Koong Seow Shin - Faculty of Business and Economics - Monash ...

Goh Soo Khoon * , Tuck Cheong Tang †
and Koong Seow Shin ‡
Asian Business and Economics Research Unit
Discussion Paper 56, 2008
ABSTRACT
This paper re-examines the degree of capital mobility for two small open economies in Southeast
Asia, namely, Malaysia and Singapore, by applying the model proposed by Shibata and Shintani
(1998) and the extension model by Cooray (2005). Three empirical contributions are derived from
the paper. First, the empirical evidences inform that capital seems to be mobile in Malaysia for the
sample period 1960-2004. But, Singapore is an exemption case. Second, the paper finds that the
interest rate differential is not related to changes in consumption in both economies. Third, this
paper also informs the importance of incorporating relevant instrumental variables in any GMM
estimations.
Keywords: Capital Mobility; GMM; Interest rate differential; Malaysia; Singapore
JEL Classification codes: F21; F32
NEW EMPIRICAL EVIDENCE ON CAPITAL
MOBILITY IN TWO SOUTHEAST ASIAN
ECONOMIES: MALAYSIA AND SINGAPORE
*
School of Business and Administration, Wawasan Open University, Penang, 10050, Malaysia. Email:
skgoh@wou.edu.my
†
Corresponding author.School of Business, Monash University Sunway Campus, Jalan Lagoon Selatan,
46150 Bandar Sunway, Selangor Darul Ehsan, Malaysia. Email: tang.tuck.cheong@buseco.monash.edu.my
‡
Graduate student, Faculty of Economics and Management, University Putra Malaysia, Serdang, 43400,
Malaysia. Email:seowshin.koong@yahoo.com
© 2008 Goh Soo Khoon, Tuck Cheong Tang and Koong Seow Shin
All rights reserved. No part of this paper may be reproduced in any form, or stored in a retrieval system,
without the prior written permission of the author.

INTRODUCTION
Using the approach developed by Shibata and Shintani (1998) and the extension by Cooray
(2005), this paper aims to reaffirm the international capital mobility in two small open economies in
Southeast Asia viz, Malaysia and Singapore. This is a more recent approach in assessing capital
mobility by focusing on the correlation between the changes in a country’s consumption and the
changes in its net output. 1 They proposed that if capital mobility is perfect, then, consumption
changes are independent of net output changes. The foundation of Shibata and Shantani (1998)’s
model is from the permanent income model of Campbell and Mankiw (1989, 1990, 1991) with a
version of an intertemporal current account.
This paper is driven by the following contributions. Firstly, a literature search informs that limited
studies have been undertaken on this model, and most of the results, in general, suggest capital
immobility. Shibata and Shintani’s (1998) study has found im-perfect capital mobility for 11 OECD
countries including the United States and Japan. In recent, Cooray (2005) had extended their
framework by further considering the effect of interest rate differential on consumption variable.
Cooray (2005) informed that if the domestic interest rate is higher than the foreign interest rate,
there will be a capital inflow, a currency appreciation and a fall in domestic consumption, and vice
versa. Using the augmented specifications of Shibata and Shintani (1998), Cooray (2005)
examined the degree of capital mobility in four South Asian countries, i.e., India, Sri Lanka,
Pakistan and Bangladesh. The empirical results inform that, in general, capital is immobile in these
countries. In fact, an empirical examination of the degree of capital mobility in the Southeast Asian
region is an important consideration. By and large, our literature survey revealed that this model
has not been applied to many small open economies, particularly in Southeast Asia such as
Malaysia and Singapore. In this relation, Malaysia and Singapore are ideal sample countries to
examine issues concerning the degree of capital mobility. Malaysia has had periods of both
liberalized and controlled capital flows since the country liberalized its capital account in the 1970s
(Goh, 2007) whereas Singapore has maintained an open capital account, in which there is virtually
no restriction on movement of portfolio capital flows since the liberalization of its capital account in
the 1970s. Hence, from an economic stand point, it is interesting to evaluate the extent to which a
liberal exchange control as opposed to a liberalized and controlled capital flows has led to
substantial integration of Malaysia’s and Singapore’s financial market with the rest of the world.
Secondly, empirical studies have often failed to provide rigorous statistical evidence about the
degree of capital mobility for developing countries such as Malaysia and Singapore. Broadly
speaking, the statistical tests to capital mobility framework involving nominal interest rate
comparisons, consumption correlation and consumption smoothing models generally indicate a
high degree of capital mobility in Malaysia (De Brouwer, 1999; Ghosh & Ostry, 1995; Goh, 2007;
Goh et al., 2006). Nevertheless, those involving saving-investment frameworks demonstrate
relatively low levels of capital mobility in Malaysia (Mamingi, 1997; Bagnai and Manzocchi, 1996).
On the other hand, applying various approaches to capital mobility, MAS (Monetary Authority of
Singapore; 2000) 2 found capital flow in Singapore to be mobile, while the saving-investment
framework shows that capital is immobile in Singapore (Anoruo, 2001; Tan, 2000).
Thirdly, this paper considers the significance of ignoring a set of relevant instrumental variables in
order to ensure the robustness of statistical properties of GMM (Generalized Method of
Movements) estimator. A set of relevant instrumental variables should be correlated with the
candidate endogenous variables, and this condition is referred to as instrument relevance in GMM
estimator. Again, this issue has not been addressed, in which, no single study has empirically
tested the instrument relevance prior to the GMM estimations. Thus, this paper fills this empirical
gap of literature in GMM estimations.
This paper is organized as follows. Section II briefly demonstrates the theoretical frameworks
proposed by Shibata and Shintani (1998) and extended by Cooray (2005). Section III describes the
data (variables) employed in this paper. The empirical results on data stationary and models
estimates are reported in Section IV. Section V concludes the paper.

THEORETICAL FRAMEWORK
The approach to capital mobility proposed by Shibata and Shintani (1998) is principally derived
from a basic model with a small open economy version of the permanent income model with
imperfect international capital mobility. Assuming a world interest rate of i, and a country’s limit
budget is given by equation (1).
Ft+ 1 (1 i) Ft GDPt Ct It Gt
= + + − − −
Ft+ 1 (1 i) Ft Vt Ct
= + + − (1)
Vt ≡GDPt −Gt − It
where t F is foreign asset holdings; GDP t is gross domestic product; t C is private consumption; t I
is investment; t G is government consumption; and V t is a country’s net output. From the national
income accounting identity, an equation for current account balance, CA can be read as:
CAt ≡ iFt + Vt − Ct
(2)
where CAt is the current account balance.
Constructively, there are two polar cases - perfect international capital mobility and financial
autarky. If we assume that the utility function is quadratic, and the consumers’ discount rate and
the world interest rate are equal, optimal consumption in the case of perfect capital mobility can be
written as:
∞
* ⎧ 1 1 n ⎫
Ct = i⎨Ft + ( ) ∑ ( ) EV t t+ n⎬
(3)
⎩ 1+ i n=
0 1+
i ⎭
where EV t t+ nrepresents
the future expectation of net output. By differentiating equation (3), we get
the following consumption equation.
∞
* 1 1 n
Δ Ct = ∑ ( ) ( Et −Et−
1)
Vt+
n
(4)
1+ i n=
0 1+
i
where ( Et − Et− 1)
Vt+
n is the change of expectations from time t-1 to t. However, the value at time t-1
is unpredictable under rational expectations. Thus, equation (4) can be expressed in this form:
*
Δ Ct = et
(5)
where e t is a rational forecast error, which is orthogonal to the information available at time t-1.
Again, under perfect capital mobility, the current account equation can be derived from equations
(1) to (3). Then, the optimal current account equation is informed by equation (6):
∞ i 1 n
CAt =−( ) ∑ ( ) Et( Vt+ n −Vt)
1+ i 1+
i
n=
0
∞ 1 n
CAt =−∑ ( ) Et( ΔVt+
n)
(6)
n=
1 1+
i
Equation (6) demonstrates that if Δ V follows a stationary process, I(0), then the current account
balance follows a stationary process, I(0) as well, under perfect capital mobility. It is worth
assuming that the current account and consumption are determined by future expectations of net
output. However, in the case of financial autarky, there are no capital movements between
countries at all. Therefore, a country’s consumption is constrained by its current net output:
a
Ct = Vt
(7)
And, the trade account balance equality is now read as:
TB ≡CA − iF
(8)
t t t

where TB t is a trade balance that always equals zero, i.e., S n = I n (where S n is national saving and
I n is national investment).
Aggregate consumption relation in the case between the two polar cases where capital is mobile
but not perfectly mobile across countries is given by:
* a
*
Ct = (1 − δ ) Ct + δCt = (1 − δ) Ct + δVt
(9)
whereδ measures the degree of international capital mobility. The estimated value of δ is ranged
between zero and unity. The smaller (larger) the estimated value ofδ , the higher (lower) the
degree of international capital mobility of a country is. If the value of δ is zero (unity), it indicates
*
perfect capital mobility (perfect capital immobile). Note that, in equation (9), C t is dependent on
unobservable terms, which is EV t t+ n (n = 1, 2, …, n). By differentiating both sides of Equation (9),
we obtain:
* a
Δ Ct = (1 −δ) Δ Ct + δΔ Ct = (1 − δ) et + δΔ
Vt
(10)
where Δ Ct
describes the changes in aggregate consumption. Shibata and Shintani (1998)
examined the degree of international capital mobility by estimating the parameter ofδ in equation
(10).
In a recent study, Cooray (2005) has extended Shibata and Shintani (1998)’s model i.e. equation
(10) by considering the interest rate differential variable. This would imply that if the domestic
interest rate is higher than the foreign interest rate, there will be a capital inflow, a currency
appreciation and a loss of competitiveness in the international market that leads to a fall in
aggregate demand and therefore consumption or vice versa. The augmented model is informed by
equation (11).
Δ Ct = (1 − δ )[ et + ( i− if )] + δΔ
Vt
(11)
where i is domestic interest rate, and i f is world interest rate. If the interest rate differential is
statistically significant (insignificant) from zero, it implies that the interest rate differential appears
(does not appear) to affect the changes in consumption. Note that δ measures the degree of
international capital mobility.
VARIABLES AND DATA
The sample period covers annual data from 1960 to 2004. The official data for the sampled
countries (Malaysia and Singapore) are obtained from the International Financial Statistics
(International Monetary Fund). The real net output variable ( V t )is constructed by subtracting the
government consumption expenditure, gross fixed capital formation and the changes in inventories
from gross domestic product (GDP) and deflating it by the Consumer Price Index (CPI). The
differences between the domestic interest rate and foreign interest rate ( i− if)
are constructed by
subtracting the world interest rate from the domestic interest rate, i ; the money market rates are
used for these two countries and the United State Federal Funds Rate is used to proxy the world
interest rate, i f .
EMPIRICAL RESULTS
Prior to the OLS (Ordinary Least Square) and GMM estimation of equations (10) and (11), we need
to examine the stationarity (or the degree of integration), I(d) of the variables of interest i.e., the
change in consumption ( Δ C ) and change in net output ( Δ V ) as well as the interest rate
differential (i-if). The results of both Augmented Dickey Fuller (ADF) and Phillips Perron (PP) test

are documented in Table 1, showing that all series are stationary in levels, or in I(0) process,
except for the interest rate differential (i-if) in Malaysia which is non-stationary or integrated with
order one, I(1). This result is in line with the study by Goh et al., (2006) who found that the i-if
variable is non-stationary or in I(1) process in the case of Malaysia.
Table 1: Unit Root Test
Country
ΔC
ADF PP
ΔV
ADF PP
i-if
ADF PP
∆( i-if )
ADF PP
Malaysia -4.63*** -4.52*** -4.09*** -4.07*** -2.876 -2.28 -4.99*** -4.61***
Singapore -4.43*** -4.32*** -3.39** -3.37** -3.38*** -3.26**
x = β + β x
T
Δ x = β + β x + ∑ β Δ x + ε . The PP test is based on the
t 0 1 t−1 2 t−i t
i = 1
+ ε . The MacKinnon (1991) t-critical values for the ADF tests for the sample
Notes: The ADF test is based on the following model,
following model, t 0 1 t−1 t
size of 50 with a constant are 1% = -3.58, 5% = -2.93 and 10% = -2.60, with a constant and trend are 1% = -
4.15, 5% = -3.50 and 10% = -3.18. *, **, and *** denotes statistical sig`nificance at the 10%, 5% and 1%
levels respectively. Only significant lags are included in the ADF and PP test.
Equations (10) and (11) were estimated by OLS and GMM estimators. According to Shibata and
Shintani (1998), the error term may be correlated with ΔV since higher output usually implies good
news for the country’s expected future net output. Hence, the GMM estimation method is employed
in this paper in order to discard any potential inconsistencies in the OLS estimates.
By the same token, it is relevant to incorporate a set of instrumental variables for GMM estimation
exercise. Again, the instrumental variables must satisfy two requirements: First, they must be
orthogonal to the error process, which is also known as “instrument exogeneity”; and, second, they
must be correlated with the included endogenous variables, more formally known as “instrument
relevance”. The formal condition of instrument exogeneity is readily tested by the J-statistics. 3
It must be noted though that the issue of relevant instrumental variables has been the subject of
much recent research in time series econometrics. If the instrumental variables are weak (i.e. they
are weakly correlated or uncorrelated with the corresponding endogenous variables), the statistical
properties of GMM can be inconsistent, that is, the coefficient estimates are biased and incorrect
statistical inferential may have occurred. A general rule of thumb for identifying the relevance
instrument variables is to estimate an OLS regression of the endogenous variable(s) on the full set
of instrumental variables, and compute the F-statistics of the excluded instruments in the
regression. 4 If the F-statistic is statistically different from zero at a conventional level of significance
(i.e., 10%), the candidate instrumental variables are weak instrumental variables. Consequently,
the GMM estimator is not a reliable estimator. In this paper, both instrument relevance and
instrument exogeneity were checked for their suitability before we determined the robustness of
the GMM estimates.
Most of the existing studies [Shibata and Shintani (1998), Bayoumi and MacDonald (1995), and
Cooray (2005)] employed the lagged endogenous variables (Xt-i) as instrumental variables. Hence,
this paper performed a series of statistical tests, and found that the lagged endogenous variables
are weak instrumental variables for sample countries viz Malaysia and Singapore. 5 In this context,
this paper considered other set of instrumental variables. The variables include the openness of
the US, the growth rate of real GDP per capita in the US, and the US consumption share of real
GDP. The justification for using these instrumental variables is intuitively informed by the fact that if
the US is a major trading partner for all sample countries, then changes or fluctuations in the
openness, growth rate of real GDP per capita and consumption in the US may affect the level of
GDP and hence consumption for small open economies such as the sampled countries – Malaysia
and Singapore.

Table 2 reports the OLS and GMM estimates for equation (10). As informed by the OLS
estimation, the estimated coefficient on net output variable is ‘inelastic’ for both countries. The tstatistics
fail to reject the null hypothesis of perfect capital mobility (i.e.δ =0) for Malaysia,
suggesting capital mobility. However, for Singapore, the null hypothesis of perfect capital mobility
can be rejected even at the 10% level of significance and the point δ estimates are not large,
0.26. The last two columns of Table 2 show the GMM estimates of δ . It is found that the high Fstatistics
reveal that the included set of instrumental variables are relevant for both countries.
Again, the J-statistics also statistically confirm that the null hypothesis of the instruments variables
are exogenous is not rejected for both countries. In general, the GMM estimations inform that the
δ estimates remain ‘small’ for Malaysia, but ‘big’ for Singapore, that is, about 0.43 although the
null hypothesis of perfect capital mobility is rejected for both sets of instruments for Singapore.
Again, the GMM estimates suggest perfect capital mobility for Malaysia. Overall, this initial analysis
indicates that while capital flow is relatively mobile in Malaysia, there are capital restrictions in
Singapore.
Table 2: Consumption-net output correlations (Equation 10)
Country
Sample
Period
OLS
estimates GMM estimates
IV(1) IV(2)
0.01
(0.11)
0.10
(0.74)
0.072
(0.50)
Malaysia 1961-2004 δ
R 2 0.0003 -0.1068 -0.077
J-statistics
3.6714 3.4734
(p-value) - (0.1594) (0.1023)
F-statistics 11.8467 10.9311
0.26 0.425 0.426
Singapore 1961-2004 δ
(3.62) (4.72) (4.72)
R 2 0.238 0.1362 0.1364
J-statistics
1.0668 1.0366
(p-value) - (0.5865) (0.3085)
F-statistics 19.292 28.196
Notes: t-statistics are reported in parentheses. Equation (10): Δ Ct = (1 − δ ) et + δΔ
Vt.
The sets of
instrumental variables (IV) used in the GMM estimation are IV (1) {US’s openness, US’s growth rate of real
GDP per capita and US consumption share of real GDP}. IV (2) {US’s growth rate of real GDP per capita and
US consumption share of real GDP}. The J statistics reported in Eviews are multiplied by n. The p-values of
the J statistics are reported in parentheses, (.).
Turning to the extended framework by Cooray (2005), this paper further estimated equation (11).
The empirical estimates are reported in Table 3. The OLS estimates (fourth column) are found to
be consistent with the estimates of equation (10) from Table 2. The estimated parameter of net
output remains small for both countries. In fact, this finding is not a surprise given the fact that the
interest rate differential variable is statistically insignificant from zero for both sample countries: this
differential (interest rate) is not associated with the growth rate of consumption. 6 For the GMM
estimates, the F-statistics show that both sets of instrumental variables are relevant for Singapore
but not for Malaysia. 7 The GMM estimator informs that the δ estimate remains small for Singapore
although the null hypothesis of perfect capital mobility is rejected for IV(1) and IV(2) specifications.
Generally speaking, the OLS and the GMM estimates both suggest that there is evidence against
perfect capital mobility in Singapore.

Table 3: Consumption-net output correlations with interest rate differential (Equation 11)
Country Sample Period OLS estimates GMM estimates
IV(1) IV(2)
Malaysia 1971-2004
δ
-0.06
-0.11 0.43
γ
(-0.55)
-0.43
(-1.04)
(-0.83)
-0.31
(-1.35)
(3.84)
0.19
(0.94)
R 2 0.04 - -
3.46 3.926
J-statistics (p-value) -
(0.3257) (0.140)
F-statistics 6.93 6.77
Singapore 1972-2004
δ
0.25
0.552 0.5568
γ
(2.81)
-0.272
(4.138)
-0.07
(4.22)
-0.106
(-1.26) (-0.538) (-0.643)
R 2 0.22 - -
J-statistics
2.222 2.147
(p-value) -
(0.5275) (0.3417)
F-statistics 10.347 15.633
Notes: t-statistics are reported in parentheses. Equation (11): Δ C = (1 − δ )[ e + ( i− i )] + δΔ
V .
t t f t
The sets of instrumental variables (IV) used in the GMM estimation are IV (1) {US’s openness, US’s
growth rate of real GDP per capita and US consumption share of real GDP, interest rate differential}.
IV (2) {US’s growth rate of real GDP per capita and US consumption share of real GDP, interest rate
differential}. The J statistics reported in Eviews are multiplied by n. The p-values of the J statistics
are reported in parentheses, (.).
By and large, the finding of capital immobility for Singapore may come as a surprise even though it
is consistent with Anoruo’s ( 2001) and Tan’s (2000) studies. In general, many countries in
Southeast Asia including Singapore and Malaysia have taken several measures in deregulating
and liberalizing their financial markets. Among them, Singapore was the first to begin liberalizing
her financial market by removing and lifting interest rate regulations and abolishing exchange
controls in the mid 1970s. And, Malaysia followed soon after, taking major steps towards
deregulating the financial markets in the late 1970s. Although financial liberalization was
undertaken at a “stop-go-stop” pace consistent with Malaysian economic needs and major
economic global events that occurred at that time, 8 our results reinforce the recent findings in the
literature (De Brouwer, 1999; Ghosh & Ostry, 1995; Goh, 2007; Goh et al., 2006) that Malaysia has
exhibited a substantial amount of, at least, de facto financial openness despite the periodic
exchange controls.
CONCLUDING REMARKS
Using the most recent measure of international capital mobility by Shibata and Shintani (1998) and
the extension model by Cooray (2005), this paper re-visits the degree of capital mobility for two
small open economies in Southeast Asia, namely, Malaysia and Singapore. Nonetheless, these
approaches focus on the correlation between consumption and net output rather than on the
conventionally applied saving-investment framework in the literature. This paper offers two findings
to be added to the existing empirical literature in international open macro-economics. Firstly, this
paper finds perfect international capital mobility for Malaysia but capital immobility for Singapore.
Secondly, it is necessary to identify a relevant set of instrumental variables in the GMM
estimations.

ACKNOWLEDGEMENT
The authors have benefited from comments made by participants at the International Economic
Conference on Trade and Industry, 3 rd -5 th December, 2007, Malaysia and from Associate
Professor Dr David Harris, University of Melbourne, Australia. Any remaining errors or omissions
are the responsibility of the authors.

REFERENCES
Anoruo, E. (2001) "Saving-investment connection: evidence from the ASEAN countries" American
Economist, 45(1), 46-53.
Bagnai, A. and Manzocchi, S. (1996) "Unit root tests of capital mobility in the less developed countries"
Wetwirschaftliches Archiv, 132(2), 545-557.
Bayoumi, T and McDonald, R. (1995) "Consumption, income, and international capital market integration"
IMF Staff Paper, 42(3), 552-576.
Campbell, J.Y., Mankiw, N.G., (1989) "Consumption, income, and interest rates: reinterpreting the time
series evidence". NBER Macroeconomics Annual, 185-216.
Campbell, J.Y., Mankiw, N.G. (1990) "Permanent income, current income, and consumption" Journal of
Business Economics Statistics, 8, 265-279.
Campbell, J.Y., Mankiw, N.G. (1991) "The response of consumption to income: a cross-country
investigation" European Economic Review, 35,723-767.
Cooray, A. (2005) "Capital mobility: evidence from four South Asian economies" Applied Economics
Letters, 12, 577-581.
DeBrouwer, G. (1999) Financial Integration in East Asia. Cambridge: New York and Melbourne: Cambridge
University Press.
Feldstein, M. and Horioka, C. (1980) "Domestic saving and international capital flows" Economic Journal,
90 (358), 314-329.
Goh, S. K., Lim, G. C., & Olekalns, N. (2006) "Deviations from uncovered interest parity in Malaysia" Applied
Financial Economics, 16, 745-759.
Goh, S.K. (2007) "Intertemporal consumption smoothing and capital mobility: evidence from Malaysia”
Journal of Business and Public Affairs, 1(1), 1-15.
Ghosh, A.R. and Ostry, J.D (1995) “The current account in developing countries: A perspective from the
consumption smoothing approach” World Bank Economic Review, 9(2), 305-33.
Ho, T. W. (1999) "Export-orientation and investment-saving correlation: a case of Taiwan" Applied
Economics 31, 805-813.
Mamingi, N. (1997) "Saving-investment correlations and capital mobility: the experience of developing
countries" Journal of Policy Modelling, 19(6), 605-26.
Monetary Authority of Singapore (MAS) (2000) Financial Market Integration in Singapore: The Narrow and
The Broad Views, Occasional Paper No.20.
Shibata, A., & Shintani, M. (1998) "Capital mobility in the world economy: an alternative test" Journal of
International Money and Finance, 17, 741-756.
Stock, J.H. and M.Watson (2006) Introduction to Econometrics (2nd edition), New York: Addison-Wesley
Tan, H. B. (2000) Savings, Investments and Capital Flows: Evidence from ASEAN 5 Ashgate Publishing
Company, Aldershot, 79-110.
Tsung, W. H. (1999) "Export-orientation and saving-investment correlation: a case for Taiwan" , 31, 805-
813.

Appendix 1: Consumption-net output correlations
Country Sample Period OLS estimates GMM estimates
I II III
Malaysia
1971-2004
δ
-0.04
(-0.42)
0.688
(3.76)
0.702
(4.66)
-0.179
(-1.15)
R 2 0.006 - - -
J-statistics
2.445 0.9814 3.657
(p-value) - (0.488) (0.805) (0.3009)
F-statistics 1.7644 1.9418 4.8232
Singapore 1972-2004
0.23 0.531 0.401 0.417
δ
(2.58) (9.75) (4.51) (6.32)
R 2 0.18 - - -
J-statistics
2.136 1.073 1.642
(p-value)
(0.544) (0.783) (0.300)
F-statistics - 9.084 5.415 8.464
Notes: t-statistics are reported in parentheses. Equation (10): Δ Ct = (1 − δ ) et + δΔ
Vt.
The sets of
instrumental variables (IV) used in the GMM estimation are {constant, ΔCt-1, ΔCt-2, ΔVt-1 , ΔVt-2}for I
and {constant, ΔVt-1,…, ΔVt-4} for II, {constant, Δct-1,…, Δct-4} for III. The J statistics reported in Eviews
are multiplied by n. The p-values of the J statistics are reported in parentheses, (.).

ENDNOTES
1 The existing available approaches are based on consumption, savings and investment behaviour. These
are based on deviations from international parity condition, i.e., uncovered interest rate parity, covered
interest parity, and real interest rate parity. Another approach, the Feldstin and Horioka’s (1980) framework,
considers the estimated parameter of savings per GDP variable on investment per GDP variable.
Consumption-smoothing in response to shocks to domestic expenditure is an alternative approach to
determine the degree of capital mobility. The final approach is to compare the consumption patterns across
countries.
2 Several models are employed in the study to examine the mobility of capital in Singapore. Models adopted
in this study include international parity condition, consumption smoothing, consumption correlation and
saving-investment correlation test.
3 J-statistic is the value of the GMM objective function, evaluated at the estimated coefficients. The J-statistics
computed by Eviews are multiplied by n.
4 For models with multiple endogenous variables, the F-statistics are no longer informative.
5 The GMM estimation results with lagged endogenous variables as instrumental variables are illustrated in
Appendix 1.
6 Similarly, Cooray (2005) documented that the interest rate differential is not related to changes in
consumption in four South Asian countries, namely, India, Sri Lanka, Pakistan and Bangladesh.
7 We assumed the interest rate differential is an exogenous variable in equation (11).
8 There have been two distinct periods of capital control in Malaysia, 1994-1996, and 1998-2004.