Content uploaded by Venus Khim-Sen Liew
Author content
All content in this area was uploaded by Venus Khim-Sen Liew on Apr 21, 2016
Content may be subject to copyright.
1
Huzaimi Hussain and Venus Khim-Sen Liew. 2006. Money Demand in Malaysia:
Further Empirical Evidence. The IUP Journal of Applied Economics, V, 6, pp. 17-27.
Money Demand in Malaysia: Further Empirical Evidence
Huzaimi Hussain (UiTM Samarahan Campus) and Venus Khim-Sen Liew
*
(University
Malaysia Sabah)
Abstract
A cointegration, error correction models and CUSUM stability test are employed in this
study aimed at analyzing the money demand in Malaysia with relations to its
determinants namely real income level, real interest rate, nominal exchange rate and the
degree of monetization. Using Malaysian monthly data covering 1979M1 to 2002M5 we
are able to find cointegration relationships among M1 as well as M2 and their
determinants. CUSUM stability test shows that both M1 and M2 are stable in this sample
period albeit the Financial Innovations and Liberalization and also Asia Financial Crisis.
A key contribution of this study is the inclusion of the degree of monetization variable,
which is found playing a the significant role in the cointegrating equation. Earlier studies
that found no cointegration between money demand and its determinants in Malaysia –
rendering Bank Negara Malaysia to shift its monetary targeting to interest rate targeting –
may be due to the omission of the degree of monetization variable. Hence, this finding
suggests that M1 and M2 can be effective monetary targeting tools in implementing
monetary policy.
Corresponding author. Email: venusliew@yahoo.com.
2
MONEY DEMAND IN MALAYSIA: FURTHER EMPIRICAL EVIDENCE
Introduction
Twenty-first century has witnessed some major changes in the conduct and order
economic activities and the way business community carries out their daily transaction.
These immense evolution and development are highly attributed to the financial
liberalization and innovations, which lead to the not only borderless but also paperless
economy. Information and communication technology (ICT) revolution in the era of
globalization has also changed the way people/investors’ management and motives of
holding money; their liquidity preference. It may be the result of the change in the
structure of the economy where new financial products were introduced like repurchase
agreements (repos) and the modernization of banking system like internet and electronic
banking. It has amplified in figure and complexities, and indirectly amounted to
instabilities of velocity of money and ultimately, money demand. Stability in money
demand function with respect to its determinants is indeed imperative in determining the
effectiveness of monetary policy implemented by Bank Negara (Central Bank) of
Malaysia in achieving its targeted objectives.
Malaysia is an open and small developing country, has been applying monetary policy
side by side with fiscal policy in promoting its growth and macroeconomic stability. To
ensure the efficacy and accuracy of the monetary policy implemented, meticulous
selection must be made of what measures of monetary aggregates give the right impact
and which determinants of money demand functions are empirically influenced. It would
also mean that money supply would have some predictable impacts in stimulating the
economy. The selection of the best monetary aggregates is based on the three important
factors
1
: First, the strength of the nexus between the changes in money supply (M1, M2,
M3) and the changes in the aggregate output or income (GNP). Second, the stability of
1
The Borneo Post 19 January 2004 (Economica, pg. 21)
3
the money-output relationship. Third, the effectiveness in the use of this relationship to
forecast the economy. The main objective of this study is to identify the significant
determinants of money demand upon several measures of monetary aggregates.
Previous Studies
The analysis of money demand function stability is highly dependent on the selection of
its aggregate measures, namely M1, M2 and M3, and its determinants such as exchange
rates, interest rates, price level, income/wealth, return on equity and degree of
monetization. Ahmed (2001) reveals that in Bangladesh, interest rate is found to be
insignificant in explaining the movements of real broad money and thus should be
allowed to be determined by the free market forces for effective monetary policy. Study
by Bahmani-Oskooee et. al. (1994) which utilizes the Korean data reveals that M1
monetary aggregate is cointegrated with its determinants but not M2. The analysis using
error correction model nevertheless shows that both M1 and M2 have short run
relationship with their determinants.
In conformity with this, Amano et.al (2000) observe the long run broad money demand
in Japan; utilizing the test specifically designed for structural instability in the presence of
I(1) process unveil the stability during the period of financial innovation and
deregulation. Similarly, study Felmingham et.al. (2001) also exhibit the stability for the
long run demand for broad money in Australia under the regime shift. Kannapiran (2001)
on the other hand confirms that the short run money demand in Papua New Guinea are
real GDP, nominal interest rate and inflation rate, besides inflation rate is the best
approach for conducting monetary policy. These evidences however contrary to Andoh
et. al. (2002) which found the instability of the demand for money in Ghana due to
reformation and structural adjustment.
In case of Malaysia, Tan (1997) analyzed money demand using broad money utilizing
cointegration approach found that the money demand function was stable even under the
period of financial liberalization for the long run, but discard the short run. This has
4
indeed been the leading motivator for us to the reexamining the data since it is important
to know the timing and extent of short run intervention by the Central Bank in ironing
and forecasting future macroeconomic movement. Long run stability is of course crucial
for the formulation of macro policy. (See Huang et. al 2001). Jusoh (1987) indicates that
the demand for narrow money in moderate inflation depends on real income, a short term
nominal rate of interest and the expected rate of inflation. (Mohamed) 2001, employing
the Johansen-Juselius (1990) likelihood ratio tests reveals that exchange rate is strong
determinant of money demand in Malaysia for M2 but not M1. In other words, the study
ascertains the currency substitution of US dollar and treasury bill for monetary assets.
Ibrahim (2001) who analyses the roles of financial factors in the behaviour of M1 and M2
demand in Malaysia reveals the presence of the long-run M1 and M2 money demands
and structural instability in the dynamic specification of the M1. Another study by
Ibrahim (1998) however finds evidence suggesting that even though the long run
relationship of M2 demand is established, the short run of M2 demand is apparently
unstable.
Using cointegration and error correction mechanism techniques, Marashdeh (1997)
reveals that factors which influence M1 demand are income, expected inflation rate, the
rate of return on 6-month. mode deposit rate, expected depreciation of the exchange rate,
seasonal dummies, and the error correction from the long-run demand for money. The
study also supports the presence of currency substitution in Malaysia, thus implying that
monetary authority should take into account the effect of exchange rate on the Malaysian
economy in its formulation of domestic monetary policy.
The abovementioned studies however have given less attention to the degree of
monetization as explanatory variables and test its significance. This study is thus initiated
and considered important since firstly, it includes non-traditional explanatory variables
namely, degree of monetization and secondly the sample size spanning before and after
the Financial Innovations, Liberalization and also Asia Financial Crisis period.
5
Theoretical Framework
Money demand has been defined as the quantity of money that the nonbank public want
to hold and not financial assets (interest bearing), thus sacrificing the interest payment
that they could obtain (see Amos 1987). The analysis of money demand is tied up closely
with interest rate; it is in fact the cost of money. People nevertheless do hold money for
three motives namely transaction, precautionary and speculative. Interest rate is
negatively related with all these motives. Higher interest rate implies higher opportunity
cost of not purchasing financial assets that pay interest for instances bonds, and by
demanding/holding it. Therefore, should interest rate is relatively high; people would be
less inclined to hold money either for transaction or precautionary motives. For the
speculative motive, the logic is that it is not difficult to transfer fund from money to
financial assets, but difficult to transfer funds from financial assets to money. Should
interest rate considerably low, and forecast a hike in the near future, they would hold the
money instead of purchasing it and stuck with low return.
Evidence in the stability of money demand function is very important since if proved to
be so, policy makers should focus on interest rate instead of money supply as its target in
the implementation of monetary policy. Stability in money demand function however is
highly dependent on the volatility of interest rate and velocity of money. Innovation in
the payment mechanism in the 1980s has sped up the transaction process, which means
people not only could easily make transaction, but also in a very fast and efficient
manner. As such, people now demand less money for transaction purpose while their
income is very liquid. This alternatively implies higher velocity of circulation and lower
degree of monetization in the economy. Degree of monetization is negatively related with
velocity since it is defined as the ratio of money stock to nominal income. It implies that
the economy incline to demand more money for transaction and precautionary purposes.
Instability in interest rate would affect the degree of monetization in such a way that
higher interest rate amount to lower money demand and higher velocity of circulation.
6
Thus variability interest is considered one of the sources of instability in money demand
function, or ultimately degree of monetization.
Data and Methodology
The data employ in this study includes money demand (M1 and M2), deposit rate (proxy
of interest rate, IR), nominal ringgit per US dollar exchange rate (ER), income level
(gross domestic product, Y) and consumer price index (CPI). These daily data covering
January 1979 to May 2002 (1979M1 to 2002M5) are collected from various issues of
International Financial Statistics published by International Monetary Fund. Following
the literature all variables are employed in their logarithmic form. From these data, we
derive (1) real M1 and M2 by separately diving M1 and M2 by CPI; (2) real interest rate
by subtracting inflation rate (that is, growth rate of CPI) from deposit rate; (3) real
income level by dividing Y by CPI; and (4) degree of monetization by dividing M2 by Y.
Johansen cointegration procedure (Johansen, 1991) is employed to determine whether or
not money demands have long-run cointegration relationship with the potential
fundamentals variables identified in this study. If cointegration is found present, we may
then proceed on to the estimation of error correction mechanism (ECM) model.
Otherwise, one should carry on with vector autoregression (VAR) estimation.
As the Johansen cointegration procedure requires that all data involved are stationary,
otherwise the inference will be misleading, this study employs the augmented Dickey-
Fuller (ADF) (Dickey and Fuller, 1979, 1981) and Philips-Perron (PP) (Phillips and
Perron, 1988) unit root tests to check the stationarity of the data. As the ADF test, PP test,
Johansen cointegration procedure, ECM and VAR are already well-known by now;
further description is omitted in this study.
As stable money demand is effective in the policy implications, we also check the
stability of our estimated models using the CUSUM of squares stability test. Briefly, the
CUSUM of squares test (Brown, Durbin, and Evans, 1975) is based on the cumulative
7
sum of the recursive squared residuals. The CUSUM of squares test provides a plot the
values the cumulative sum together with the 5 percent critical lines against time. The test
finds parameter instability if the cumulative sum of the recursive squared residuals goes
outside the area between the two critical lines.
Results and Study
The unit root tests results are summarized in Table 1. Table 1 show that all variables are
non-stationary at conventional significance level, except the real interest rate and real
income level variables. The former has been shown mean and trend stationary by the PP
test, and the latter is trend stationary based on both ADF and PP tests. On the other hand,
all variables are first difference stationary, implying that our analysis should be
conducted in the first difference series instead of in their levels.
Table 1: Unit Roots Test
Variable
Levels
First Differences
ADF Test
PP Test
ADF Test
PP Test
u
t
t
u
t
t
u
t
t
u
t
t
M1
-0.576
-1.516
-0.532
-1.710
-7.447
I
-7.434
I
-17.213
I
-17.181
I
M2
-0.522
-1.144
-0.758
-1.469
-7.397
I
-7.395
I
-16.570
I
-16.558
I
IR
-2.467
-2.703
-4.553
I
-4.688
I
-7.499
I
-7.524
I
-27.254
I
-27.332
I
ER
-0.683
-2.021
-0.461
-2.018
-7.282
I
-7.298
I
-14.502
I
-14.494
I
Y
-0.417
-3.229
X
-0.735
-5.427
I
-9.213
I
-9.198
I
-31.995
I
-31.934
I
DM
-1.503
-1.144
-0.758
-1.469
-7.397
I
-7.395
I
-16.570
I
-16.558
I
Critical Values
1%
-3.54
-3.99
-3.54
-3.99
-3.54
-3.99
-3.54
-3.99
5%
-2.87
-3.43
-2.87
-3.43
-2.87
-3.43
-2.87
-3.43
10%
-2.57
-3.14
-2.57
-3.14
-2.57
-3.14
-2.57
-3.14
Note: Superscripts (
X
) and (
I
) denotes statistical significant at 10% and 1% level respectively.
8
The Johansen cointegration test results are tabulated in Table 2. Table 2 shows that M1 is
cointegrated with its potential determinants by the trace test with two cointegrating
equations, but not the maximum eigenvalue test. Since at least one statistics has shown
cointegration relationship, this study will proceed to estimate the error correction
mechanism (ECM) model for M1. As for M2, both trace test and maximum eigenvalue
test suggest that it is cointegrated with its potential determinants, although minor
disagreement occurs in the number of cointegraiton equations. Similar to M1, as
cointegration relationship are found, ECM model will also be estimated for M2 demand
function.
Table 2: Cointegration Rank Test M1
Hypothesized
No. of CE(s)
Eigenvalue
Trace Test
Maximum Eigenvalue Test
Statistic
Critical Values
Statistic
Critical Values
5%
1%
5%
1%
Variables: M1, interest rate, exchange rate, income, degree of monetization
None
0.104
82.605
I
68.52
76.07
30.310
33.46
38.77
At most 1
0.080
52.295
V
47.21
54.46
22.935
27.07
32.24
At most 2
0.057
29.360
29.68
35.65
16.317
20.97
25.52
At most 3
0.045
13.043
15.41
20.04
12.822
14.07
18.63
At most 4
0.001
0.220
3.76
6.65
0.220
3.76
6.65
Variables: M2, interest rate, exchange rate, income, degree of monetization
None
0.119
90.347
I
68.52
76.07
35.017
V
33.46
38.77
At most 1
0.091
55.330
I
47.21
54.46
26.389
27.07
32.24
At most 2
0.056
28.940
29.68
35.65
15.835
20.97
25.52
At most 3
0.046
13.105
15.41
20.04
12.999
14.07
18.63
At most 4
0.000
a
0.106
3.76
6.65
0.106
3.76
6.65
Notes: The VAR lag length is determined to be 4 by the Akaike Information Criterion
(AIC). Superscripts (
I
)
and (
V
) denote statistical significant at 1% and 5% level
respectively. Superscript (
a
) denotes extremely small value.
9
The estimated money demand functions for M1 and M2 based on the ECM model are
summarized in Table 4. The estimated money demand functions are deemed adequate as
their residuals have good properties as required by the classical linear regression
assumptions such as no autocorrelation (by the Breusch-Godfrey LM test, see Breusch,
1978 and Godfrey, 1978), no heteroscedasticity (White heterocsedasticity test, see White,
1980), no misspecification (Ramsey’s RESET test, see Ramsey, 1969). Moreover, the
Jarque-Bera normality test (Jarque and Bera, 1987) shows that the residuals are normality
distributed. To sum, the residuals diagnostics suggest that the normal interpretation of
least squares regression model is reliable in our estimated models and the inferences are
therefore trustworthy.
One important finding from Table 4 is that both the error correction terms (ECT) for M1
and M2 demand function models are significant at conventional significance level. This
implies that while money demand may temporary deviate from its long-run equilibrium,
the deviations are adjusting towards the equilibrium level in the long-run. From the
estimated coefficients of ECT for M1 and M2, we know that 6.5% and 6.4% of the short-
run deviations will be adjusted each month (amounting to an annual total of nearly 80%)
towards the equilibrium levels of M1 and M2 respectively. This finding reinforces the
inference of Johansen cointegration relationship between M1 and M2 with their
respective fundamentals.
10
Table 3: Estimated Money Demand Functions
D(M1)
D(M2)
Variable
Coefficient
Std. Error
p-value
Coefficient
Std. Error
p-value
D(M1(-1))
-0.073
0.065
0.261
--
--
--
D(M1(-2))
0.000
0.064
0.998
--
--
--
D(M1(-3))
0.010
0.064
0.875
--
--
--
D(M1(-4))
0.058
0.064
0.366
--
--
--
D(M2(-1))
--
--
--
0.084
0.107
0.435
D(M2(-2))
--
--
--
0.089
0.111
0.426
D(M2(-3))
--
--
--
0.117
0.109
0.285
D(M2(-4))
--
--
--
-0.137
0.111
0.220
D(Y)
0.003
0.106
0.980
0.106
0.092
0.250
D(Y(-1))
0.184
0.107
0.087
0.052
0.094
0.585
D(Y(-2))
-0.014
0.107
0.897
0.060
0.092
0.514
D(Y(-3))
0.216
0.105
0.042
0.195
0.091
0.033
D(Y(-4))
0.249
0.104
0.017
0.108
0.056
0.056
D(IR)
-0.001
0.001
0.070
0.000
0.000
0.638
D(IR(-1))
-0.001
0.001
0.206
0.000
0.000
0.861
D(IR(-2))
-0.001
0.001
0.111
0.000
0.000
0.564
D(IR(-3))
0.000
0.001
0.579
0.000
0.000
0.630
D(IR(-4))
0.001
0.001
0.314
0.000
0.000
0.791
D(ER)
0.038
0.012
0.002
-0.007
0.007
0.312
D(ER(-1))
-0.014
0.013
0.262
-0.003
0.007
0.638
D(ER(-2))
-0.021
0.013
0.102
0.002
0.007
0.730
D(ER(-3))
-0.014
0.013
0.291
-0.006
0.007
0.416
D(ER(-4))
-0.018
0.013
0.158
-0.004
0.007
0.515
D(DM)
0.012
0.102
0.904
0.069
0.092
0.451
D(DM(-1))
0.195
0.101
0.055
0.002
0.095
0.981
D(DM(-2))
-0.067
0.102
0.511
0.021
0.094
0.823
D(DM(-3))
0.136
0.100
0.172
0.211
0.092
0.022
D(DM(-4))
0.208
0.099
0.037
0.093
0.053
0.082
ECTM1
-0.065
0.022
0.004
--
--
--
ECTM2
--
--
--
-0.064
0.038
0.090
Diagnostic
2
R
0.209
0.001
2
R
0.134
-0.094
AIC
-5.578
-6.840
SC
-5.250
-6.512
DW d
2.016
2.028
AR [4]
1.206 [0.309]
1.789 [0.130]
ARCH [4]
0.599 [0.664]
0.293 [0.883]
HETERO
0.818 [0.801]
0.956 [0.661]
RESET [4]
1.290 [0.275]
0.293 [0.883]
JB
4.468 [0.107]
0.437 [0.112]
Notes: AR[4] and ARCH [4] is the Lagrange Multiplier test of 4
th
order autoregression and ARCH effects
respectively. HETERO, RESET and JB are correspondingly White Heteroscedasticity test, Ramsey RESET
specification and Jarqur-Bera Normality test.
11
Another interesting insight depicted in Table 3 is that both M1 and M2 seem exogenous
as their lagged values are not contributive at all in determining they respectively present
values. Moreover, by the p-value of the t-test of individual significance (variable with p-
value less than 0.10 is considered statistically significant), we find that changes in all
independent variables (or their lagged values) have short-run impact on M1 demand.
Surprisingly, only lagged values of real income and degree of monetization has
significant short-run influence on the M2 demand. The last two findings suggest that
policy makers may predict or manage, first, the M1 money demand via monitoring the
real and nominal gross domestic product (the latter is an element of degree of
monetization), nominal exchange rate and deposit rate, and second, the M2 money
demand through real and nominal gross domestic product only.
By appropriately conducting Wald test of restrictions on the estimated ECM models, one
may identify the overall short-run effect of a variable the money demand variable. If the
first difference variable and its lagged values in sum has no significant effect (by the F
test of overall significance in the Wald test framework) on the M1 or M2 (in first
difference), than one may conclude that the variable does not Granger cause M1 or M2 in
the short-run. Thus, the above test is effectively a short-run causality test. The results of
this test are summarized in Table 4.
12
Table 4: Short-run Causality Tests
Null Hypothesis
F statistic of Wald Test [ p-value]
D(M1)
D(M2)
1
4
))(1(
i
iMD
0.498 [0.737]
--
1
4
))(2(
i
iMD
--
0.948 [0.437]
0
4
))((
i
iYD
3.018 [0.012]
1.655 [0.146]
0
4
))((
i
iIRD
1.318 [0.259]
0.164 [0.976]
0
4
))((
i
iERD
3.332 [0.006]
0.573 [0.720]
0
4
))((
i
iDMD
2.553 [0.028]
1.690 [0.158]
Table 4 shows that M1 is Granger caused by real income (Y), ringgit-US dollar exchange
rate (ER) and degree of monetization (DM), whereas no variable is found to Granger
cause M2 in the short-run. We argue here that one need not be pessimistic in this regard
as the overall test of restrictions may mask the individually influence of a lagged
dependent variable on the M2. Recall from Table 3 that the coefficients of D(Y(-3)) and
D(Y(-4)) are separately significant by the t-test at 5% and 10% respectively. Actually, the
combined effect of these two lagged dependent variables is significant at 5% level by the
same F test (F value=3.724, p-value=0.026). Meanwhile, D(DM(-2)) and D(DM(-3)) are
significant at 5% and 10% respectively by the t-test and they are jointly significantly at
5% level by the F test (F value=3.857, p-value=0.022]. Disregarding this fact, if one were
to choose between M1 and M2 as monetary policy target variable, we would suggest M1
13
variable because the impact of its determinants is more substantial by the short-run
causality test.
Finally, CUSUM stability test results for M1 and M2 are accordingly plotted in Figure 1
and Figure 2. Figure 1 shows that the cumulative sum of the squared residuals of the
estimated ECM model as depicted in Table 3 is inside the 5% critical lines. This implies
that the estimated M1 money demand function is stable within the period of study, even
though this period seems to have contained structural reforms such as the early 80s’
Financial Innovations and Liberalization and the 1997 Asian Financial Crisis. Similarly,
our estimated M2 demand function is also stable in this sample period of study (Figure
2). We note here that early studies (see Ibrahim, 2001, for example) which reported
structural instability in money demand function in the Malaysia context may be due to the
exclusion of degree of monetization variable, a variable determined to be relevant to
Malaysian money demand models in this current study. Our study has an advantage of
simplicity but reliable over others in the sense that we do not need to resort to more
complicated econometrics procedures -- such as the explicit inclusion of dummy
variables or structural breaks -- in searching for stable money demand function.
14
Figure 1: CUSUM stability Test for M1 ECM model
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
82 84 86 88 90 92 94 96 98 00
CUSUM of Squares 5% Significance
15
Figure 2: CUSUM stability Test for M2
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
82 84 86 88 90 92 94 96 98 00
CUSUM of Squares 5% Significance
Conclusions and Policy Suggestions
This study investigates the money demand functions in the context of Malaysia with the
inclusion of the degree of monetization variable, which has been neglected in previous
related studies. Our results, among others find long-run cointegration relationship
between money demand as measured in M1 and M2 and the set of fundamental variables
including real income level, real interest rate, nominal ringgit-US dollar exchange rate
and degree of monetization. Error correction mechanism models have been estimated for
M1 and M2 demand functions. One cheerful finding regarding these models is that both
16
estimated M1 and M2 money demand functions are stable in the period of study, which
has undergone various structural reforms such as in the early 80s’ Financial Innovations
and Liberalization and the late 90s’ Asian Financial Crisis.
A key contribution of this study is the inclusion of the degree of monetization variable,
which is found playing a significant role in the cointegrating equation. Earlier studies that
found no stable relationship between money demand and its determinants in Malaysia --
rendering Bank Negara Malaysia to shift its monetary targeting to interest rate targeting -
- may be due to the omission of the degree of monetization variable. Hence, this finding
suggests that M1 and M2 can be effective monetary targeting tools in implementing
monetary policy. The policy makers must also be conscious of the movement and level of
interest rate since the degree of monetization variable has a positive effect on money
demand. They must ensure that money supply is at the appropriate level, as increases in
the degree of monetization would amount to increases in money demand. (Ahmed 2001)
It is noteworthy that if policy makers were to choose between M1 and M2 for monetary
policy targeting tool, we would suggest the use of M1 for the reason that behaviour of
M1 may be forecastable or monitored via all the independent variables under study,
whereas the movement of M2 is not traceable from these variables at all at least in the
short-run.
17
Acknowledgement: The authors would like to thank Ching-Hong Puah and the
participants of the First International Borneo Business Conference (IBBC) 2004 for their
helpful comments and suggestions. The remaining errors are no doubts ours.
References
Ahmed M. 2001. Demand for money in Bangladesh: An econometric investigation into
some basic issues. The Indian Economic Journal, 48, 1.
Amano RA, Wirjanto TS. 2000. On the stability of long-run M2 demand in Japan. The
Japanese Economic Review, 51, 4.
Amos OM. 1987. Economics Concept, Analysis, and Application. Wadworth Publishing
Company Belmont, California.
Bahami-Oskooee M and Rhee H. 1994. Long-run elacticities of the demand for money in
Korea:Evidence for cointegration analysis. International Economic Journal, 8, 2.
Breusch TS. 1978. Testing for autoregression in dynamic linear models. Australian
Economic Papers, 17, 334 – 355.
Brown RL, Durbin J and Evans JM. 1975. Techniques for testing the constancy of
regression relationships over time. Journal of the Royal Statistical Society, 37, 149-192.
Dickey DA and Fuller WA. 1979. Distribution for the estimation for auto-regressive time
series with unit root. Journal of American Statistical Association, 74 – 427 – 431.
Felmingham B and Zhang Q. 2001. The long-run demand for broad money in Australia
subject to regime shifts. Australian Economic Papers, June, 2001
Godfrey LG. 1978. Testing for higher order serial order correlation in regression
equations when the regressors include lagged dependent variables. Econometrica, 46,
1303 – 1310.
Handa J. 2000. Monetary Economics. Routledge: Taylor & Francis Group
Huang CJ, Lin CJ and Cheng J. 2001. Evidence on nonlinear error correction in money
demand: the case of Taiwan. Applied Economics, 33, 1727-1736.
18
Ibrahim MH. 1998. An econometric analysis of money demand and its stability in the
Malaysian economy. Indian Economic Review, 33, 1. 53-66
Ibrahim MH. 2001. Financial factors and the empirical behaviour of money demand: A
case study of Malaysia. International Economic Journal, 15, 55 – 72.
Johansen S. 1991. Estimation of hypothesis testing of cointegration vectors in Gaussian
vector autoregressive models. Econometrica, 58, 165 – 188.
Jusoh M. 1987. Inflationary expectations and the demand for money in moderate
inflation: Malaysian evidence. Jurnal Ekonomi Malaysia 15, 3-14.
Marashdeh O. 1997. The demand for money in an open economy: the case of Malaysia.
Southern Finance Association Annual Meeting, 19-22 November 1997, Baltimore,
Maryland, USA.
Phillip PCB. and Perron P. 1988. Testing for a unit root in time series regressions.
Biometrika, 65, 335 – 346.
Kannapiran CA. 2001. Stability of money demand and monetary policy in Papua New
Guinea (PNG): An error Correction Model Analysis. International Economic Journal,
15, 3.
Ramsey JB. 1969. Tests for specification errors in classical linear least squares regression
analysis. Journal of Royal Statistical Society, 31, 350 – 371.
Tan EC. 1997. Money demand amid financial sector developments in Malaysia. Applied
Economimcs, 29, 1201-1215.
White H. 1980. A heteroscedasticity consistent covariance matrix estimator and a direct
test of heteroscedasticity. Econometrica, 48, 817 – 818.
