9
[
1
]
*
-
9
.
7
8
8
[
1
]
*
-
1
6
.
3
3
7
*
-
1
6
.
3
4
0
*
-
9
0
.
0
2
9
*
-
8
.
0
2
9
*
0
.
9
8
0
0
.
7
1
0
A
L
R
-
8
.
9
6
7
[
0
]
*
-
8
.
9
5
0
[
0
]
*
-
9
.
3
2
3
*
-
9
.
3
0
9
*
-
8
7
.
9
1
2
*
-
6
.
6
2
9
*
0
.
0
7
5
0
.
2
7
8
S
A
V
-
1
9
.
8
0
5
[
0
]
*
-
1
9
.
7
7
1
[
0
]
*
-
1
9
.
7
8
1
*
-
1
9
.
7
4
8
*
-
3
7
.
8
5
1
*
-
4
.
3
5
0
*
0
.
1
1
4
2
.
4
0
7
F
D
3
M
-
1
2
.
5
5
0
[
0
]
*
-
1
2
.
5
2
8
[
0
]
*
-
1
2
.
7
2
7
*
-
1
2
.
7
0
7
*
-
5
7
.
2
7
3
*
-
5
.
3
5
1
*
0
.
0
9
3
0
.
4
2
7
F
D
6
M
-
1
3
.
1
2
5
[
0
]
*
-
1
3
.
1
0
1
[
0
]
*
-
1
3
.
4
6
0
*
-
1
3
.
4
4
0
*
-
3
9
.
0
4
9
*
-
4
.
4
1
8
*
0
.
1
1
3
0
.
6
2
7
F
D
9
M
-
1
2
.
9
2
9
[
0
]
*
-
1
2
.
9
0
7
[
0
]
*
-
1
3
.
2
7
0
*
-
1
3
.
2
5
1
*
-
4
2
.
2
5
4
*
-
4
.
5
8
8
*
0
.
1
0
8
0
.
6
0
1
F
D
1
2
M
-
1
2
.
9
9
2
[
0
]
*
-
1
2
.
9
7
5
[
0
]
*
-
1
3
.
2
5
3
*
-
1
3
.
2
3
7
*
-
4
4
.
6
4
2
*
-
4
.
7
1
6
*
0
.
1
0
5
0
.
5
6
9
F
i
n
a
n
c
e
C
o
m
p
a
n
i
e
s
B
L
R
-
7
.
7
9
8
[
1
]
*
-
7
.
7
9
5
[
1
]
*
-
1
2
.
2
5
1
*
-
1
2
.
2
3
9
*
-
7
8
.
0
8
7
*
-
6
.
2
4
8
*
0
.
0
8
0
0
.
3
1
3
A
L
R
-
1
6
.
8
9
0
[
0
]
*
-
1
6
.
8
5
2
[
0
]
*
-
1
6
.
8
9
0
*
-
1
6
.
8
5
2
*
-
7
2
.
5
6
3
*
-
6
.
0
1
9
*
0
.
0
8
2
0
.
3
4
6
S
A
V
-
8
.
7
7
4
[
2
]
*
-
8
.
7
5
4
[
2
]
*
-
1
9
.
2
4
3
*
-
1
9
.
2
0
8
*
-
6
7
.
0
7
3
*
-
5
.
7
9
1
*
0
.
0
8
6
0
.
3
6
5
F
D
3
M
-
1
3
.
6
4
5
[
0
]
*
-
1
3
.
6
2
0
[
0
]
*
-
1
3
.
6
9
1
*
-
1
3
.
6
6
6
*
-
1
3
0
.
9
5
4
*
-
8
.
0
9
1
*
0
.
0
6
1
0
.
1
8
7
F
D
6
M
-
1
1
.
7
6
1
[
0
]
*
-
1
1
.
7
3
9
[
0
]
*
-
1
1
.
8
4
2
*
-
1
1
.
8
2
0
*
-
1
2
1
.
4
0
0
*
-
7
.
7
9
1
*
0
.
0
6
4
0
.
2
0
1
F
D
9
M
-
1
1
.
9
3
2
[
0
]
*
-
1
1
.
9
0
9
[
0
]
*
-
1
2
.
0
6
4
*
-
1
2
.
0
4
2
*
-
1
2
2
.
4
8
4
*
-
7
.
8
2
5
*
0
.
0
6
3
0
.
2
0
0
F
D
1
2
M
-
1
2
.
0
8
9
[
0
]
*
-
1
2
.
0
6
8
[
0
]
*
-
1
2
.
1
7
5
*
-
1
2
.
1
5
5
*
-
1
2
3
.
4
6
4
*
-
7
.
8
5
6
*
0
.
0
6
3
0
.
1
9
8
N
o
t
e
:
*
a
n
d
*
*
d
e
n
o
t
e
r
e
j
e
c
t
i
o
n
a
t
t
h
e
1
%
a
n
d
5
%
l
e
v
e
l
s
r
e
s
p
e
c
t
i
v
e
l
y
.
N
u
m
b
e
r
s
i
n
[
]
i
n
d
i
c
a
t
e
t
h
e
l
a
g
l
e
n
g
t
h
s
.
1
%
c
r
i
t
i
c
a
l
v
a
l
u
e
s
f
o
r
t
(
a
l
l
o
w
s
f
o
r
a
d
r
i
f
t
t
e
r
m
)
a
n
d
t
t
(
a
l
l
o
w
s
f
o
r
a
d
r
i
f
t
a
n
d
d
e
t
e
r
m
i
n
i
s
t
i
c
t
r
e
n
d
)
a
r
e
-
3
.
4
9
a
n
d
-
4
.
0
4
r
e
s
p
e
c
t
i
v
e
l
y
w
h
i
c
h
a
r
e
o
b
t
a
i
n
e
d
f
r
o
m
M
a
c
K
i
n
n
o
n
(
1
9
9
1
)
f
o
r
t
h
e
A
D
F
a
n
d
P
P
w
h
i
l
e
t
h
e
c
r
i
t
i
c
a
l
v
a
l
u
e
f
o
r
t
h
e
N
g
P
t
e
s
t
i
s
o
b
t
a
i
n
e
d
f
r
o
m
N
g
a
n
d
P
e
r
r
o
n
(
2
0
0
1
)
.
T
h
e
N
g
P
t
e
s
t
s
w
e
r
e
b
a
s
e
d
o
n
e
q
u
a
t
i
o
n
s
i
n
c
l
u
d
i
n
g
a
c
o
n
s
t
a
n
t
,
w
h
i
l
e
w
e
d
o
n
o
t
r
e
p
o
r
t
t
h
e
N
g
P
t
e
s
t
b
a
s
e
d
o
n
b
o
t
h
a
c
o
n
s
t
a
n
t
a
n
d
t
r
e
n
d
.
T
h
e
d
e
t
a
i
l
s
o
f
t
h
e
r
e
s
u
l
t
s
c
a
n
b
e
o
b
t
a
i
n
e
d
f
r
o
m
t
h
e
a
u
t
h
o
r
s
.
F
u
r
t
h
e
r
m
o
r
e
,
w
e
u
s
e
t
h
e
M
o
d
i
e
d
A
I
C
t
o
s
e
l
e
c
t
t
h
e
a
d
e
q
u
a
t
e
l
a
g
l
e
n
g
t
h
.
13
Table 5: Cointegration Test and Long-Run Relationship
Null Hypothesis Eigenvalue Trace Statistic Mark-up
Degree of Pass-
through
Commercial Banks
BLR
) 0 ( = r
) 1 ( s r
0.05
0.02
20.41*
6.15**
0.05 0.53
ALR
) 0 ( = r
) 1 ( s r
0.09
0.02
21.78*
3.74**
0.06 0.51
SAV
) 0 ( = r
) 1 ( s r
0.06
0.01
16.18*
1.49**
0.01 0.39
FD3M
) 0 ( = r
) 1 ( s r
0.06
0.02
19.69*
4.37**
0.005 0.88
FD6M
) 0 ( = r
) 1 ( s r
0.12
0.02
36.66*
4.06**
0.009 0.85
FD9M
) 0 ( = r
) 1 ( s r
0.09
0.02
28.99*
4.63**
0.01 0.83
FD12M
) 0 ( = r
) 1 ( s r
0.07
0.02
22.36*
4.42**
0.02 0.78
Finance Companies
BLR
) 0 ( = r
) 1 ( s r
0.06
0.02
19.82*
5.12**
0.06 0.53
ALR
) 0 ( = r
) 1 ( s r
0.08
0.01
15.56*
0.90
0.09 0.29
SAV
) 0 ( = r
) 1 ( s r
0.08
0.01
21.81*
0.74
0.02 0.48
FD3M
) 0 ( = r
) 1 ( s r
0.08
0.01
24.51*
5.87**
0.02 0.88
FD6M
) 0 ( = r
) 1 ( s r
0.07
0.02
30.79*
4.37**
0.008 0.91
FD9M
) 0 ( = r
) 1 ( s r
0.09
0.02
26.77*
3.93**
0.01 0.88
FD12M
) 0 ( = r
) 1 ( s r
0.07
0.02
17.42*
4.82**
0.01 0.83
Notes: r stands for the rank number. Rejection of r = 0 is evidence in favour of at least one cointegrating vector. * and
** indicate signicance at the 5% and 10% critical values. Critical values are based on MacKinnon-Haug-Michelis
(1999). BLR = base lending rate, ALR = average lending rate, SAV = savings rate, FD3M = 3-month xed deposit rate
FD6M = 6-month xed deposit rate, FD9M = 9-month xed deposit rate and FD12M = 12-month xed deposit rate.
Mark-up is measured by the intercept, o
0
and the degree of pass-through is measured by the slope, o
1
based on the
equation i
R
= o
0
+ o
1
i
P
+ c.
Among the deposit rates, the AML for both
banks and nance companies savings
deposit rates are much longer than that for
xed deposit rates. It takes about 16.3 and
15.8 months respectively, for the banks
and the nance companies savings deposit
rates to adjust to the long-term equilibrium
level while their xed deposit rates take
only about 4.7 and 3.7 months respectively.
The likely reason for this may be that cost
switching considerations more likely drove
the adjustment process in the short run.
For example, savings deposits are generally
held for savings/investment purposes and
are thus less sensitive to changes in interest
rates. The longer maturity xed deposits,
in contrast, are more sensitive to changes
in interest rates as they are mostly held for
investment purposes.
The short-run adjustment results also
show that nance companies are quicker in
adjusting their deposit rates but slower in
adjusting their lending rates in comparison
with commercial banks. In the case of xed
deposit rates, nance companies AML is
slightly shorter than that of banks. The AML
14
is around 3.7 and 4.7 months for nance
companies and banks respectively. The
results suggest that nance companies in
comparison to commercial banks rely more
heavily on deposits as a primary source
of funding. However, being smaller and
riskier, nance companies have to compete
aggressively for deposits. Finance companies
not only offer higher interest rates but are
also quicker in adjusting the deposit rates
back to the equilibrium level in response to
changes in market interest rates. In general,
our nding is consistent with previous
ndings on bank lending channels where
monetary policy is found to be transmitted
mostly through small banks that are either
illiquid or undercapitalised (Kakes & Sturm,
2002).
The results on the dynamics of lending rates,
however, show that, on average, nance
companies are slower than commercial
banks in adjusting their lending rates.
The AML is about 14.1 months for nance
companies lending rates and about 10.9
months for commercial banks lending rates.
The difference may be attributed to the fact
that nance companies loan customers face
greater switching costs in comparison to the
banks loan customers. Another explanation
is that customers who borrow from nance
companies are likely to be riskier and
cannot switch easily to borrowing from
banks. The ndings show that generally,
in the deposit market, monetary policy is
transmitted mostly through the smaller
nancial institutions. On the contrary,
in the loan market, the larger banking
institutions are quicker in transmitting
the effect of a change in monetary policy
via changes in the lending rates while the
nance companies are relatively slower in
adjusting their lending rates.
Finally, the results in Table 7 show that the
AML for all deposit rates and lending rates
are shorter when the rates are above the long-
term equilibrium level than when below. For
commercial banks, the lending rates AML is
about 7.8 months when they are above their
equilibrium level and about 11.9 months
when below. The mean adjustment lag for
nance companies is about 7.3 months
when rates are above their equilibrium level
and about 15.4 months below. This nding
is in contrast with Scholnik (1996) who also
found a downward rigidity for lending rates
in Malaysia. The upward rigidity in lending
rates, in particular, is consistent with the
credit rationing hypothesis cited in Stiglitz
and Weiss (1981) where nancial institutions
are more reluctant to adjust their loan rates
upwards because of the adverse selection
and moral hazard problems associated
with higher loan rates. Rather than
charging borrowers higher rates, banks
may be reluctant to lend, resulting in credit
rationing and slower loan growth.
There is similarly an upward rigidity in the
deposit rates. On average, the deposit rates
AML for commercial banks and nance
companies is about 2.9 and 2.8 months
respectively when their rates are above their
equilibrium level and about 9.9 months
and 7.5 months respectively when below.
This nding is consistent with both the
switching costs and imperfect competition
hypotheses. Customer deposits, for example,
are subject to signicant switching costs
if held primarily for transaction purposes.
Financial institutions are able to exploit
customer deposit inertia by being quicker in
reducing deposit rates when they are above
the equilibrium level than in raising them
when below. The existence of imperfect
competition in the deposit market, if any,
can give rise to collusive pricing practices
among the nancial institutions. Financial
institutions are, therefore, likely to reduce
deposit rates much faster than they increase
the rates when there is imperfect competition
in the deposit market.
VI. CONCLUSION
In this paper we provide a more uniform
analysis of the interest rate pass-through
process in the Malaysian nancial sector.
It is a well-known feature of monetary
policy operation that authorities aim to
exercise control over short-term interest
rates by adjusting the ofcial rate and that
it is commonly assumed there is complete
transmission to the short-term rate within
a short period of time. With complete pass-
through, monetary policy can be more
efcient in its ability to control ination
although incomplete pass-through can still
be effective if it is predictable.
The evidence provided in this paper shows
that for the vast majority of retail lending
rates pass-through is less than complete,
while the speed of adjustment varies across
retail interest rates. These empirical results
are consistent with previous studies,
although the degree of pass-through in
this study is found to be relatively lower.
Adjustment in lending rates than that of
deposit rates tended to be more sluggish.
Although nance companies are quicker
in adjusting their deposit rates, they are
slower in adjusting their loan rates than
commercial banks.
15
Note: *, ** and *** indicate signicance at the 1%, 5% and 10% levels. T-statistics are in parentheses. The model is
given by the following equation: .
Larger credit institutions (commercial
banks) adjust their lending rates to changes
in the market rates faster than smaller
credit institutions (nance companies). This
shows that smaller credit institutions are
more dependent on bank credit, compete
less intensely with market terms and adjust
their lending rates comparatively slowly;
while larger credit institutions, having better
access to the capital market, may offset
deposit changes in the event of a monetary
policy change by exible borrowing on
market terms, which have a comparatively
rapid effect on their lending rates.
The ndings of this paper also suggest
the existence of asymmetric adjustment of
interest rates in the short run. The interest
rates adjustment process is faster in a
period of monetary easing rather than in
one of monetary tightening. This empirical
nding has strong implications for the
conduct of monetary policy as it implies
that the speed of monetary transmission
is not uniform across both retail rates and
nancial institutions.
Table 6: Symmetric Model of Short-Run Dynamic
Short-term Error-correction Adjustment Mean Log
Pass-though ( ) Adjustment ( ) (AML)
Commercial Banks
BLR 0.09* -0.09*
(2.36) (-5.31) 10.1
ALR 0.07** -0.08*
(2.02) (5.75) 11.6
SAV 0.02*** -0.06*
(1.91) (-4.29) 16.3
FD3M 0.12* -0.22*
(3.14 (-9.17) 4.0
FD6M 0.11* -0.19*
(3.14) (-8.61) 4.7
FD9M 0.12* -0.18*
(3.38) (-8.32) 4.9
FD12M 0.12* -0.17*
(3.63) (-7.50) 5.2
Finance Companies
BLR 0.06*** -0.11*
(1.67) (-6.09) 11.8
ALR 0.01** -0.06*
(1.98) (-3.36) 16.5
SAV 0.05*** -0.06*
(1.86) (-3.55) 15.8
FD3M 0.29* -0.26*
(6.69) (-8.92) 2.7
FD6M 0.17* -0.22*
(4.70) (-9.41) 3.8
FD9M 0.15* -0.21*
(4.17) (-9.27) 4.0
FD12M 0.16* -0.19*
(4.34) (-8.16) 4.4
16
Commercial Banks
BLR 0.081** -0.116* -0.076* 0.13 7.8 11.9 9.529*
(2.192) (-4.834) (-2.584)
ALR 0.062*** -0.108* -0.068*)
(1.743) (-6.483) (-3.352) 0.20 8.6 13.6 8.830*
SAV 0.024** -0.089* -0.048*
(2.082) (-5.389) (-3.634) 0.13 10.8 20.8 8.634*
FD3M 0.079** -0.368* -0.114*
(2.143) (-10.579) (-3.623) 0.35 2.4 8.0 28.502*
FD6M 0.114* -0.291* -0.089*
(3.046) (-8.839) (-2.655) 0.27 3.0 9.9 18.166*
FD9M 0.086** -0.287* -0.287*
(2.415) (-9.193) (-2.709) 0.29 3.1 10.0 17.808*
FD12M 0.092* -0.278* -0.076**
(2.580) (-9.048) (-2.289) 0.28 3.3 11.9 21.998*
Finance Companies
BLR 0.052*** -0.128* -0.061**
(1.717) (-6.666) (-2.307) 0.18 8.0 15.4 26.150*
ALR 0.008*** -0.093* -0.054**
(1.811) (-3.430) (-1.971 0.11 10.5 18.2 5.583***
SAV 0.060** -0.073* -0.041***
(2.219) (-4.445) (-1.728) 0.13 12.7 22.7 5.718***
FD3M 0.269* -0.343* -0.161*
(5.850) (-8.482) (-3.643) 0.36 2.1 4.5 9.102*
FD6M 0.152* -0.308* -0.090*
(3.969) (-8.918) (-2.641) 0.31 2.7 9.3 19.757*
FD9M 0.131* -0.289* -0.107*
(3.435) (-8.061) (-2.980) 0.27 2.9 8.0 10.639*
FD12M 0.132* -0.269* -0.102*
(3.540) -0.269* (-2.797) 0.27 3.2 8.4 11.420*
Note: *, ** and *** indicate signicance at the 1%, 5% and 10% levels.
T-statistics are in parentheses.
The model is given by the following equation: .
Table 7: Asymmetric Model of Short-Run Dynamics
1
o
2
o
3
o
2
R
+
AML
t
is the trend component of the series,
t
is the corresponding cyclical component,
t
is the seasonal component and
t
is the
irregular component (assumed to be white
noise). The above equation is subject to the
following two restrictions: cov (
t
,
t
,
t
) = 0
and
t
~ NID (0,
2
c
o ). These restrictions imply
that the components cannot in any way be
correlated and that the error term must be
normally and independently distributed.
In its most general form, the trend
component (representing the long-term
movement in a series) can be written as a
stochastic linear process.
t
=
t-1
+
t-1
+
t
[2]
t
=
t-1
+ t [3]
subject to
t
~ NID (0,
2
c
o
) and t ~ NI D
( 0,
2
c
o ) , which implies that
t
follows a
random walk with a drift
t
(following a rst-
order autoregressive process as represented
by Equation 3).
The stochastic properties of the level and
slope are driven by
t
and
t
, which are
uncorrelated random errors. Each of the
unobserved components is assumed to
8 Harvey and Todd (1983) identied a number of other advantages for using the structural time series modelling
approach.
9 The same conclusion was reached by Al-Sad and Moosa (2005).
10 In this paper, unobserved components entail trend, cyclical and seasonal component.
33
be normally distributed with mean 0 and
variances
2
c
o ,
2
q
o
, and
2
o
. If any of these
variances converges to zero, the stochastic
component becomes purely deterministic
and the stochastic model collapses into a
purely deterministic model.
The seasonal component can be represented
by a trigonometric specication, as
dened by Harvey and Scott (1994). This
specication is commonly known to allow a
smoother change in seasonals. Nevertheless,
in this paper this specication will not be
used as the estimates for the seasonal
component do not allow for a straightforward
interpretation
11
. Consequently, stochastic
dummies will be used to represent the
seasonal component as:
t
=_
1
1
s
j
j t
+ K
t
[4]
where s is the number of possible seasonal
factors in one year and K
t
(random term) K
t
~ NID (0,
2
K
o ). For stochastic seasonality
to be satised
0
2
=
K
o
, whereas deterministic
seasonality requires the condition 0
2
=
K
o to
be satised.
Finally, the cyclical component is assumed
to be a stationary linear process
, sin cos
t t t
b a u u | + =
[5]
where u
t
is the frequency of the cycle and
a and b are parameters. In order to make
the cycle stochastic, the parameters a
and b are allowed to change over time,
while preserving continuity is achieved by
writing down a recursion for constructing
before introducing the stochastic elements.
By introducing a dampening factor and
disturbances the following are obtained:
,
*
1 1
) sin cos (
t t t t
e u | u | | + + =
[6]
,
* *
1 1
*
) cos sin (
t
t t
t
e u | u | | + + =
[7]
where
*
t
| appears by construction such
that
t
e and
*
t
e are uncorrelated white
noise disturbances with variances and
-
, respectively. The parameters
t u s s 0
and
1 0 s s
are the frequency of the cycle
and the damping factor on the amplitude,
respectively. The period of the cycle, which
is the time taken by the cycle to go through
its complete sequence of values, is u t / 2
(Harvey, 1990, p 38).
The above specied model is put into a state
space form and estimated by the maximum
likelihood procedure using the Kalman
ltering technique
12
. The actual estimations
are conducted by STAMP 6.0 software,
which is a Windows-based econometric
modelling programme.
III. EMPIRICAL RESULTS
The structural time series model represented
by Equation 1 is applied to actual monthly
home loan interest rate data for NAB,
between 1994:06 and 2004:03. The data
have been sourced from CANNEX
13 14
. For
consistency, the sample for each variable is
standardised to start with the rst available
July observation and end with the latest
available June observation.
Before going into the actual results of
the modelling, a short elaboration on the
meaning of the elements of the nal state
vector might be benecial.
t
represents the
level of the series while
1t
is a seasonal factor
that relates to the last month in the sample,
i.e. June. That means that
2t
corresponds to
May,
3t
corresponds to April, and so forth.
t
|
and
*
t
| represent cyclical components of
the series.
In addition to other results, Table 1
also reports on the goodness of t and
diagnostics. The goodness of t measures
entail the coefcient of determination
15
(R
2
),
a modied coefcient of determination
16
(R
s
2
), the standard error of the estimation
17
11 Koopman et al (1995, p 226) discusses the difculties of interpreting trigonometric seasonals.
12 Detailed explanations of the procedure are provided in Harvey (1985, 1990) and Koopman (1997).
13 Due to often sizeable discounts applicable to the advertised rates and condentiality surrounding those discounts,
more recent reliable data was not available at the time of the analysis.
14 CANNEX is Australia and New Zealands specialist research service company used by over 450 nancial institutions,
government, media and nance professionals. Since 1992, CANNEX has been highly regarded as the premier supplier
of data and independent and objective research and analysis for banks, building societies, credit unions, life and
general insurance and other nancial services providers, as well as the media.
15 A statistic that is widely used to determine how close the actual observations are to the line of the best t.
16 Unlike the conventional coefcient of determination, the modied coefcient of determination is calculated on the
basis of the seasonal mean.
17 The standard error is the standard deviation of the sampling distribution of the estimator, calculated as the square
root of the one-step ahead prediction error variance.
*
t
e
34
(o), Akaikes Information Criterion
18
(AIC)
and the Schwarz Bayesian Criterion
(BIC). On the other hand, diagnostic tests
cover serial correlation, normality and
heteroscedasticity. They are represented
by the Durbin-Watson statistic (DW), the
Bowman-Shenton (1975) test for normality
of the residual
19
(N), the Ljung-Box (1978)
test for serial correlation
20
(Q) and a test for
heteroscedasticity
21
(H).
The estimation set covers 10 variables,
namely, the interest rates for the following
home loans
22
: Basic Home Loan (BHL),
Standard Variable Rate Home Loan (SVRHL),
6-month xed rate Introductory home loan,
12-month xed rate Introductory home
loan, 1-year Fixed Rate Home Loans (FRHL),
2-year FRHL, 3-year FRHL, 4-year FRHL,
5-year FRHL and Line of Credit (LOC) home
loan.
With respect to the goodness of t, the
model is generally well dened. Overall, the
diagnostic tests are predominately passed.
The only exceptions are the test for serial
correlation (Q) (for BHL, SVRL, 6-month
Introductory FRHL, 1-Year FRHL and the
LOC home loan) and the test for normality
(N) (for the 12-month xed rate Introductory
home loan), which are slightly above the
statistically acceptable level.
As illustrated in Table 1, most of the
variables analysed, with the exception of the
3- and 4-year FRHLs, imply the presence of
a statistically signicant cyclical component
(shown as variables with t statistics values
above 1.96). Similarly, with respect to the
11 presented seasonal factors, most of the
variables analysed, with the exception of the
3- and 5-year FRHLs, imply the presence
of statistically signicant seasonal factors
(shown as variables with t statistics values
above 1.96). Figures 2 to 11 plot observed
interest rates and trend, cycle components
and seasonal components, individually for
each variable.
The seasonal factors presented appear to
be rather regular for all variables except the
12-month xed rate Introductory home loan,
implying the deterministic nature of all the
seasonal factors except those corresponding
to the 12-month xed rate Introductory
home loan, and the stochastic nature of
the seasonal factors corresponding to the
12-month xed rate Introductory home loan.
IV. FINDINGS
Home Loan Interest Rates Cyclicality
The result of the estimating Equation 5,
shown in Table 1, indicates that statistically
signicant cyclicality is observed for almost
all the variables analysed, except for the 3-
and 4-year FRHLs. Therefore, the hypothesis
that cyclicality found in Australian nancial
markets interest rates should also be found
in home loan interest rates, can, overall, be
conrmed. Due to the scope of the analysis,
a conclusive explanation for the absence
of statistically signicant cyclicality in the
3- and 4-year FRHLs cannot be provided.
A possible explanation, however, may be
that during the observed period, NAB,
through application of its pricing strategies,
particularly targeted 3- and 4-year FRHL
interest rates and thus managed to remove
cyclical behaviours from the interest rates
for those two products. More detailed
analysis would be needed to test such a
hypothesis, indicating an interesting topic
for future research.
Home Loan Interest Rates Seasonality
Compared to the interpretation of the
modelling results for the home loan interest
rates cyclicality, the interpretation of
modelling results for home loan interest
rates seasonality is a bit more complex.
Here, the focus is not only on nding
whether the seasonality exists or not, but on
examining the existence, or not, of the three
seasonal effects in home loan interest rates
(spring, autumn and the end of the nancial
year). Even after establishing the existence
of a statistically signicant seasonal factor
corresponding to one of the considered
seasonal effects, an additional condition
applicable to all three seasonal effects is for
the statistically signicant seasonal factor to
be expressed as a product-specic seasonal
rate reduction. For this reason, while at
risk of being somewhat repetitive, instead
of providing a summary interpretation,
brief product-level interpretations of the
modelling results are offered.
18 The procedure applied to calculate o is also used to calculate AIC, by adjusting for the number of estimated parameters.
19 The N test measures the departure of the third and fourth moments from their expected values under normality.
20 The Q statistic is distributed as x
2 (n+1-k)
where n is the number of autocorrelation coefcients and k is the number of
estimated parameters.
21 The H test is calculated as the ratio of the squares of the last h residual to the squares of the rst h residuals. The h
is dened as the closest integer to one-third of the sample size and distributed as F(h,h).
22 For more details on the presented home loan products reefer to Karamujic (2008).
35
T
a
b
l
e
1
:
F
i
n
a
l
S
t
a
t
e
V
e
c
t
o
r
s
6
m
t
h
.
x
e
d
r
a
t
e
1
2
m
t
h
.
x
e
d
r
a
t
e
B
H
L
S
V
R
H
L
I
n
t
r
o
d
.
h
o
m
e
I
n
t
r
o
d
.
h
o
m
e
1
-
y
e
a
r
F
R
H
L
2
-
y
e
a
r
F
R
H
L
3
-
y
e
a
r
F
R
H
L
4
-
y
e
a
r
F
R
H
L
5
y
e
a
r
F
R
H
L
L
O
C
H
o
m
e
L
o
a
n
l
o
a
n
l
o
a
n
E
s
t
i
m
a
t
e
d
t
E
s
t
i
m
a
t
e
d
t
E
s
t
i
m
a
t
e
d
t
E
s
t
i
m
a
t
e
d
t
E
s
t
i
m
a
t
e
d
t
E
s
t
i
m
a
t
e
d
t
E
s
t
i
m
a
t
e
d
t
E
s
t
i
m
a
t
e
d
t
E
s
t
i
m
a
t
e
d
t
E
s
t
i
m
a
t
e
d
t
V
a
l
u
e
s
S
t
a
t
i
s
t
i
c
s
V
a
l
u
e
s
S
t
a
t
i
s
t
i
c
s
V
a
l
u
e
s
S
t
a
t
i
s
t
i
c
s
V
a
l
u
e
s
S
t
a
t
i
s
t
i
c
s
V
a
l
u
e
s
S
t
a
t
i
s
t
i
c
s
V
a
l
u
e
s
S
t
a
t
i
s
t
i
c
s
V
a
l
u
e
s
S
t
a
t
i
s
t
i
c
s
V
a
l
u
e
s
S
t
a
t
i
s
t
i
c
s
V
a
l
u
e
s
S
t
a
t
i
s
t
i
c
s
V
a
l
u
e
s
S
t
a
t
i
s
t
i
c
s
t
6
.
1
2
3
9
0
5
8
.
2
4
6
.
5
7
4
0
0
5
6
.
2
8
5
.
1
8
6
5
0
3
5
.
0
5
5
.
4
8
6
0
0
4
0
.
8
9
6
.
1
1
1
2
0
4
7
.
2
8
6
.
5
7
3
9
0
3
9
.
2
2
6
.
0
8
1
2
0
8
3
.
4
2
6
.
5
4
8
3
0
4
6
.
2
9
6
.
9
9
3
7
0
4
1
.
1
4
6
.
7
5
7
3
0
6
3
.
7
6
|
t
0
.
0
1
3
5
8
0
.
1
6
-
0
.
0
1
7
4
9
-
0
.
1
7
-
0
.
0
1
1
4
0
-
0
.
2
9
0
.
0
4
2
2
1
0
.
4
3
0
.
0
7
3
8
9
1
.
4
8
0
.
0
3
0
6
8
0
.
4
5
-
0
.
0
3
2
1
0
-
0
.
9
9
-
0
.
0
2
7
9
2
-
0
.
3
5
0
.
0
2
5
5
9
0
.
4
7
0
.
0
2
1
1
8
0
.
2
5
|
t
0
.
0
1
3
6
2
0
.
4
4
0
.
0
0
9
7
2
0
.
3
3
0
.
0
0
2
4
4
0
.
0
9
-
0
.
1
3
2
1
1
-
3
.
3
5
-
0
.
1
8
2
2
3
-
2
.
0
7
-
0
.
2
5
3
0
5
-
2
.
7
6
0
.
0
2
7
1
7
1
.
4
3
-
0
.
0
4
4
7
5
-
0
.
4
8
0
.
0
6
9
7
1
2
.
1
1
0
.
0
4
7
7
9
1
.
5
5
|
t *
-
0
.
0
9
4
8
2
-
3
.
0
8
-
0
.
0
9
3
0
0
-
3
.
1
2
-
0
.
0
9
3
6
9
-
3
.
4
1
0
.
1
3
0
5
4
1
.
8
-
0
.
3
3
7
0
8
-
3
.
8
3
-
0
.
1
6
9
0
9
-
1
.
9
-
0
.
0
1
3
6
1
-
0
.
7
2
-
0
.
0
6
5
9
5
-
0
.
7
3
-
0
.
3
0
0
7
9
-
2
.
6
1
-
0
.
0
9
4
7
5
-
3
.
0
9
1
t
-
0
.
0
6
7
6
3
-
0
.
9
5
-
0
.
0
7
5
4
9
-
1
.
0
8
-
0
.
1
4
0
9
5
-
1
-
0
.
0
7
8
0
6
-
0
.
8
7
0
.
0
1
4
3
7
0
.
2
9
0
.
0
0
1
6
1
0
.
0
3
-
0
.
0
0
8
3
9
-
0
.
1
2
0
.
0
1
1
9
8
0
.
1
5
-
0
.
1
3
2
0
7
-
1
.
6
4
-
0
.
0
7
3
3
0
-
1
.
0
4
2
t
-
0
.
1
4
4
1
3
-
2
.
1
4
-
0
.
1
4
9
8
8
-
2
.
2
6
-
0
.
1
7
7
2
6
-
1
.
2
7
-
0
.
0
7
2
0
8
-
0
.
8
4
0
.
0
2
5
5
4
0
.
5
3
-
0
.
0
2
2
2
0
-
0
.
3
6
0
.
0
1
9
4
8
0
.
2
8
-
0
.
0
9
6
2
0
-
1
.
2
7
-
0
.
0
6
8
0
8
-
0
.
8
7
-
0
.
1
4
3
8
1
-
2
.
1
6
3
t
-
0
.
1
4
6
9
4
-
2
.
2
3
-
0
.
1
5
0
9
1
-
2
.
3
2
-
0
.
1
4
2
3
8
-
1
.
0
4
-
0
.
1
0
7
6
6
-
1
.
2
7
-
0
.
0
4
2
1
4
-
0
.
8
9
-
0
.
0
6
2
1
4
-
0
.
9
9
0
.
0
0
3
6
1
0
.
0
5
-
0
.
1
5
1
4
0
-
2
.
0
0
-
0
.
0
9
0
3
4
-
1
.
1
8
-
0
.
1
3
9
3
9
-
2
.
1
4
4
t
-
0
.
0
7
3
9
3
-
1
.
1
2
-
0
.
0
7
6
4
2
-
1
.
1
8
-
0
.
4
2
3
5
3
-
3
.
1
2
-
0
.
1
7
2
0
8
-
2
.
0
4
-
0
.
0
4
6
1
2
-
0
.
9
8
-
0
.
1
2
8
8
7
-
2
.
0
6
-
0
.
1
0
3
5
8
-
1
.
4
9
-
0
.
1
5
1
1
1
-
1
.
9
9
-
0
.
0
8
0
0
0
-
1
.
0
5
-
0
.
0
2
1
8
6
-
0
.
3
4
5
t
0
.
0
7
1
3
1
1
.
0
8
0
.
0
7
0
1
0
1
.
0
8
-
0
.
1
9
8
1
2
-
1
.
4
7
-
0
.
0
9
3
8
4
-
1
.
1
1
-
0
.
0
6
5
1
9
-
1
.
3
8
-
0
.
1
1
1
3
0
-
1
.
7
8
-
0
.
0
2
9
5
9
-
0
.
4
3
-
0
.
1
2
0
9
0
-
1
.
6
0
-
0
.
0
8
8
3
4
-
1
.
1
6
0
.
0
5
3
9
8
0
.
8
3
6
t
-
0
.
0
3
1
2
9
-
0
.
4
7
-
0
.
0
3
1
3
3
-
0
.
4
8
0
.
0
6
6
7
8
0
.
5
0
0
.
0
2
7
9
1
0
.
3
3
-
0
.
1
0
1
1
4
-
2
.
1
4
-
0
.
0
7
8
0
6
-
1
.
2
5
0
.
0
1
8
1
1
0
.
2
6
-
0
.
0
5
3
0
8
-
0
.
7
0
-
0
.
0
0
4
5
4
-
0
.
0
6
-
0
.
0
4
4
9
4
-
0
.
6
9
7
t
0
.
0
4
7
8
2
0
.
7
2
0
.
0
4
8
9
2
0
.
7
5
0
.
0
9
9
5
9
0
.
7
4
0
.
0
6
0
4
7
0
.
7
1
-
0
.
0
6
5
0
4
-
1
.
3
8
-
0
.
1
0
2
4
3
-
1
.
6
5
-
0
.
0
6
8
5
1
-
0
.
9
9
-
0
.
1
0
6
8
4
-
1
.
4
2
-
0
.
0
0
8
3
7
-
0
.
1
1
0
.
0
3
6
3
5
0
.
5
5
8
t
0
.
0
5
4
0
8
0
.
8
2
0
.
0
5
6
2
9
0
.
8
6
-
0
.
1
4
6
5
4
-
1
.
0
9
0
.
0
9
1
3
7
1
.
0
8
0
.
0
4
7
6
1
1
.
0
1
0
.
0
1
3
9
0
0
.
2
2
-
0
.
0
4
6
3
5
-
0
.
6
7
0
.
0
4
2
5
5
0
.
5
6
0
.
0
5
1
1
1
0
.
6
7
0
.
0
4
4
3
2
0
.
6
8
9
t
0
.
0
5
1
0
2
0
.
7
7
0
.
0
5
4
3
0
0
.
8
4
0
.
3
0
8
0
2
2
.
2
7
0
.
0
9
4
5
1
1
.
1
2
0
.
0
8
0
9
0
1
.
7
1
0
.
0
2
5
9
3
0
.
4
1
-
0
.
0
0
9
6
0
-
0
.
1
4
0
.
0
4
2
8
0
0
.
5
6
0
.
0
4
3
2
1
0
.
5
7
0
.
0
4
3
6
4
0
.
6
7
1
0
t
0
.
1
0
7
2
5
1
.
6
2
0
.
1
1
1
4
8
1
.
7
2
0
.
3
6
8
7
1
2
.
6
8
0
.
1
1
0
6
1
1
.
3
1
0
.
0
5
0
1
3
1
.
0
5
0
.
1
5
6
6
6
2
.
5
1
0
.
0
2
1
4
3
0
.
3
1
0
.
1
9
3
4
8
2
.
5
5
0
.
1
3
5
4
6
1
.
7
6
0
.
1
0
3
6
2
1
.
5
9
1
1
t
0
.
1
0
6
0
3
1
.
5
8
0
.
1
1
1
0
4
1
.
6
8
0
.
2
9
9
0
9
2
.
1
5
0
.
0
6
4
3
4
0
.
7
5
0
.
0
5
0
7
5
1
.
0
6
0
.
1
4
7
7
3
2
.
3
7
0
.
0
9
6
0
7
1
.
3
7
0
.
2
2
1
3
8
2
.
9
2
0
.
1
3
0
9
4
1
.
6
7
0
.
1
0
7
5
5
1
.
6
2
R
2
0
.
9
4
9
5
0
0
.
9
5
0
2
6
0
.
8
6
8
4
5
0
.
9
5
1
7
8
0
.
9
5
2
6
0
0
.
9
1
6
8
9
0
.
9
7
2
6
3
0
.
9
1
5
6
2
0
.
8
9
5
1
0
0
.
9
5
4
2
6
R
s
2
0
.
2
3
8
5
1
0
.
2
4
9
2
3
0
.
3
5
9
2
8
0
.
1
8
0
2
9
0
.
3
3
3
1
0
0
.
1
3
1
9
6
-
0
.
0
0
1
1
7
0
.
1
4
4
4
8
0
.
1
9
4
5
8
0
.
2
8
4
9
6
o
0
.
1
4
4
1
3
0
.
1
4
2
6
8
0
.
1
9
4
1
5
0
.
1
4
8
9
0
0
.
1
2
5
7
4
0
.
1
5
6
4
5
0
.
1
8
7
8
0
0
.
1
5
8
1
4
0
.
1
6
9
0
1
0
.
1
4
4
3
7
A
I
C
-
2
.
8
3
2
4
6
-
2
.
8
5
2
6
9
-
1
.
8
8
9
3
4
-
2
.
4
2
0
0
7
-
3
.
3
1
3
7
4
-
3
.
0
1
5
6
3
-
2
.
8
2
3
9
6
-
2
.
6
4
6
8
8
-
2
.
5
1
3
9
6
-
2
.
8
2
9
1
B
I
C
-
1
.
8
5
7
8
7
-
1
.
8
7
8
1
0
-
0
.
7
8
9
6
7
-
1
.
3
2
0
4
1
-
2
.
4
4
1
1
0
-
2
.
2
2
5
1
2
-
2
.
1
5
6
1
6
-
1
.
6
7
2
2
9
-
1
.
5
3
9
3
7
-
1
.
8
5
4
5
1
Q
*
2
0
.
7
7
0
0
0
2
0
.
6
2
0
0
0
1
4
.
0
0
0
0
0
1
0
.
2
1
0
0
0
1
8
.
7
0
0
0
0
8
.
3
3
3
0
0
8
.
6
8
3
0
6
.
5
4
2
0
1
1
.
4
6
0
0
1
4
.
9
4
0
0
H
#
0
.
1
9
9
8
4
0
.
2
0
6
5
3
0
.
5
4
0
7
1
0
.
2
1
0
9
3
0
.
5
4
2
7
7
0
.
3
9
9
9
5
1
.
0
1
1
0
0
0
.
4
5
0
3
4
0
.
4
2
1
2
0
0
.
2
2
9
5
4
N
4
.
7
8
8
0
0
4
.
5
6
6
0
0
0
.
0
2
7
0
5
4
6
.
8
0
0
0
0
2
.
4
5
3
0
0
0
.
4
4
5
0
4
0
.
8
3
4
2
9
1
.
1
9
0
0
0
1
.
8
5
5
0
0
5
.
4
1
8
0
0
*
D
i
s
t
r
i
b
u
t
e
d
a
s
_
2
(
6
)
*
D
i
s
t
r
i
b
u
t
e
d
a
s
_
2
(
6
)
*
D
i
s
t
r
i
b
u
t
e
d
a
s
_
2
(
6
)
*
D
i
s
t
r
i
b
u
t
e
d
a
s
_
2
(
6
)
*
D
i
s
t
r
i
b
u
t
e
d
a
s
_
2
(
6
)
*
D
i
s
t
r
i
b
u
t
e
d
a
s
_
2
(
6
)
*
D
i
s
t
r
i
b
u
t
e
d
a
s
_
2
(
6
)
*
D
i
s
t
r
i
b
u
t
e
d
a
s
_
2
(
6
)
*
D
i
s
t
r
i
b
u
t
e
d
a
s
_
2
(
6
)
*
D
i
s
t
r
i
b
u
t
e
d
a
s
_
2
(
6
)
#
D
i
s
t
r
i
b
u
t
e
d
a
s
F
(
l
l
,
l
l
)
#
D
i
s
t
r
i
b
u
t
e
d
a
s
F
(
1
1
,
1
1
)
#
D
i
s
t
r
i
b
u
t
e
d
a
s
F
(
7
,
7
)
#
D
i
s
t
r
i
b
u
t
e
d
a
s
F
(
7
,
7
)
*
#
D
i
s
t
r
i
b
u
t
e
d
a
s
F
(
1
5
,
1
5
)
#
D
i
s
t
r
i
b
u
t
e
d
a
s
F
(
1
9
,
1
9
)
#
D
i
s
t
r
i
b
u
t
e
d
a
s
F
(
2
7
,
2
7
)
#
D
i
s
t
r
i
b
u
t
e
d
a
s
F
(
1
1
,
1
1
)
#
D
i
s
t
r
i
b
u
t
e
d
a
s
F
(
7
,
7
)
#
D
i
s
t
r
i
b
u
t
e
d
a
s
F
(
7
,
7
)
C
o
m
p
o
n
e
n
t
36
Fig 2: BHL, Interest Rates and Trend, Cyclical and Seasonal Components
Observed Interest Rates and Trend
5
5.5
6
6.5
7
7.5
8
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Cyclical Component
Cys
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Seasonal Component
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
37
Fig 3: SVRHL, Interest Rates and Trend, Cyclical and Seasonal Components
Observed Interest Rates and Trend
5
5.5
6
6.5
7
7.5
8
8.5
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Cyclical Component
Cys
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Seasonal Component
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
38
Fig 4: 6-month Fixed Rate Introductory Home Loan, Interest Rates and Trend,
Cyclical and Seasonal Components
Observed Interest Rates and Trend
3.5
4
4.5
5
5.5
6
6.5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Cyclical Component
Cys
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
2
0
0
0
-
7
2
0
0
0
-
8
2
0
0
0
-
9
2
0
0
0
-
1
0
2
0
0
0
-
1
1
2
0
0
0
-
1
2
2
0
0
1
-
1
2
0
0
1
-
2
2
0
0
1
-
3
2
0
0
1
-
4
2
0
0
1
-
5
2
0
0
1
-
6
2
0
0
1
-
7
2
0
0
1
-
8
2
0
0
1
-
9
2
0
0
1
-
1
0
2
0
0
1
-
1
1
2
0
0
1
-
1
2
2
0
0
2
-
1
2
0
0
2
-
2
2
0
0
2
-
3
2
0
0
2
-
4
2
0
0
2
-
5
2
0
0
2
-
6
2
0
0
2
-
7
2
0
0
2
-
8
2
0
0
2
-
9
2
0
0
2
-
1
0
2
0
0
2
-
1
1
2
0
0
2
-
1
2
2
0
0
3
-
1
2
0
0
3
-
2
2
0
0
3
-
3
2
0
0
3
-
4
2
0
0
3
-
5
2
0
0
3
-
6
Seasonal Component
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
2
0
0
0
-
7
2
0
0
0
-
9
2
0
0
0
-
1
1
2
0
0
1
-
1
2
0
0
1
-
3
2
0
0
1
-
5
2
0
0
1
-
7
2
0
0
1
-
9
2
0
0
1
-
1
1
2
0
0
2
-
1
2
0
0
2
-
3
2
0
0
2
-
5
2
0
0
2
-
7
2
0
0
2
-
9
2
0
0
2
-
1
1
2
0
0
3
-
1
2
0
0
3
-
3
2
0
0
3
-
5
39
Fig 5: 12-month Fixed Rate Introductory Home Loan, Interest Rates and Trend,
Cyclical and Seasonal Components
Observed Interest Rates and Trend
4.5
5
5.5
6
6.5
7
7.5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Cyclical Component
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Seasonal Component
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
2
0
0
0
-
7
2
0
0
0
-
9
2
0
0
0
-
1
1
2
0
0
1
-
1
2
0
0
1
-
3
2
0
0
1
-
5
2
0
0
1
-
7
2
0
0
1
-
9
2
0
0
1
-
1
1
2
0
0
2
-
1
2
0
0
2
-
3
2
0
0
2
-
5
2
0
0
2
-
7
2
0
0
2
-
9
2
0
0
2
-
1
1
2
0
0
3
-
1
2
0
0
3
-
3
2
0
0
3
-
5
40
Fig 6: 1-year FRHL, Interest Rates and Trend, Cyclical and Seasonal Components
Observed Interest Rates and Trend
5.5
6
6.5
7
7.5
8
1
9
9
8
-7
1
9
9
8
-9
1
9
9
8
-1
1
1
9
9
9
-1
1
9
9
9
-3
1
9
9
9
-5
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Cyclical Component
NAB's 1 Year Fixed - Cycle 3
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
1
9
9
8
-7
1
9
9
8
-9
1
9
9
8
-1
1
1
9
9
9
-1
1
9
9
9
-3
1
9
9
9
-5
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Seasonal Component
NAB's 1 Year Fixed - Seasons
-0.15
-0.1
-0.05
0
0.05
0.1
1
9
9
8
-
7
1
9
9
8
-
9
1
9
9
8
-
1
1
1
9
9
9
-
1
1
9
9
9
-
3
1
9
9
9
-
5
1
9
9
9
-
7
1
9
9
9
-
9
1
9
9
9
-
1
1
2
0
0
0
-
1
2
0
0
0
-
3
2
0
0
0
-
5
2
0
0
0
-
7
2
0
0
0
-
9
2
0
0
0
-
1
1
2
0
0
1
-
1
2
0
0
1
-
3
2
0
0
1
-
5
2
0
0
1
-
7
2
0
0
1
-
9
2
0
0
1
-
1
1
2
0
0
2
-
1
2
0
0
2
-
3
2
0
0
2
-
5
2
0
0
2
-
7
2
0
0
2
-
9
2
0
0
2
-
1
1
2
0
0
3
-
1
2
0
0
3
-
3
2
0
0
3
-
5
41
Fig 7: 2-year FRHL, Interest Rates and Trend, Cyclical and Seasonal Components
Observed Interest Rates and Trend
5
5.5
6
6.5
7
7.5
8
8.5
1
9
9
7
-
7
1
9
9
7
-
9
1
9
9
7
-
1
1
1
9
9
8
-
1
1
9
9
8
-
3
1
9
9
8
-
5
1
9
9
8
-
7
1
9
9
8
-
9
1
9
9
8
-
1
1
1
9
9
9
-
1
1
9
9
9
-
3
1
9
9
9
-
5
1
9
9
9
-
7
1
9
9
9
-
9
1
9
9
9
-
1
1
2
0
0
0
-
1
2
0
0
0
-
3
2
0
0
0
-
5
2
0
0
0
-
7
2
0
0
0
-
9
2
0
0
0
-
1
1
2
0
0
1
-
1
2
0
0
1
-
3
2
0
0
1
-
5
2
0
0
1
-
7
2
0
0
1
-
9
2
0
0
1
-
1
1
2
0
0
2
-
1
2
0
0
2
-
3
2
0
0
2
-
5
2
0
0
2
-
7
2
0
0
2
-
9
2
0
0
2
-
1
1
2
0
0
3
-
1
2
0
0
3
-
3
2
0
0
3
-
5
Cyclical Component
Cy3
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
1
9
9
7
-7
1
9
9
7
-9
1
9
9
7
-1
1
1
9
9
8
-1
1
9
9
8
-3
1
9
9
8
-5
1
9
9
8
-7
1
9
9
8
-9
1
9
9
8
-1
1
1
9
9
9
-1
1
9
9
9
-3
1
9
9
9
-5
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Seasonal Component
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
1
9
9
7
-
7
1
9
9
7
-
9
1
9
9
7
-
1
1
1
9
9
8
-
1
1
9
9
8
-
3
1
9
9
8
-
5
1
9
9
8
-
7
1
9
9
8
-
9
1
9
9
8
-
1
1
1
9
9
9
-
1
1
9
9
9
-
3
1
9
9
9
-
5
1
9
9
9
-
7
1
9
9
9
-
9
1
9
9
9
-
1
1
2
0
0
0
-
1
2
0
0
0
-
3
2
0
0
0
-
5
2
0
0
0
-
7
2
0
0
0
-
9
2
0
0
0
-
1
1
2
0
0
1
-
1
2
0
0
1
-
3
2
0
0
1
-
5
2
0
0
1
-
7
2
0
0
1
-
9
2
0
0
1
-
1
1
2
0
0
2
-
1
2
0
0
2
-
3
2
0
0
2
-
5
2
0
0
2
-
7
2
0
0
2
-
9
2
0
0
2
-
1
1
2
0
0
3
-
1
2
0
0
3
-
3
2
0
0
3
-
5
42
Fig 8: 3-year FRHL, Interest Rates and Trend, Cyclical and Seasonal Components
Observed Interest Rates and Trend
6
6.5
7
7.5
8
8.5
9
9.5
10
10.5
1
9
9
5
-
7
1
9
9
5
-
1
0
1
9
9
6
-
1
1
9
9
6
-
4
1
9
9
6
-
7
1
9
9
6
-
1
0
1
9
9
7
-
1
1
9
9
7
-
4
1
9
9
7
-
7
1
9
9
7
-
1
0
1
9
9
8
-
1
1
9
9
8
-
4
1
9
9
8
-
7
1
9
9
8
-
1
0
1
9
9
9
-
1
1
9
9
9
-
4
1
9
9
9
-
7
1
9
9
9
-
1
0
2
0
0
0
-
1
2
0
0
0
-
4
2
0
0
0
-
7
2
0
0
0
-
1
0
2
0
0
1
-
1
2
0
0
1
-
4
2
0
0
1
-
7
2
0
0
1
-
1
0
2
0
0
2
-
1
2
0
0
2
-
4
2
0
0
2
-
7
2
0
0
2
-
1
0
2
0
0
3
-
1
2
0
0
3
-
4
Cyclical Component
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
1
9
9
5
-7
1
9
9
5
-1
1
1
9
9
6
-3
1
9
9
6
-7
1
9
9
6
-1
1
1
9
9
7
-3
1
9
9
7
-7
1
9
9
7
-1
1
1
9
9
8
-3
1
9
9
8
-7
1
9
9
8
-1
1
1
9
9
9
-3
1
9
9
9
-7
1
9
9
9
-1
1
2
0
0
0
-3
2
0
0
0
-7
2
0
0
0
-1
1
2
0
0
1
-3
2
0
0
1
-7
2
0
0
1
-1
1
2
0
0
2
-3
2
0
0
2
-7
2
0
0
2
-1
1
2
0
0
3
-3
Seasonal Component
-0.12
-0.08
-0.04
0
0.04
0.08
0.12
1
9
9
5
-
7
1
9
9
5
-
1
1
1
9
9
6
-
3
1
9
9
6
-
7
1
9
9
6
-
1
1
1
9
9
7
-
3
1
9
9
7
-
7
1
9
9
7
-
1
1
1
9
9
8
-
3
1
9
9
8
-
7
1
9
9
8
-
1
1
1
9
9
9
-
3
1
9
9
9
-
7
1
9
9
9
-
1
1
2
0
0
0
-
3
2
0
0
0
-
7
2
0
0
0
-
1
1
2
0
0
1
-
3
2
0
0
1
-
7
2
0
0
1
-
1
1
2
0
0
2
-
3
2
0
0
2
-
7
2
0
0
2
-
1
1
2
0
0
3
-
3
43
Fig 9: 4-year FRHL, Interest Rates and Trend, Cyclical and Seasonal Components
Observed Interest Rates and Trend
5
5.5
6
6.5
7
7.5
8
8.5
1
9
9
9
-
7
1
9
9
9
-
9
1
9
9
9
-
1
1
2
0
0
0
-
1
2
0
0
0
-
3
2
0
0
0
-
5
2
0
0
0
-
7
2
0
0
0
-
9
2
0
0
0
-
1
1
2
0
0
1
-
1
2
0
0
1
-
3
2
0
0
1
-
5
2
0
0
1
-
7
2
0
0
1
-
9
2
0
0
1
-
1
1
2
0
0
2
-
1
2
0
0
2
-
3
2
0
0
2
-
5
2
0
0
2
-
7
2
0
0
2
-
9
2
0
0
2
-
1
1
2
0
0
3
-
1
2
0
0
3
-
3
2
0
0
3
-
5
Cyclical Component
Cys
-0.3
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
1
9
9
9
-
7
1
9
9
9
-
9
1
9
9
9
-
1
1
2
0
0
0
-
1
2
0
0
0
-
3
2
0
0
0
-
5
2
0
0
0
-
7
2
0
0
0
-
9
2
0
0
0
-
1
1
2
0
0
1
-
1
2
0
0
1
-
3
2
0
0
1
-
5
2
0
0
1
-
7
2
0
0
1
-
9
2
0
0
1
-
1
1
2
0
0
2
-
1
2
0
0
2
-
3
2
0
0
2
-
5
2
0
0
2
-
7
2
0
0
2
-
9
2
0
0
2
-
1
1
2
0
0
3
-
1
2
0
0
3
-
3
2
0
0
3
-
5
Seasonal Component
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
1
9
9
9
-
7
1
9
9
9
-
9
1
9
9
9
-
1
1
2
0
0
0
-
1
2
0
0
0
-
3
2
0
0
0
-
5
2
0
0
0
-
7
2
0
0
0
-
9
2
0
0
0
-
1
1
2
0
0
1
-
1
2
0
0
1
-
3
2
0
0
1
-
5
2
0
0
1
-
7
2
0
0
1
-
9
2
0
0
1
-
1
1
2
0
0
2
-
1
2
0
0
2
-
3
2
0
0
2
-
5
2
0
0
2
-
7
2
0
0
2
-
9
2
0
0
2
-
1
1
2
0
0
3
-
1
2
0
0
3
-
3
2
0
0
3
-
5
44
Fig 10: 5-year FRHL, Interest Rates and Trend, Cyclical and Seasonal Components
Observed Interest Rates and Trend
6.5
6.7
6.9
7.1
7.3
7.5
7.7
7.9
8.1
8.3
1
9
9
9
-
7
1
9
9
9
-
9
1
9
9
9
-
1
1
2
0
0
0
-
1
2
0
0
0
-
3
2
0
0
0
-
5
2
0
0
0
-
7
2
0
0
0
-
9
2
0
0
0
-
1
1
2
0
0
1
-
1
2
0
0
1
-
3
2
0
0
1
-
5
2
0
0
1
-
7
2
0
0
1
-
9
2
0
0
1
-
1
1
2
0
0
2
-
1
2
0
0
2
-
3
2
0
0
2
-
5
Cyclical Component
-0.000025
-0.00002
-0.000015
-0.00001
-0.000005
0
0.000005
0.00001
0.000015
0.00002
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
Seasonal Component
Sea
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
1
9
9
9
-
7
1
9
9
9
-
8
1
9
9
9
-
9
1
9
9
9
-
1
0
1
9
9
9
-
1
1
1
9
9
9
-
1
2
2
0
0
0
-
1
2
0
0
0
-
2
2
0
0
0
-
3
2
0
0
0
-
4
2
0
0
0
-
5
2
0
0
0
-
6
2
0
0
0
-
7
2
0
0
0
-
8
2
0
0
0
-
9
2
0
0
0
-
1
0
2
0
0
0
-
1
1
2
0
0
0
-
1
2
2
0
0
1
-
1
2
0
0
1
-
2
2
0
0
1
-
3
2
0
0
1
-
4
2
0
0
1
-
5
2
0
0
1
-
6
2
0
0
1
-
7
2
0
0
1
-
8
2
0
0
1
-
9
2
0
0
1
-
1
0
2
0
0
1
-
1
1
2
0
0
1
-
1
2
2
0
0
2
-
1
2
0
0
2
-
2
2
0
0
2
-
3
2
0
0
2
-
4
2
0
0
2
-
5
2
0
0
2
-
6
45
Fig 11: LOC Home Loan, Interest Rates and Trend, Cyclical and Seasonal Components
Observed Interest Rates and Trend
5
5.5
6
6.5
7
7.5
8
8.5
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Cyclical Component
Cy3
-0.000004
-0.000003
-0.000002
-0.000001
0
0.000001
0.000002
0.000003
0.000004
0.000005
1
9
9
9
-7
1
9
9
9
-9
1
9
9
9
-1
1
2
0
0
0
-1
2
0
0
0
-3
2
0
0
0
-5
2
0
0
0
-7
2
0
0
0
-9
2
0
0
0
-1
1
2
0
0
1
-1
2
0
0
1
-3
2
0
0
1
-5
2
0
0
1
-7
2
0
0
1
-9
2
0
0
1
-1
1
2
0
0
2
-1
2
0
0
2
-3
2
0
0
2
-5
2
0
0
2
-7
2
0
0
2
-9
2
0
0
2
-1
1
2
0
0
3
-1
2
0
0
3
-3
2
0
0
3
-5
Seasonal Component
Sea
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
1
9
9
9
-
7
1
9
9
9
-
9
1
9
9
9
-
1
1
2
0
0
0
-
1
2
0
0
0
-
3
2
0
0
0
-
5
2
0
0
0
-
7
2
0
0
0
-
9
2
0
0
0
-
1
1
2
0
0
1
-
1
2
0
0
1
-
3
2
0
0
1
-
5
2
0
0
1
-
7
2
0
0
1
-
9
2
0
0
1
-
1
1
2
0
0
2
-
1
2
0
0
2
-
3
2
0
0
2
-
5
2
0
0
2
-
7
2
0
0
2
-
9
2
0
0
2
-
1
1
2
0
0
3
-
1
2
0
0
3
-
3
2
0
0
3
-
5
46
Basic Home Loan (BHL)
With respect to NABs BHL, of the 11
seasonal factors presented in Table 1, only
those corresponding to May (
2
) and April (
3
)
were found to be statistically signicant. The
two observed seasonal factors correspond to
the end of the nancial year season. The end
of the nancial year season generally starts
by the end of April or the beginning of May,
and nishes at the end of the rst week in
July. The primary cause of this seasonal
effect is that the Australian Taxation Ofce
(ATO) accepts prepayment of the interest
payable on investment properties. Hence,
towards the end of the nancial year,
investors seek nancial tools that minimise
their tax exposure, and, as such, optimise
their wealth creation opportunities. The
same is applicable to a lesser extent to
owner-occupiers, who may come in larger
numbers to the market, attracted by lower
interest rates.
In summary, both of the statistically
signicant seasonal factors observed relate
to the end of the nancial year season and
have negative values, indicating seasonal
rate reductions. Although, at a face value,
it seems that the existence of the end of the
nancial year seasonal effect is conrmed
for BHL, it (BHL) is not a product that was
expected to experience increased demand
during the end of the nancial year season.
More precisely, although the factors for
NABs BHL exhibit statistically signicant
seasonality in those 2 months during the
observed period, it is not offered as an
interest only (IO) product and does not allow
for prepayment of interest. Furthermore, a
BHL being a variable rate home loan is not
a product that should have been used for
price-driven initiatives. Using a variable
rate home loan and offering sizeable rate
discounts would be too expensive for a bank
such as NAB, as it would, at a minimum,
re-price its entire BHL book
23
. The above
arguments refute the possibility of linking
the observed seasonality to the end of the
nancial year season.
Another possible explanation is that, during
the observed period, NAB reduced the BHL
interest rates during the end of the nancial
year season, hoping to attract an increased
number of borrowers. This explanation is
also easily refutable in view of the product
functionalities and the way it is priced. In
addition to the arguments raised thus far,
it is important to note that because the
cost of funds (COFs) for BHL are primarily
determined by the cash rate, a BHLs
interest rate principally moves only with the
movement of the cash rate. Furthermore, a
BHL is already a heavily discounted product;
hence it is almost never used for rate-driven
initiatives.
Having refuted the above two possible
explanations and knowing that the BHLs
interest rate moves principally with the
movement of the cash rate only, it is
logical to check whether the cause of the
observed seasonality bears any relationship
to movements in the cash rate. As already
established, the COFs for a BHL are
predominantly inuenced by the cash rate,
which is directly controlled by the RBA.
Hence, this interpretation can be conrmed
or refuted by considering whether there has
been a concentration of the RBAs cash rate
reductions coinciding with the end of the
nancial year season and the time horizon
of the analysis. The cash rate reductions
would have to be of a particular magnitude
and time distribution to be able to cause
the statistically signicant seasonal factors
observed.
The result of checking whether the RBA
has reduced the cash rate in April and May,
during the observed period, is encouraging.
The cash rate changes observed support
this interpretation, revealing that in four
of the 5 years observed, the RBA has
actually reduced the cash rate during one
or both of those two months or the two
months surrounding the period, i.e. March
and June. To conclude, it seems that the
explanation for the observed seasonality is
related to the particular stage of the interest
rate cycle and is not related to the end of the
nancial year season.
Standard Variable Rate Home Loan (SVRHL)
The modelling results obtained for the
SVRHL are almost identical to those obtained
for the BHL
24
, i.e. of the 11 seasonal factors
presented in Table 1, the same two seasonal
factors were found to be statistically
signicant (the factors corresponding to
May (
2
) and April (
3
)). As in the case of the
BHL, both of the statistically signicant
23 The impact may be even larger as pricing of other variable rate home loans is commonly linked to the BHL interest
rate. For example, NAB has established a relationship between the BHL and the SVRHL, where the BHL is always
priced at a discount of 0.5% to the SVRHL.
24 The only difference is that the estimated values are somewhat different.
47
seasonal factors observed have a negative
value, implying the presence of the end of
the nancial year seasonal effect.
Because of the product similarity (among
other things, both products, being variable
rate home loans, are priced using the same
COFs) and the exact correspondence of
the data available for both the BHL and
the SVRHL, for the same reasons outlined
when interpreting the BHLs statistically
signicant seasonal factors, SVRHLs
statistically signicant seasonal factors
cannot be attributed to the end of the
nancial year seasonal effect. Consequently,
the interpretation for the observed
statistically signicant seasonal factors is
the same as that provided for the BHLs
statistically signicant factors, i.e. it seems
to be related to the particular stage of the
interest rate cycle.
6-Month Fixed Rate Introductory Home Loan
Contrary to the BHL and the SVRHL, the
modelling results for NABs 6-month xed
rate Introductory home loan, presented in
Table 1, show that of the 11 seasonal factors,
four were found to be statistically signicant,
namely, the factors corresponding to March
(
4
), August (
11
), September (
10
) and October
(
9
).
Primarily due to the product functionalities,
the expectation for both the 6- and 12-month
xed rate Introductory home loans was that
they would exhibit statistically signicant
seasonal interest rate reductions for
both the Spring and Autumn seasons.
Nevertheless, the modelling results obtained
for NABs 6-month xed rate Introductory
home loan signicantly deviate from the
expected results. This relates, in particular,
to the observed statistically signicant
interest rate increases for September
(
10
) and October (
9
), and the statistically
insignicant seasonal factor for November
(
8
).
The absence of statistically signicant
seasonal rate reductions for the 6-month
xed rate Introductory home loan, during
both the spring and autumn seasons, is
particularly interesting because, during
the observed period, NAB was the only
bank among the major four banks that had
a 6-month xed rate Introductory home
loan, and, as such, it was expected to use
it. A possible explanation for the modelling
results obtained is that the 6-month xed
rate Introductory home loan is relatively
new on the market, i.e. only a few years of
monthly data were available at the time of
the analysis. For this reason, it would be
interesting to replicate the analysis after
more data become available.
The nding of the presence of a statistically
signicant rate increase in August (
11
)
is also contradictory to expectations. An
explanation for the statistically signicant
interest rate increase in August (
11
) may
be a post-end of the nancial year season
correction. Although not typically used
during the end of the nancial year season,
the 6-month xed rate Introductory home
loan has been occasionally used as a part of
a pricing strategy during this time. Another
possible explanation is the possibility that
the smaller amount of data distorted the
results obtained.
With respect to the statistically signicant
March (
4
) seasonal factor, the result is as
expected and corresponds to the autumn
seasonal effect. During this low home loan
demand season, lenders tend to reduce
interest rates aiming to reinvigorate this
otherwise low home loan demand period.
12-Month Fixed Rate Introductory Home Loan
Of the 11 seasonal factors for the 12-month
xed rate Introductory home loan, presented
in Table 1, only the one corresponding to
March (
4
) was found to be statistically
signicant. The absence of statistically
signicant seasonal factors during the
spring season is surprising. The same
explanation offered for the 6-month xed rate
Introductory home loan is also applicable to
the 12-month xed rate Introductory home
loan. As previously stated, the absence of
statistically signicant seasonal factors
for the end of the nancial year season is
as expected, as the 12-month xed rate
Introductory home loan is very rarely used
during the end of the nancial year season.
In summary, as with the 6-month xed
rate Introductory home loan, the presence
of a statistically signicant seasonal rate
reduction for March (
4
) implies the existence
of the Autumn seasonal effect.
1-Year Fixed Rate Home Loan (FRHL)
Of the 11 seasonal factors for the 1-year
FRHL, shown in Table 1, only one was
found to be statistically signicant, namely
the seasonal interest rate reduction
corresponding to January (
6
).
With the introduction of Introductory home
loans, the importance of the 1-year FRHL
has signicantly diminished. In particular,
48
after the introduction of the 12-month xed
rate Introductory home loan, borrowers
could opt for a more functional and more
competitively priced 12-month xed rate
Introductory home loan instead of a 1-year
FRHL. Therefore, it is not surprising to
nd the absence of statistically signicant
seasonal factors during the spring season.
The product could have been utilised for
the end of the nancial year season, as a
1-year FRHL allows for interest prepayment
together with an IO payment. However,
this is not shown in the modelling results
as factors corresponding to April (
3
), May
(
2
) and June (
1
) were found not to exhibit
statistically signicant seasonality. As
stated above, the only statistically signicant
factor is the one for January (
6
), which
corresponds to the beginning of the autumn
season. This is somewhat unexpected as,
for reasons outlined above, this product is
not typically used for either the spring or
autumn seasons.
2-Year FRHL
The modelling results obtained for the 2-year
FRHL show that the March (
4
), August
(
11
) and September (
10
) seasonal factors
were found to be statistically signicant.
As with the interpretation offered for the
August (
11
) result for the 6-month xed rate
Introductory home loan, an explanation for
the statistically signicant factors in August
(
11
) and September (
10
) may be a post-end
of the nancial year season correction, as
this product is known to be used for the
end of the nancial year season. A difculty
with this explanation is that statistically
signicant interest rate reductions were not
observed during the end of the nancial year
season. A more conclusive explanation for
the observed statistically signicant rate
increases in August (
11
) and September
(
10
) could only be reached through a more
detailed analysis of associated changes in
the yield curve and pricing strategies that
used the 2-year FRHL during the observed
period.
With respect to the statistically signicant
March (
4
) seasonal factor, this is as
expected, as it corresponds with the
autumn seasonal effect. The estimation
results strongly support the claim, showing
a statistically signicant rate reduction
during that month.
3-Year FRHL
Of the 11 seasonal factors presented in
Table 1 for the 3-year FRHL, none is found
to be statistically signicant. This result
is contrary to what is expected because
the 3-year FRHL is, anecdotally, known
to be heavily used by NAB as a part of its
xed rate pricing strategy during all three
identied seasons (spring, autumn and the
end of the nancial year).
To be able to provide a more conclusive
explanation for the observed modelling
results, a more detailed analysis of
associated changes in the yield curve and
pricing strategies that use the 3-year FRHL
during the observed period would be needed.
4-Year FRHL
The modelling results obtained for the
4-year FRHL show that of the 11 seasonal
factors presented in Table 1, the factors
corresponding to April (
3
), March (
4
),
September (
10
) and August (
11
) are found
to be statistically signicant. The factors
corresponding to April (
3
) and March (
4
)
showed statistically signicant interest rate
reductions, while the factors corresponding
to September (
10
) and August (
11
) show
statistically signicant interest rate
increases.
Statistically signicant interest rate
reductions for the March (
4
) and April (
3
)
factors correspond to the autumn season.
On the other hand, because the 4-year
FRHL is known to be used for the end of
the nancial year season, an explanation
for the statistically signicant rate increases
in August and September may be a post-
end of the nancial year season correction.
In line with the interpretation provided for
the 6-month xed rate Introductory home
loan and the 2-year FRHL, a difculty
with this explanation is that statistically
signicant interest rate reductions have
not been observed during the end of the
nancial year season. Therefore, the scope
of the analysis does not allow for a more
conclusive explanation, hence the need for
additional analysis of, among other things,
associated changes in the yield curve and
pricing strategies that used a 4-year FRHL
during the observed period.
5-Year FRHL
With respect to the 11 seasonal factors for
the 5-year FRHL, shown in Table 1, none is
found to be statistically signicant. As for
the 3-year FRHL, this result is surprising,
primarily because the 5-year FRHL is,
anecdotally, known to be used by NAB as
a supporting product to the 3-year FRHL
for its xed rate pricing strategy. Therefore,
49
the future research possibility suggested for
the 3-year FRHL also applies equally to the
5-year FRHL.
LOC Home Loan
As with the other variable rate home loans
(BHL and SVRHL), with respect to the 11
seasonal factors for the LOC home loan,
presented in Table 1, the same two factors
are found to be statistically signicant,
namely, the factors corresponding to May
(
2
) and April (
3
). Due to product similarity
(all products being variable rate products
and being priced by the same COFs) and
the same time period considered, the same
explanation offered for the other two variable
rate home loan products applies to the LOC
home loan as well. Therefore, an explanation
for the seasonality observed is most likely
unrelated to the banks intervention but
related to a particular stage of the interest
rate cycle.
V. CONCLUSION
The purpose of this paper has been to model
product-level home loan interest rates for
one of the biggest banks in Australia (NAB)
using a structural time series modelling
approach. The modelling focus has been
to establish whether or not selected home
loan interest rates (product-level monthly
home loan interest rates for NAB) exhibit the
expected cyclical and seasonal variations
and whether seasonality, if present, is
stochastic or deterministic.
The modelling results, with respect to the
expected cyclicality of home loan interest
rates, largely conrm the existence of
cyclicality in home loan interest rates
for NAB. As illustrated in Table 1, the
magnitude of possible cycle-related changes
on home loan interest rates ranges from
around 0.10% to almost 0.41%. The nding
is not conclusive as two variables indicate
the absence of a statistically signicant
cyclical factor. To that end, a possible
explanation for those factors is offered, and
several possibilities for future research that
may lead to a more conclusive explanation
are indicated.
With respect to home loan interest rate
seasonality, most of the variables analysed
(with the exception of NABs 3- and 5-year
FRHL) show the presence of statistically
signicant seasonal factors. The magnitude
of these seasonal changes range from around
0.10% to almost 0.90%. Furthermore,
the seasonal factors presented appear to
be regular for all variables, except for the
12-month xed rate Introductory home loan,
implying the deterministic nature of all the
other seasonal factors observed, and, the
stochastic nature of the 12-month xed rate
Introductory home loan seasonal factors.
Although most of the variables analysed
show the presence of statistically signicant
seasonal factors, the majority of the observed
statistically signicant seasonal factors
cannot be attributed to any of the three
considered seasonal effects. Nonetheless,
even for the observed statistically signicant
seasonal factors that cannot be attributed
to any of the three seasonal effects, a
possible explanation has been offered.
Moreover, since a number of statistically
signicant seasonal factors do not allow for
a conclusive explanation (primarily due to
the scope of the analysis and the relatively
short period of data available), so several
interesting possibilities for future research
(such as extending the scope of the analysis
by testing the same data set for another
of the major four Australian banks) are
indicated. Finally, due to a recent signicant
turbulence on nancial markets world wide,
it will be exceptionally interesting, when
data becomes available, to include the time
series for the current nancial crisis into the
analysis.
References
Al-Sad K. and Moosa I. A. (2005). Seasonality
in stock returns: Evidence from an emerging
market. Applied Financial Economics, 15:
6371
Berger-Thomson, L. and Ellis, L. (2004).
Housing construction cycles and interest
rates. Economic Group, Reserve Bank of
Australia, www.rba.gov.au
Campbell, F. and Lewis, E. (1998). What moves
yields in Australia. Reserve Bank of Australia,
www.rba.gov.au
Ellis, L. and Lowe, P. (2002). The smoothing
of official interest rates. Reserve Bank of
Australia, www.rba.gov.au
Harvey, A. C. and Todd, P. H. J. (1983).
Forecasting economic time series with
structural and Box-Jenkins models: A case
study. Journal of Economic and Business
Statistics, 1: 299307
Harvey, A. C. (1985). Trends and cycles in
macroeconomic time series. Journal of
Business and Economic Statistics, 13(3):
216227
Harvey, A. C. (1990). Forecasting, Structural
Time Series Models and the Kalman Filter,
Cambridge University Press, Cambridge
50
Harvey, A. C. (1997). Trends, cycles and
autoregressions. Economic Journal, 107:
192201
Harvey, A. C. and Scott, A. (1994) Seasonality in
dynamic regression models. Economic Journal,
104: 13241345
Harvey, A. C. and Trimbur, T. M. (2003). General
model-based lters for extracting cycles and
trends in economic time series. Review of
Economics and Statistics, 85(2): 244255
Australian Taxation Of fice. Income Tax
Assessment Act 1997, http://law.ato.gov.
au/atolaw/execute_search_java.htm
Karamujic, M. H. (2008). A classication of
home loan products in Australia, refereed
proceedings of 2009 PRRES Conference,
University of Technology, Sydney, 1821
January
Karamujic, M. H. (2009). The Determination of
Home Loan Interest Rates and an Empirical
Assessment of their Cyclicality and Seasonality
in Australia, VDM, Saarbrucken, Germany
Koopman, S. J., Harvey, A. C., Doornik, J. A.
and Shephard, N. (1995). Stamp 5.0 Structural
Time Series Analyser, Modeller and Predictor,
Chapman and Hall, London
Koopman, S. J. (1997). Exact initial Kalman
ltering and smoothing for non-stationary
time series. Journal of American Statistical
Association, 92(440): 16301638
Moosa, I. A. (1998). An investigation into the
cyclical behaviour of output, money, stock
prices and interest rates, Applied Economics
Letters, 5(4): 235238
Moosa, I. A. (1999). On the cyclical behaviour of
prices. Economic Issues, 4(2): 107122
51
Impact of oil price shocks on government
revenue and expenditure: Evidence for Malaysia
Dr Tan Juat Hong
Associate Professor, College of Graduate Studies,Universiti Tenaga Nasional, Malaysia
ABSTRACT: The paper investigates the short-run impact of oil price shocks on government revenue
and expenditure of Malaysia. The empirical results from the impulse response functions and variance
decomposition suggest that oil price shocks have signicant positive impact on both government revenue and
spending. The oil price shocks are further segregated into positive and negative shocks. Evidence suggests that
asymmetric oil price shocks exert positive impact on government revenue and expenditure.
I. INTRODUCTION
World oil price uctuations have been the main contentious issue among net oil-exporting
as well as oil-importing countries vis--vis their economies. Many studies especially in the
1980s indicate that oil price uctuations were associated with adverse economic activities.
For instance, Hamilton (1983), Burbidge and Harrison (1984) and Mork (1989) apprise that
oil price increases have a negative impact on economic activity. However, studies in the 2000s
such as Abeysinghe (2001), Barsky and Kilian (2004) and Cunado and Gracia (2005) posit
that the impact of oil price rises on economic activity may be somewhat limited. Perhaps this
phenomenon as suggested by Doroodian and Boyd (2002) could be due to better efciency in
the utilisation of energy, intensifying energy conservation policy, decreasing energy intensity
and greater usage of alternative energy sources.
By using both linear and non-linear specications, Jbir and Zourari-Ghorbel (2009) conclude
that there is no direct impact of oil price shocks on economic activity in Tunisia, but
shocks indirectly affect economic activity in the Tunisian economy. They also conclude that
government spending is the most signicant channel by which the effects of oil price shocks
are transmitted.
Lorde et al (2009) investigate the macroeconomic effects of oil price uctuations on
Trinidad and Tobago. They nd that oil price volatility brings about random swings in the
macroeconomy with government revenue and price level exhibiting signicant responses.
Although the response of government spending to oil price uctuations is much smaller than
for revenue, it remains insignicant statistically.
Given the lack of empirical studies focusing on the effects of oil price uctuations on
government revenue and spending, the paper purports to ll the research gap for a small
net-oil exporting country like Malaysia. The main objective of this study is to investigate
the response of the government scal budget to oil price innovations by using linear and
non-linear specications. It provides empirical evidence on how government revenue and
expenditure respond to oil price shocks over the horizon.
The paper is structured as follows: Section II details the data sources and variables used.
Section III describes the methodology employed, while the statistical results are presented in
Section IV. The conclusion is provided in Section V.
II. DATA AND VARIABLES
The study uses annual data for the period 1970 till 2008 to capture the effects of oil price
uctuations on government revenue and spending. The world oil price (oilp) is expressed in
US$/barrel. The world oil prices are based on crude oil US prices (19701990) and Asia-
Pacic Tapis prices (19912007). The annual government revenue (rev) and spending (exp)
(including operating and net development expenditures) data is compiled from the various
issues of the Monthly Statistical Bulletin published by Bank Negara Malaysia (BNM). These
endogenous variables are transformed into natural logarithms for the purpose of analysis.
52
In order to accommodate the possibility of
an asymmetrical response of government
revenue and spending to the world oil
price (oilp), the world oil price variable is
further segregated into positive and negative
price shocks. Following Mork (1989), the
asymmetric oil price shocks are shown by
the expressions given as follows:
For positive oil price shocks at time t:
oilp +
t
= max (0, oilp
t
oilp
t1
) (1a)
For negative oil price shocks at time t:
oilp
t
= min (0, oilp
t
oilp
t1
) (1b)
III. THE VAR MODEL
To capture the world crude oil price pass-
through into government revenue and
expenditure, the study uses the vector
autoregression (VAR) methodology specied
as follows:
UY
t
=o+ E
p
j=1
|
j
UY
t-j
+
t
[2]
where A is the lag operator, Y
t
is a (n x 1)
vector of the endogenous variables of
the VAR system at time t, o is the (n x 1)
intercept vector, |
j
is the (n x n) matrix of
the autoregressive coefcients of j = 1, 2,
... p lags, and
t
is the (n x 1) white noise
term. The optimal lag period in the VAR
system is determined using the Akaike
Information Criterion (AIC). In this study,
the optimal lag length of 5 periods is used
for the VAR estimation. The lag operator
denotes that these series used in equation
(2) are rst-differenced to ensure time-series
stationarity.
The VAR system will then be transformed
into a moving average (MA) representation
as illustrated below:
UY
t
=c+ E
j=0
o
j
t-j
[3]
where c is the mean process and o
j
(j =
0, 1 ... ) the identity matrix of the MA
representation. The MA representation
is used to generate the impulse response
functions and the variance decomposition of
the endogenous variables.
The impulse response functions trace the
impact of symmetric and asymmetric oil price
shocks on the government scal budget.
The orthogonal innovations are obtained
by factoring the variance-covariance matrix
of the VAR model by using the Cholesky
decomposition method. The condence
levels of the impulse response functions
are generated by using 1000 Monte Carlo
simulations for a 10-year forecast horizon.
The Cholesky decomposition used in the
regression follows the order given as: oilp,
rev and exp.
IV. EMPIRICAL RESULTS
(a) World Oil Price Trend
Figure 1 depicts the world crude oil prices as
expressed in US$ per barrel. The statistics
show that oil prices dipped in 1985/86 as
the commodity market collapsed in mid-
1980 and 1997/98 (perhaps due to the East
Asian nancial crisis). On the other hand,
world oil prices peaked in 1973/74 (the
Arab oil embargo), in 1978/79 (the Iranian
revolution), in 1980 (with the start of Iran-
Iraq war), in 1990/91 (the Gulf war), in
2002/03 (the Venezuelan crisis) and was
further impounded by the Iraq war in 2003.
The instability of world oil prices is further
compounded by the avaricious demand for
oil imports from consuming countries like
China and India to fuel their rapidly growing
economies. Meanwhile oil production is
affected by the uncertainties in oil producing
countries such as Nigeria, Venezuela and as
well as the geopolitical tensions between
Iran and the US on nuclear enrichment. The
declining value of the US Dollar has also
created an environment of uncertainty in oil
market futures due to massive speculation.
(b) Government Fiscal Budget
Figure 2 shows the rst-difference of
government revenue and spending
(inclusive of operating and net development
expenditures) expressed in Ringgit Malaysia.
The volatility trend for both series indicates
a relatively high positive correlation. In his
study on causality of government revenue
and expenditure for Malaysia, Tan (2009)
concludes that government revenue and
spending are cointegrated and empirical
results support the spend-and-tax
hypothesis.
(c) Impulse Response Functions
In this sub-section, Figures 3a through
3c depict the statistical results of the
orthogonal impulse response functions of
oil price shocks on government scal budget
for a 10-year forecast horizon.
Figure 3a displays the impulse response
functions of symmetric oil price shocks (oilp)
on government revenue and expenditure for
a 10-year horizon. The results indicate that
government revenue and spending respond
positively to oil price shocks in the second-
53
-.2
-.1
.0
.1
.2
.3
.4
1970 1975 1980 1985 1990 1995 2000 2005
expenditure revenue
0
20
40
60
80
100
120
1970 1975 1980 1985 1990 1995 2000 2005
Figure 1: World crude oil prices (1970 - 2008)
Figure 2: Volatility of government revenue and expenditure (1970 - 2008)
Source: Monthly Statistical Bulletin, BNM (various issues)
Source: Monthly Statistical Bulletin, BNM (various issues)
World crude oil prices (US$/barrel)
revenue expenditure
54
year, and are statistically signicant. At
the zenith in year 2, a one-standard (sd)
deviation of symmetric innovations exerts
a positive impact of 0.053% and 0.057%
on government revenue and spending
respectively. Thereafter in the longer
horizon, the results are not signicant
statistically. Thus, the impact of symmetric
oil price shocks on government revenue
and spending is deemed a short-trend
phenomenon.
Figure 3b displays the impulse response
functions of positive oil price shocks (oilp+)
on government revenue and expenditure.
Despite the positive shocks, the results
indicate that government revenue responds
negatively in the rst period and then,
positively in the second-year. However,
-.4
-.3
-.2
-.1
.0
.1
.2
.3
.4
1 2 3 4 5 6 7 8 9 10
Response of oilp to oilp
-.15
-.10
-.05
.00
.05
.10
.15
1 2 3 4 5 6 7 8 9 10
Response of rev to oilp
-.20
-.16
-.12
-.08
-.04
.00
.04
.08
.12
.16
1 2 3 4 5 6 7 8 9 10
Response of exp to oilp
-.4
-.3
-.2
-.1
.0
.1
.2
.3
.4
1 2 3 4 5 6 7 8 9 10
Response of oilp+ to oilp+
-.12
-.08
-.04
.00
.04
.08
.12
1 2 3 4 5 6 7 8 9 10
Response of rev to oilp+
-.3
-.2
-.1
.0
.1
.2
.3
1 2 3 4 5 6 7 8 9 10
Response of exp to oilp+
-.6
-.4
-.2
.0
.2
.4
.6
1 2 3 4 5 6 7 8 9 10
Response of oilp- to oilp-
-.4
-.3
-.2
-.1
.0
.1
.2
.3
.4
1 2 3 4 5 6 7 8 9 10
Response of rev to oilp-
-.4
-.3
-.2
-.1
.0
.1
.2
.3
.4
1 2 3 4 5 6 7 8 9 10
Response of exp to oilp-
government spending continues to respond
positively to positive oil price shocks in both
the rst and second year. At the zenith, a one-
standard deviation positive oil price shocks
exerts a proportion of 0.049% and 0.046%
on government revenue and spending
respectively in the second year. After the
second year, the responses of government
revenue and spending, though positive from
oil price shocks remain insignicant.
Figure 3c displays the impulse response
functions of negative oil price shocks (oilp-)
on government revenue and expenditure.
Despite the negative oil price shocks,
government revenue and spending respond
positively to the effects in the rst and
second year. In the second year, a one-
standard deviation positive oil price shocks
Figure 3a: Response to Cholesky One S.D. Innovations 2 S.E.
(Linear Oil Price Shocks)
Figure 3b: Response to Cholesky One S.D. Innovations 2 S.E.
(Positive Oil Price Shocks)
Figure 3c: Response to Cholesky One S.D. Innovations 2 S.E.
(Negative Oil Price Shocks)
55
exerts a proportion of 0.038% and 0.045%
on government revenue and spending
respectively. After the second year, both
these responses remain statistically
insignicant.
(d) Variance Decomposition
In this sub-section, Figure 1a through
Figure 1c display the results of forecast
error variance decomposition for oil price
shocks, and government revenue and
expenditure. The statistical results highlight
the proportion of unanticipated change
in oil price and government revenue and
expenditure attributable to an oil price
shock (encompassing linear, positive and
negative shocks).
Table 1a depicts the results of the estimated
orthogonal variance decomposition of
oil price and government revenue and
expenditure attributed to a linear oil price
shock (oilp). Comparing the oil price shocks
on government revenue and expenditure
in the 10th-period, oil price shocks have a
Table 1a: Estimated Orthogonal Variance Decomposition
(Linear Oil Price Shocks)
Linear oil price Shocks (oilp)
Period S.E oilp rev exp
1 0.2438 100.000 0.000 0.000
2 0.2709 86.561 7.047 6.392
3 0.2754 83.945 6.822 9.233
4 0.2807 81.589 6.782 11.629
5 0.3046 71.378 18.629 9.994
10 0.3429 63.083 24.117 12.800
Government revenue (rev)
Period S.E oilp rev exp
1 0.0760 4.817 95.183 0.000
2 0.1025 46.223 53.753 0.024
3 0.1041 47.331 52.528 0.141
4 0.1051 47.414 51.921 0.665
5 0.1139 46.910 45.053 8.037
10 0.1321 40.507 46.196 13.297
Government spending (exp)
Period S.E oilp rev exp
1 0.0783 3.785 6.727 89.489
2 0.1112 45.053 3.348 51.599
3 0.1200 50.479 3.446 46.075
4 0.1210 50.272 3.406 46.321
5 0.1226 49.308 3.319 47.373
10 0.1398 43.203 16.840 39.957
greater impact on expenditure than revenue.
In the same period, linear oil price shocks
contributed 40.5% to government revenue;
and 43.2% on government expenditure.
Table 1b depicts the results of the estimated
orthogonal variance decomposition of
oil price and government revenue and
expenditure attributed to a positive oil
price shock (oilp+). At the 10th year, a
positive oil price shock contributed 32.2% to
government revenue; while exerting 46.1%
on government expenditure. For a positive
oil price shock, the variation of contribution
is greater for government expenditure than
revenue.
Table 1c shows the results of the estimated
orthogonal variance decomposition of
oil price and government revenue and
expenditure attributed to a negative oil
price shock (oilp-). Comparing the results
at the 10th year, a negative oil price shock
contributed 44.5% to government revenue;
while exerting 32.8% on government
56
Table 1b: Estimated Orthogonal Variance Decomposition
(Positive Oil Price Shocks)
Positive oil price shocks (oilp+)
Period S.E oilp+ rev exp
1 0.1499 100.000 0.000 0.000
2 0.1744 94.034 5.171 0.795
3 0.1793 90.459 8.717 0.824
4 0.1840 87.225 8.481 4.293
5 0.1906 81.340 13.923 4.737
10 0.2075 76.988 15.233 7.779
Government revenue (rev)
Period S.E oilp+ rev exp
1 0.0775 3.266 96.734 0.000
2 0.0977 35.328 61.000 3.672
3 0.1004 38.197 58.158 3.645
4 0.1044 39.724 56.546 3.730
5 0.1092 37.834 53.040 9.126
10 0.1232 32.174 53.775 14.051
Government spending (exp)
Period S.E oilp+ rev exp
1 0.0787 15.078 17.816 67.106
2 0.1044 46.975 14.372 38.653
3 0.1129 49.703 12.882 37.414
4 0.1148 49.604 13.990 36.406
5 0.1153 49.178 14.044 36.778
10 0.1306 46.067 21.180 32.753
expenditure. For a negative oil price shock,
the variation of contribution is greater for
government revenue than expenditure.
V. CONCLUSION
The paper deals with the impact of world
oil price shocks on government revenue
and expenditure. The study uses the linear
world price shocks (symmetric) as well as
the positive and negative oil price shocks as
a non-linear model (asymmetric) in the VAR
estimation.
As analysed by the impulse response
functions, the linear model of oil price
shocks suggests that government revenue
and expenditure are affected positively
by world oil price volatility. However, the
shocks remain a short-term phenomenon.
The variable for oil price shocks is further
segregated into positive and negative oil
price shocks in order to measure their
asymmetric impact on the government
budget. The asymmetric results show a
reciprocal relationship between oil price
shocks and the scal budget. A positive oil
price shock would exert a greater magnitude
on government expenditure than revenue.
Conversely, a negative oil price shock would
exert a greater impact on government revenue
than expenditure. All in all, the ndings
suggest that the government scal budget
is affected positively by both symmetric and
asymmetric oil price uctuations.
The results from the analysis of variance
decomposition suggest that a linear oil
price shock contributes 40.5% and 43.2%
in the 10th year horizon to the variability
of government revenue and expenditure
respectively. For the positive oil price shock,
the results indicate that positive oil price
shock contributes 32.2% to the variability
of government revenue and 46.1% to
government expenditure in the 10th year
horizon. In the case of a negative price shock,
the results show that its impact of 44.5%
and 32.8% on government revenue and
expenditure respectively occurs in the 10th
year horizon. In short, a positive oil price
shock would contribute greater variability
to government expenditure, while a negative
oil price shock would exert a greater impact
on the variability of government revenue.
57
Table 1c: Estimated Orthogonal Variance Decomposition
(Negative Oil Price Shocks)
Negative oil price shocks (oilp-)
Period S.E oilp- rev
exp
1 0.1543 100.000 0.000 0.000
2 0.1601 93.601 3.251 3.148
3 0.1619 92.443 4.298 3.259
4 0.1648 90.775 5.816 3.409
5 0.1705 85.769 5.758 8.473
10 0.1804 80.435 7.266 12.299
Government revenue (rev)
Period S.E oilp- rev
exp
1 0.0896 40.475 59.525 0.000
2 0.1042 53.410 44.065 2.525
3 0.1132 45.495 52.240 2.265
4 0.1152 44.344 51.790 3.866
5 0.1191 47.176 48.672 4.152
10 0.1251 44.531 46.267 9.202
Government spending (exp)
Period S.E oilp- rev
exp
1 0.0828 0.854 30.630 68.516
2 0.1174 27.094 24.085 48.821
3 0.1270 27.247 28.284 44.470
4 0.1284 27.678 27.723 44.599
5 0.1331 31.553 26.925 41.522
10 0.1399 32.791 27.475 39.734
References
Abeysinghe, T. (2001). Estimation of direct
and indirect impact of oil prices on growth.
Economic Letters, 73: 147-153
Barsky, R. B. and Kilian, L. (2004). Oil and the
macroeconomy since the 1970s. Journal of
Economic Perspectives, 18(4): 115-134
Burbidge, J. and Harrison, A. (1984). Testing
for the effects of oil-price rise using vector
autoregressions. International Economic
Review, 25: 459-484
Cunado, J. and Gracia, F. (2005). Oil prices,
economic activity and ination: Evidence for
some Asian countries, The Quarterly Review
of Economics and Finance, 45: 65-83
Doroodian, K. and Boyd, R. (2002). The linkage
between oil price shocks and economic growth
with ination in the presence of technological
advances: A CGE model. Energy Policy, 31:
989-1006
Hamilton, J. D. (1983). Oil and the macroeconomy
since World War II, Journal of Political
Economy, 91: 228-248
Jbir, R. and Zouari-Ghorbel, S. (2009). Recent
oil price shock and the Tunisian economy.
Energy Policy, 37: 1041-1051
Lorde, T., Jackman, M. and Thomas, C. (2009).
The macroeconomic ef fects of oil price
uctuations on a small open oil-producing
country: The case of Trinidad and Tobago,
Energy Policy, (in press, accepted and available
online March 2009)
Mork, K. A. (1989). Oil and the macroeconomy
when prices go up and down: An extension
of Hamiltons results. Journal of Political
Economy, 97(3): 740-744
Tan, J. H. (2009). Tax-and-spend or spend-and-
tax? Empirical evidence from Malaysia Journal
of Accounting and Finance, vol. 5(1), 2009)
BANKERS JOURNAL MALAYSIA
Malaysias only journal on banking and nance
SUBSCRIPTION ORDER
Yes, I would like to sign up for a series of 2 (two) consecutive issues to the
Bankers Journal Malaysia.
Name:
IBBM Membership No: (if applicable)
Job title:
Organisation:
Postal Address ( Home Ofce):
Postcode: Town/City:
Country: Date:
Tel: Fax:
E-mail :
Publication Months : June and December
Subscription Instruction
Please commence my subscription with:
Issue #134 (current)
Issue #135
Issue #136
(In BLOCK LETTERS)
Payment Method
(Payment by credit card is applicable only for purchases of RM50.00 and above)
Malaysia
I enclose a cheque / bankers draft / money order no. for the amount: RM
payable to Institut Bank-Bank Malaysia
Overseas
I authorise you to debit my VISA / MasterCard for the amount: RM
Credit Card no: Card Expiry Date: /
Cardholders name as per name on the credit card Cardholders Signature as per credit card
Fill out the subscription order form and return it with your payment to:
INSTITUT BANK-BANK MALAYSIA (35880-P)
Wisma IBI
5 Jalan Semantan, Damansara Heights
50490 Kuala Lumpur, Malaysia
Tel: +603-20956833 Fax: +603-20952322 E-mail: publish@ibbm.org.my
BJM/Ord/#134
Subscription Rates (2 Issues)
Malaysia Overseas
IBBM Member RM25.00 RM50.00 *
Non- Member RM35.00 RM110.00 *
* This rate applies to payment by credit card by
overseas subscribers. For other payment modes,
please contact IBBM for the rates.
(mm/yy)