This article was downloaded by: [University of Illinois Chicago]

On: 17 October 2014, At: 20:05

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK

Applied Economics
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/raec20

Export response to trade liberalization in

Bangladesh: a cointegration analysis
Nasiruddin Ahmed
Published online: 04 Oct 2010.

To cite this article: Nasiruddin Ahmed (2000) Export response to trade liberalization in Bangladesh: a cointegration
analysis, Applied Economics, 32:8, 1077-1084, DOI: 10.1080/000368400322138

To link to this article: http://dx.doi.org/10.1080/000368400322138

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any
purpose of the Content. Any opinions and views expressed in this publication are the opinions and views
of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources of information.
Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs,
expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in
connection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 Applied Economics, 2000, 32, 1077 ± 1084

Export response to trade liberalization in

Bangladesh: a cointegration analysis
N A S I R U D D I N A H ME D
Prime Minister’s O ce, Dhaka 1215, Bangladesh, India

This paper investigates the response of Bangladesh’s aggregate merchandise exports

Downloaded by [University of Illinois Chicago] at 20:05 17 October 2014

to a real exchange rate-based trade liberalization programme during the period

1974 ± 1995. The cointegration and error correction modelling approaches have
been applied. The empirical results suggest that there exists a unique long-run or
equilibrium relationship among real quantities of export, relative export price and
export-weighted real e€ ective exchange rate. The short-term dynamic behaviour of
Bangladesh’ s export supply has been investigated by estimating an error correction
model in which the error correction term has been found to be correctly signed and
statistically signi® cant. Relative export price (lagged two quarters), real e€ ective
exchange rate, predicted values of real GDP (lagged one quarter) and a dummy
variable capturing the e€ ects of trade liberalization programme have all emerged
as important determinants of an aggregate export supply function for Bangladesh.
The error correction model has also been found to be robust as it satis® es all relevant
diagnostic tests.

I . I N T R OD U C T I ON Helleiner, 1994; Joshi and Little, 1996). The rationale is

that trade liberalization reduces anti-export bias and
The relation between trade policy and economic perform- makes exports more competitive in international markets.
ance is a controversial issue in economics. Several com- This is done primarily by correcting an overvalued
parative studies provide evidence about the positive e€ ect exchange rate and creating incentives for expanding the
of trade liberalization on economic growth and exports tradeable sector. On the other hand, three studies for
(Little et al., 1970; Balassa, 1971, 1982; Bhagwati, 1978; UNCTAD (UNCTAD, 1989; Agosin, 1991; Shafaeddin,
Krueger, 1978; Michaely et al., 1991). Besides these aggre- 1994), studies by Clarke and Kirkpatrick (1992),
gate studies, some studies used manufacturing ® rm-level Greenaway and Sapsford (1994) , and Jenkins (1996) have
data to provide empirical evidence on the e€ ects of trade expressed scepticism and found little evidence to support a
liberalization on productivity growth (Harrison, 1994; link between trade liberalization and export performance.
Tybout and Westbrook, 1995; Gokcekus, 1997). These In view of the above contradictory ® ndings, it may be
studies show that trade liberalization, or decline in protec- worthwhile to re-examine the fundamental determinants
tion, has a signi® cant positive e€ ect on productivity of export expansion in the context of trade liberalization.
growth. These studies also shed light on the channels This paper is mainly motivated by two considerations . In
through which productivity growth manifests itself. the ® rst place, in keeping with the objective of greater
A number of developing countries which embarked on openness and outward orientation, trade liberalization
liberalization over the past two decades experienced measures have been adopted as part of structural adjust-
improvements in export performance, with signi® cant ment policies since 1986 in Bangladesh. 1 In fact, signi® cant
increases in manufactured exports (Thomas and Nash, steps have been taken during the 1990s to liberalize
1991; Weiss, 1992; Arslan and van Wijnbergen, 1993; Bangladesh’ s trade regime in order to solve the problem

1
Structural adjustment programmes were undertaken in Bangladesh along the prescriptions of the Bretton Woods Institutions ± the
World Bank and the International Monetary Fund (IMF).
Applied Economics ISSN 0003± 6846 print/ISSN 1466± 4283 online # 2000 Taylor & Francis Ltd 1077
http://www.tandf.co.uk/journals
 1078 N. Ahmed.
of persistent trade de® cit. However, while structural 1995). This is because the limiting distribution of the
reforms in such areas as the ® nancial sector, legal reform, asymptotic variance of the parameter estimates is not
infrastructure and energy are di cult and politically sensi- ® nitely de® ned (Fuller, 1985). Phillips (1986) found that
tive, trade policy reforms, the bulk of which consists of the distributions of the student’s t-statistic diverge so that
liberalizing imports, is perceived to be relatively easy to there are no asymptotically correct critical values for these
implement (World Bank, 1996). The primary objective of tests. This study applies cointegration-erro r correction
trade liberalization policies has been to move towards a framework for estimation of the model. This is because
liberalized trade regime with ¯ exible exchange rate man- this framework helps identify long-run relationships as
agement. This paper is motivated by the recent attempts in well as the short-run dynamics between external sector
Bangladesh to liberalize its external sector which accounts variables and other macroeconomic variables for trade pol-
for more than 30% of Bangladesh’ s GDP at current prices. icy modelling.
As the available literature reviewing trade policy reforms in With the above backgroun d in mind, this paper intends
Bangladesh is scanty, the present study will ® ll some gaps. 2 to undertake an econometric analysis of the response of
One of the major goals of trade policy reform is to elim- Bangladesh’ s aggregate merchandise exports to a real
inate the anti-export bias of the past protectionist policies exchange rate-based trade liberalization programme during
Downloaded by [University of Illinois Chicago] at 20:05 17 October 2014

and thus improve the economy’ s export performance as the period 1974± 1995. The remainder of this paper is
well as its overall e ciency in resource allocation. A real organized as follows. Section II presents theoretical frame-
exchange rate-based trade liberalization policy helps to work of the study. A modelling of the export supply func-
achieve a reduction in the anti-export bias. The reason tion for Bangladesh is presented in Section III. The
for this is that in the presence of quantitative import empirical results are reported and discussed in Section
restrictions, a (real) devaluation will result in a lower pre- IV. Section V presents summary and conclusions.
mium (PR) accrued to those with import licence alloca-
tions, and thus in a reduction in the anti-export bias.
Available empirical evidence shows that successful trade
I I . TH E TH E OR E T I C A L F R A ME W OR K
liberalization has been associated with devaluations either
at the same time or beforehand (Bhagwati, 1978; Krueger, Following Dornbusch (1974) and Corden (1997), we con-
1978; Michaely et al., 1991; Edwards, 1993; Little, et al., sider a country that consumes and produces three cate-
1993; Corden, 1997). In this paper we employ real e€ ective gories of commodities, namely, exportables (X ),
exchange rate (REER),3 the key variable in this paper, as importables (M ) and nontradables (N ). The relative
an index of export competitiveness. A REER includes the price of traded goods in the world market is taken as
nominal exchange rate, tari€ s, trade subsidies, and domes- given ± the usual assumption for a small country. Excess
tic and world market prices; as such it is obviously involved demand for M will be imported and excess supply of X will
in the process of liberalization. A REER re¯ ects the change be exported. There is a single tari€ applied to M, the tari€
in a country’s competitive position relative to its trading rate being t, and there are no export taxes and subsidies.
partners. The domestic prices of M and X are pm and px respectively.
Secondly, we ® nd methodological inadequacies in the These two prices depend on foreign prices, on the exchange
available literature on the subject. This paper expresses rate …e† and, in the case of M, on the tari€ rate. Thus,
scepticism about the validity of the empirical results of pm ˆ ep¤m …1 ‡ t† and px ˆ ep¤x . Assuming p¤m and p¤x con-
most of the earlier studies on aggregate export demand stant, pm can be increased only by a rise in e or in t, and
function for Bangladesh. This scepticism is based on the px only by a rise in e. In addition, it is assumed that pn (the
now universal consensus that it is inappropriate to apply price of nontradables) is constant, so that either a rise in pm
conventional econometric techniques to nonstationary time or in px must lead to a rise in S.
series. 4 This is due to the fact that to estimate a regression In Fig. 1 the axes show pm and px . The line OT0 repre-
model containing nonstationary variables at best ignores sents free trade and the free trade equilibrium is at point
important information about the underlying (statistical Q1 , where the line OT0 , intersects the S0 schedule (the
and economic) processes generating the data, and at switching ratio). The imposition of an import tari€ raises
worst leads to nonsensical (or spurious) results (Harris, pm relative to px by the amount of the tari€ and accord-

2
Such studies include, inter alia, Rahman (1992) , Bayes et al. (1995), Hossain et al. (1997) , the World Bank (1996) and Dutta and Ahmed
(1999) .
3
The estimation of the REER is problematic because of the di culty in ® nding appropriate domestic and international price indices.
4
The empirical evidence provided by Nelson and Plosser (1982) and Meese and Singleton (1983) have shown that in reality, aggregate
economic time-series are not stationary in their levels and therefore contain variances that explode with time (DeJong and Whiteman,
1991).
 T rade liberalization in Bangladesh 1079
pm and 1996 respectively. As the exporters in a small country5
T1 like Bangladesh take the demand conditions in the import-
ing countries as given, Bangladesh’ s export demand is in® -
Q/O
nitely price elastic. This in® nite elasticity of export demand
T0 allows us for the estimation of a single equation export
supply function.
Q2 Q0 In modelling our export trade we shall follow the imper-
S1 fect substitutes model, in which the key assumption is that
Q1
So
neither imports nor exports are perfect substitutes for
domestic goods. The supply function is derived from the
assumption of pro® t maximization on the part of the pro-
0 px ducer. In this model we estimate price elasticity of exports
Fig. 1. Real exchange rate-based trade liberalization policies because it shows the responsiveness of a country’ s volume
of exports to the relative export price.
ingly, yields the line OT1 . The new equilibrium will be at Secondly, since exports are supply-constrained , an
Downloaded by [University of Illinois Chicago] at 20:05 17 October 2014

point Q00 . The new equilibrium could have been at the point increase in production capacity of the economy is likely
Q2 . But since t is added to pm , the ratio of traded goods to have a positive e€ ect on export performance. The use
changes causing changes in the prices of traded goods rela- of such a scale variable is justi® ed on the ground that a
tive to nontradables. As such the switching ratio schedule country’ s exports will not only depend on export prices but
shifts from S0 to S1 . also on excess supply which in turn is dependent on the
We now consider trade liberalization in terms of reduc- output capacity of the country. Following Bayes et al.
tion of import tari€ rate, starting at Q00 . A fall in the import (1995) and Hossain et al. (1997), we use a capacity utiliza-
tari€ rate (to zero in the diagram), with the exchange rate tion variable given by the trend output obtained from the
constant, brings the economy to Q1 , representing the new predicted values of an estimated exponential growth func-
(zero) tari€ level. However, in order to keep S constant, tion. The greater utilization of capacity leads to higher
trade liberalization will require a real depreciation of the exports, although it is likely that higher exports lead to a
domestic currency, bringing the economy to Q0 . greater utilization of available capacity.
The movement from Q00 to Q0 has two implications. Thirdly, real e€ ective exchange rate (REER), the key
First, there is the direct e€ ect within the tradables sector explanatory variable in this model, brings together changes
as a whole. The reduction in the import tari€ rate reduces in the nominal exchange rate, the e€ ective value of ® nan-
pm relative to px . The decline in pm relative to px moves cial incentives, and domestic and world prices. It is, there-
resources from imports …M† to exports …X†. Secondly, re- fore, an index of export competitiveness. Trade
sources move out of nontradable s …N† into exports …X†, the liberalization is likely to depreciate REER.
incentive being the real depreciation, which raises the price Fourthly, an intercept dummy variable is included in the
of exportables …px † relative to that of nontradables …pn †. In model to capture the distinction (if any) between pre- and
the absence of quantitative restrictions on some imports, a post-trade liberalization periods with regard to export per-
real devaluation increases the price not only of tradables formance which is not expressed in REER.
relative to nontradables but also of exportables relative to The long-run aggregate export supply function for
importables, thereby reducing anti-export bias and and Bangladesh is speci® ed as follows:
thus increases export competitiveness. This theoretical L Xt ˆ ¬0 ‡ ¬1 L Pt ‡ ¬2 L PGDPt ‡ ¬3 L REERt
analysis brings out the importance of real exchange rate-
‡ ¬4 D t ‡ ut …1†
based trade liberalization policies for increasing export
competitiveness. where X is real quantity of aggregate merchandise exports;
P is relative price of exports (unit value index of exports for
Bangladesh to that of industrial countries) ; PGDP is pre-
dicted values of real GDP …1990 ˆ 100† used as a proxy
I I I . MOD E L LI N G E X POR T S U PPL Y variable for the production capacity of the economy;
F U N C T I ON F OR B A N G L A D E S H REER is export-weighted real e€ ective exchange rate
…1990Q4 ˆ 100†; D is an intercept dummy variable with
Bangladesh’ s exports constitute a small proportion of values 0 for 1974± 91 and 1 for 1992± 95; u is random dis-
world exports, being as low as 0.04% and 0.06% in 1975 turbance term with its usual classical properties; and L is

5
The assumption that domestic producers and consumers face given world prices for goods and services is referred to as the small country
assumption.
 1080 N. Ahmed.
natural logarithm. It is expected that ¬3 < 0; and …¬1 ;¬2 where ECt-1 ˆ error-correction term lagged one period.
and ¬4 † > 0: The modelling strategy adopted in this study involves
If the time series variables of L Xt , L Pt , L PGDPt and three steps:
L REERt have unit roots, then we need to take the ® rst
di€ erence of the variables (as in Equation 2) in order to (i) determining the order of integration of the variables
obtain a stationary series: by employing Augmented Dickey± Fuller (ADF)
and Phillips± Perron unit-root tests;
DL Xt ˆ ¬0 ‡ ¬1 DL Pt ‡ ¬2 DL PGDPt ‡ ¬3 DL RFERt
(ii) if the variables are integrated of the same order, we
‡ ¬ 4 D t ‡ ut …2† test for cointegration by applying the Johansen±
Juselius (1990) approach;6 and
Equation 2 ignores any reference to the long-run aspects of
(iii) if the variables are cointegrated, we can specify an
decision-making. That is, this procedure of di€ erencing
error correction model and estimate it using stan-
results in a loss of valuable `long-run information’ in the
dard methods and diagnostic tests.
data (Maddala, 1992). The theory of cointegration
addresses this issue by introducing an error-correction
(EC) term. The EC term lagged one period (i.e., ECt-1 )
Downloaded by [University of Illinois Chicago] at 20:05 17 October 2014

I V . EM PIR I C A L A N A L Y S I S
integrates short-run dynamics in the long-run export sup-
ply function. This leads us to the speci® cation of a general
T ime-series properties of the data
error correction model (ECM):
n
X n
X Unit root tests for stationarity are performed on levels of all
DL Xt ˆ ­ 0 ‡ ­ 1i DL Xt¡i ‡ ­ 2i DL Pt¡i ‡ the four variables and ® rst di€ erences for all the variables
iˆ1 iˆ0 other than PGDP which is found stationary in its level. The
n
X n
X DF± ADF tests (Table 1) show the existence of unit roots,
­ 3i DL PGDPt¡i ‡ ­ 4i DL REERt¡i and therefore nonstationarity , in the levels of L X, L P and
iˆ0 iˆ0
L REER which is con® rmed by the Phillips± Perron unit-root
‡­ 5 ECt¡1 ‡ ­ 6 D t ‡ "t test (Table 2). However, the ® rst di€ erences of these vari-
Table 1. DF and ADF unit root tests for stationarity

Variables Level/First Di€ . Type of test Without trend With trend Conclusion

LX Level DF 70.96 74.03

ADF(1) 70.71 73.45
ADF(2) 70.16 71.82
First di€ . DF 79.12 79.01 I…1†
ADF(1) 78.67 78.52
ADF(2) 78.18 78.01
LP Level DF 71.63 71.38
ADF(1) 72.09 72.26
ADF(2) 71.85 71.88
First di€ . DF 76.03 76.01 I…1†
ADF(1) 75.86 75.86
ADF(2) 73.97 73.98
L REER Level DF 73.01 75.43
ADF(1) 72.17 75.15
ADF(2) 72.24 75.03
First di€ DF 79.12 79.01 I…1†
ADF(1) 78.67 78.52
ADF(2) 78.18 78.01
L PGDP Level DF 784.03 7127.83 I…0†
ADF(1) 753.88 716.48
ADF(2) 75.16 711.05

Notes: (i) Unit root tests were performed using Micro® t 4.0.
(ii) 95% critical values for ADF statistic (without trend) ˆ 72.90
(iii) 95% critical values for ADF statistic (with trend) ˆ 73.46

6
The Johansen± Juselius approach (1990) is preferred in the case of more than one cointegrating vector because the Engle± Granger
approach (1987) has several limitations in the case of more than one cointegrating vector.
 T rade liberalization in Bangladesh 1081
Table 2. Phillips ± Perron (PP ) unit root test for stationarity

Variables Level/First di€ . Constant, no trend Constant, trend Conclusion

LX Level 70.86 72.15

(72.57) (73.13)
First di€ . 713.05 712.94 I…1†
(72.57) (73.13)
LP Level 71.77 71.96
(72.57) (73.13)
First di€ . 76.09 76.07 I…1†
(72.57) (73.13)
L REER Level 72.44 72.93
(72.57) (73.13)
First di€ . 76.73 76.85 I…1†
(72.57) (73.13)
L PGDP Level 737.74 758.74 I…0†
Downloaded by [University of Illinois Chicago] at 20:05 17 October 2014

(72.57) (73.13)

Notes: (i) PP test was performed using SHAZAM 8.0.

(ii) Figures in brackets indicate critical values at 10% level.

Table 3. Johansen± Juselius maximum likelihood cointegration tests

Maximal eigenvalue test Trace test

Null Alternative Statistic 95% critical value Null Alternative Statistic 95% critical value

rˆ0 rˆ1 31.09 17.68 rˆ0 r¶1 39.28 24.05

rµ1 rˆ2 5.27 11.03 rµ1 r¶2 8.19 12.36
rµ2 rˆ3 2.93 4.16 rµ2 r¶3 2.93 4.16

Notes: (i) The test was performed using Micro® t 4.0

(ii) r stands for the number of cointegrating vectors.

ables are stationary under the test. Hence we conclude that Turning to the trace test, the null hypothesis of no coin-
these three variables are integrated of order 1. tegration …r ˆ 0† is rejected at 5% level of signi® cance in
favour of the general alternative, that there is one cointe-
grating vector, r ˆ 1. But the test fails to reject the null
Cointegration test hypotheses of r 4 1 and r 4 2. This con® rms the conclu-
sion that there is only one cointegrating relationship
Before undertaking the cointegration tests, let us ® rst specify
amongst the three I…1† variables.
the relevant order of lags …p† of the VAR model. Given the
fact that the sample size is relatively small, we select 2 for the
order of the VAR (Pesaran and Pesaran, 1997). On the basis Estimation of an error-correction model
of the above unit-root tests, we apply the procedure of
Johansen and Juselius (JJ) (1990) to determine whether Once a cointegrating relationship is established, then an
any combinations of the variables are cointegrated. ECM can be estimated to determine the short-run dynamic
Starting with the null hypothesis of no cointegration behaviour of money demand. Following Hendry’ s (1995)
…r ˆ 0† among the variables, the maximal eigenvalue general-to-speci® c modelling approach, we ® rst include
statistic is 31.09, which is above the 95% critical value of four lags of the explanatory variables and of the error
correction (EC) term, and then gradually eliminate the
17.68. Hence it rejects the null hypothesis r ˆ 0, in favour
insigni® cant variables. After experimenting with the gen-
of the general alternative r ˆ 1. As is evident in Table 3, the
eral form of ECM (Equation 3) , the following ECM is
null hypotheses of r 4 1 and r 4 2 cannot be rejected at a
found to ® t the data best (Table 4).
5% level of signi® cance. Consequently, we conclude that In this model, relative export price, REER, predicted
there is only one cointegrating relationship among the values of real GDP and an intercept dummy variable
three variables of L X, L P and L REER. have all emerged as signi® cant determinants of an aggre-
 1082 N. Ahmed.
Table 4. Estimated error-correction model (UNCTAD, 1989). Between 1991 and 1994 there was a net
depreciation of the REER of about 12% which, with lower
Dependent variable D LX
in¯ ation, improved external competitiveness (Shand and
Regressors Parameter estimates T -ratio Alauddin, 1997). The country has successfully maintained
a competitive, e€ ective real exchange rate (Quibria, 1997).
Intercept 0.23 0.20
D L P…¡2† 0.65 2.06
In a ¯ exible exchange rate regime, import liberalization
D L REER 70.96 72.17 through tari€ reductions is expected to lead to a deprecia-
L PGDP…¡1† 0.41 3.21 tion of REER (that is, a real devaluation) through a
D 0.10 1.65* strengthening of import demand. The net e€ ect of this
EC…¡1† 70.33 73.50 action will be to reduce incentives for import-substitutin g
Adj. R2 ˆ 0:20
D:W: ˆ 2:17
industries which have previously enjoyed high rates of pro-
RESET ˆ 1:48 tection, but also to increase incentives for export industries.
NORM ˆ 0:15 The coe cient of the intercept dummy variable used to
HET ˆ 2:13 represent regime shift is statistically signi® cant at 10% level
with the expected (positive) sign. This result provides sta-
Note: *Signi® cant at 10%
Downloaded by [University of Illinois Chicago] at 20:05 17 October 2014

tistical support for the view that, apart from its impact
operating through the incentives captured in REER, the
gate export supply function for Bangladesh. The error liberalization policy package has been instrumental in
correction coe cient, estimated at ¡0:33 is statistically sig- creating a conducive environment for export expansion.
ni® cant at 1% level, has the correct sign, and suggests a However, the coe cient estimate of the dummy variable
high speed of convergence to equilibrium. The diagnostic is very low (0.10) in the model. This result seems to be
test statistics show no evidence of misspeci® cation, no expected in Bangladesh, where the liberalization policies
serial correlation, nor any problem of heteroscedasticity cannot be fully e€ ective until various structural bottlenecks
and no problem of non-normality in residuals. responsible for low export expansion are dealt with.
The policy implications are straightforward. For rapid
expansion of exports, trade liberalization policies need to
be associated with devaluations. This is because a real
V . S U MMA R Y A N D C ON C L U S I ON S exchange rate-based trade liberalization policy helps to
achieve a reduction in the anti-export bias. The ® ndings
In our empirical investigation of the aggregate export sup- also point out the need for adopting public policies for
ply function for Bangladesh, cointegration and error cor- greater utilization of productive capacity of the economy,
rection modelling approaches have been applied, we ® nd a leading to higher growth of exports. However, for e€ ective
unique equilibrium relationship among real quantities of policy analysis studies may be undertaken using data at the
export, relative export price and the export-weighted real disaggregate level.
e€ ective exchange rate. In order to determine the short-
term dynamics around the equilibrium relationship, we
estimated an error correction model (ECM). In the
model, relative export price (lagged two quarters), real A C K N OW L E D G E ME N T S
e€ ective exchange rate, production capacity of the econ- The author is deeply grateful to Professor W. Max Corden
omy proxied by the predicted values of the real GDP and Dr. Dilip Dutta for useful discussions on earlier drafts
(lagged one quarter), and an intercept dummy variable and an anonymous referee for helpful comments.
capturing the e€ ects of trade liberalization programme
have all emerged as important determinants of the short-
term dynamics of the aggregate export supply function for
Bangladesh. The error correction term in the model is R EFER ENCES
found to be statistically signi® cant, con® rming the validity Agosin, M. R. (1991) Trade policy reform and economic perform-
of the long-run equilibrium relationship. The coe cient ance: a review of the issues and some preliminary evidence,
UNCT AD Discussion Papers No. 41, UNCTAD, Geneva.
estimate of the error correction term is ¡0:33 indicating a Arslan, I. and Wijnbergen, S. v. (1993) Export incentives,
high speed of adjustment to equilibrium. exchange rate policy and export growth in Turkey, Review
Our econometric estimates of the aggregate export sup- of Economics and Statistics, 75(1) , 128± 33.
ply function for Bangladesh suggest a high negative elasti- Bahmani-Oskooee, M. (1995) Real and nominal e€ ective
city of our export earnings to the export-weighted REER exchange rates for 22 LDCs: 1971: 1± 1990:4, Applied
Economics, 27(7), 591 ± 604.
indicating an increase in our export competitiveness. This is Balassa, B. and Associates (1971) (Ed.), T he Structure of
consistent with the fact that the Bangladesh currency (taka) Protection in Developing Countries, Johns Hopkins
was depreciated about 20% between 1984/85 and 1987/88 University Press, Baltimore.
 T rade liberalization in Bangladesh 1083
Balassa, B. and Associates (1982) Developing Strategies in Semi- demand for money, Oxford Bulletin of Economics and
Industrial Economies, Johns Hopkins University Press for the Statistics, 52, 169± 210.
World Bank, Baltimore. Joshi, V. and Little, I. M. D. (1996) India’s Economic Reforms
Bayes, A. , Hossain, M. I. and Rahman, M. (1995) Trends in the 1991± 2001, Oxford University Press, Oxford.
external sector: trade and aid in Experiences with Economic Krueger, A. O. (1997) Trade policy and economic development:
Reform: A Review of Bangladesh’ s Development 1995, (Ed.) how we learn, T he American Economic Review, 87(1), 1± 22.
R. Sobhan, Centre for Policy Dialogue and University Press Little, I. M. D., Scitovsky, T. and Scott, M. (1970) Industry and
Limited, Dhaka, pp. 243± 97. T rade in Some Developing Countries: A Comparative Study,
Bhagwati, J. N. (1978) Anatom y and Consequences of Exchange Oxford University Press for OECD, London.
Control Regimes, Ballinger for the National Bureau of Little, I. M. D., Cooper, R. N., Corden, W. M. and Rajapatirana,
Economic Research, Cambridge, MA. S. (1993) Boom, Crisis and Adjustment: The Macroeconomic
Clarke, R. and Kirkpatrick, C. (1992) Trade policy reform and Experience of Developing Countries, Oxford University Press
economic performance in developing countries: assessing the for the World Bank, Oxford.
empirical evidence, in Industrial and Trade Policy Reform in Maddala, G. S. (1977) Econometrics , McGraw-Hill, New York.
Developing Countries, (Eds) R. Adhikari, C. Kirkpatrick and Meese, R. A. and Singleton, K. J. (1983) On unit roots and the
J. Weiss, Manchester: Manchester University Press. empirical modeling of exchange rates, International Economic
Corden, W. M. (1997) T rade Policy and Economic W elfare, Review, 24, 1029 ± 35.
Second edn. Clarendon Press, Oxford. Michaely, M., Papageorgiou, D. and Choksi, A. M. (Eds) (1991)
Downloaded by [University of Illinois Chicago] at 20:05 17 October 2014

DeJong, D. N. and Whiteman, C. H. (1991) Reconsidering trends L iberalising Foreign T rade: L essons of Experience in the
and random walks in macroeconomic time series, Journal of Developing W orld, Basil Blackwell, Oxford.
Monetary Economics, 28, 221 ± 54. Nelson, C. R. and Plosser, C. I. (1982) Trends and random walks
Dornbusch, R. (1974) Tari€ s and nontraded goods, Journal of in macroeconomic time series, Journal of Monetary
International Economics, 4, 177± 85. Economics, 10, 139 ± 62.
Dutta, D. and Ahmed, N. (1999) An aggregate import demand Pesaran, M. H. and Pesaran, B. (1997) W orking with Microsoft
function for Bangladesh: a cointegration approach, Applied 4.0: Interactive Econometric Analysis, Oxford University
Economics, 31, 465± 72. Press, Oxford.
Edwards, S. (1993) Openness, trade liberalization, and growth in Phillips, P. C. B. (1986) Understanding spurious regressions in
developing countries, Journal of Economic L iterature, econometrics, Journal of Econometrics, 33, 311± 40.
XXXI(3), 1358 ± 93. Quibria, M.G. (1997) Introduction, in T he Bangladesh Econom y
in T ransition, (Ed.) M. G. Quibria, Oxford University Press,
Engle, R. F. and Granger, C. W. J. (1987) Cointegration and
Delhi, pp. 1± 6.
error correction: representation, estimation and testing,
Rahman, S. H. (1992) Structural adjustment and macroeconomic
Econometrica , 55, 251 ± 76.
performance in Bangladesh in the 1980s , The Bangladesh
Fuller, W. A. (1985) Nonstationary autoregressive time series, in
Development Studies, XX(2&3), 89± 125.
Handbook of Statistics, 5 (Eds) E. J. Hannan, P. R.
Shafaeddin, S.M. (1994) The impact of trade liberalization on
Krishnaiah and M. M. Rao, North Holland, Amsterdam.
export and GDP in least developed countries, UNCT AD
Gokcekus, O. (1997) Trade liberalization and productivity
Discussion Papers No. 85, UNCTAD, Geneva.
growth: new evidence from the Turkish rubber industry, Shand, R. and Alauddin, M. (1997) Economic Pro® les in South
Applied Economics, 29(5), 639 ± 45. Asia: Bangladesh, Australian National University, Canberra.
Greenaway, D. and Sapsford, D. (1994) What does liberalisation Thomas, V., Nash, J. and Associates (1991) Best Practices in
do for exports and growth, W eltwirtschaftliches Archiv, T rade Policy Reform, Oxford University Press for the
130(1) , 152± 74. World Bank, Oxford.
Harris, R. I. D. (1995) Using Cointegration Analysis in Tybout, J. R. and Westbrook, M. D. (1995) Trade liberalisation
Econometric Modelling, Harvester Wheatsheaf, London. and the dimensions of e ciency change in Mexican manu-
Harrison, A. E. (1994) Productivity, imperfect competition and facturing industries, Journal of International Economics, 39
trade reform, Journal of International Economics, 36(1/2), 53± (1/2), 53± 78.
73. UNCTAD (1989) T rade and Development Report 1989.
Harvey, A. C. (1981) Econometric Analysis of T ime Series, Philip Wallis, K. F. (1974) Seasonal adjustment and relations between
Allan Publishers Limited, Oxford. variables, Journal of the American Statistical Association, 69,
Helleiner, G. K. (1994) (Ed.) T rade Policy and Industrialization in 18± 31.
T urbulent T imes, Routledge, London and New York. Weiss, J. (1992) Export response to trade reforms: recent Mexican
Hendry, D. F. (1995) Dynamic Econometrics. Oxford University experience, Development Policy Review, 10(1), 43± 60.
Press, Oxford. The World Bank (1996) Bangladesh: T rade Policy Reform for
Hossain, M. I., Rahman, M. A. and Rahman, M. (1997) Current Improving the Incentive Regime, Report no.15900-BD.
external sector performance and emerging issues in Growth or
Stagnation? A Review of Bangladesh’s Development 1996.
(Ed.) R. Sobhan, Centre for Policy Dialogue and
University Press Limited, Dhaka, pp. 161± 220.
International Monetary Fund (1998) International Financial A PPE N D I X 1 : TH E D A T A
Statistics on CD-ROM.
International Monetary Fund (1997) Direction of T rade Statistics The source of our data comprises the International
Y earbook 1996, IMF, Washington, DC.
Monetary Fund’ s (IMF) various issues of International
Jenkins, R. (1996) Trade liberalization and export performance in
Bolivia, Development and Change, 27(4) , 693± 716. Financial Statistics (IFS). Twenty-two years of quarterly
Johansen, S. and Juselius, K. (1990) Maximum likelihood estima- data (1974Q2± 1995Q4) of the Bangladesh economy have
tion and inference on cointegration with applications to the been used in the present study. The nominal values of all
 1084 N. Ahmed.
the variables (expressed in terms of domestic currency) trading partners (country i’ s) are calculated from the
have been converted into real ones. The nominal values exchange rate data of International Financial Statis-
of aggregate merchandise exports are de¯ ated by the unit tics (IFS) involving the US dollar.
value index of exports in order to obtain real quantities of (2) We calculate the real bilateral exchange rates
exports. Secondly, unit value indices of Bangladesh’s …REXij † using EXij ’ s and CPI indexes (from IFS):
exports are de¯ ated by unit value indices of exports of REXij ˆ …REXijt · CPIj =CPIi † i 6ˆ j
industrial countries to obtain relative exports. Thirdly, (3) We construct the index of real bilateral exchange
the trend values of production capacity of the rates …IREXij † by selecting the fourth quarter of
Bangladesh economy are obtained by regressing GDP at 1990 as the base period: IREXij ˆ
1990 prices on a time trend. Finally, export-weighted …REXijt =REXij90 †· 100
REER is computed using Bahmani-Oskooee (1995) method (4) We take the weighted average of IREXij in order to
(Appendix 2). obtain the index of real e€ ective exchange rates
Finally, we plotted the data on real export, relative P
…REERj † for Bangladesh: REERj ˆ niˆ1 ¬ij IREXij
export price, predicted values of real GDP, and the real where ¬ij are the trading weights (export shares) of
e€ ective exchange rate in order to check seasonality in Bangladesh’s export trading partners and
Downloaded by [University of Illinois Chicago] at 20:05 17 October 2014

the data. On inspection of the graphs, we found no strong Pn

¬
iˆ1 ij
ˆ 1. These shares are taken from Directory
seasonal e€ ects. Moreover, there are widespread criticisms of T rade Statistics Y earbook 1997 published by the
over the use of seasonally adjusted data (Wallis, 1974; IMF. We have selected nine major export trading
Maddala 1977; Harvey, 1981) ; so we have not made any partners of Bangladesh, which are USA, UK,
seasonal adjustment to the data. Canada, Japan, Belgium, France, Germany, Italy
and Netherlands. About 80% of Bangladesh’s
A PPE N D I X 2 : C ON S TR U C T I N G T H E R E A L exports go to these countries.
E F F E C T I V E EX C H A N G E R A T E F OR
B A N G L A D ES H U S I N G B A H M A N I ±
OS K OOE E ( 1 9 9 5 ) ME T H OD
(1) The bilateral exchange rates …EXij † between Bangla-
desh’s (country j) currency and the currencies of its