Professional Documents
Culture Documents
MOHSEN BAHMANI-OSKOOEE
Department of Economics, The University of Wisconsin–Milwaukee,
Milwaukee, Wisconsin, USA
HANAFIAH HARVEY
Department of Economics, Pennsylvania State University Mont Alto,
Mont Alto, Pennsylvania, USA
I. INTRODUCTION
The response of the trade balance to exchange rate changes still continues
to attract interest from many. Old studies claimed that if the Marshall-Lerner
condition is satisfied, currency devaluation under a fixed exchange rate sys-
tem or currency depreciation under a flexible exchange rate system will
103
104 M. Bahmani-Oskooee and H. Harvey
improve the trade balance in the long run.1 However, in the early 1970s,
Magee (1973) argued that, due to adjustment lags, the trade balance could
deteriorate after devaluation and only improve after the lags are realized,
hence the J-Curve pattern. Bahmani-Oskooee (1985) provided an empirical
counterpart to Magee (1973) by introducing a reduced form trade balance
model that was estimated using regression analysis. Backus et al. (1994)
introduced an alternative approach that was based on cross-correlation
between current exchange rate and past as well as future values of the trade
balance that came under the heading of the S-Curve.2
Since this article is concerned with the response of Indonesia’s trade
balance to currency depreciation, a short account of the literature is in order.
Indonesia is one of the member countries in the Association of South East
Asian Nations (ASEAN) and it is also one of the largest economies in ASEAN
with a nominal GDP of $846 billion in 2011. The Bank of Indonesia (central
bank) allows the rupiah to float freely. However, it is committed to intervene
should the rupiah become more volatile.3
Studies that estimated price elasticities to infer the Marshall-Lerner con-
dition include Houthakker and Magee (1969), Khan (1974), Warner and
Kreinin (1983), Gylfason and Risager (1984), Rose (1991), Bahmani-Oskooee
(1986), Bahmani-Oskooee (1998), Bahmani-Oskooee and Niroomand (1998),
Bahmani-Oskooee and Kara (2005), and Sweidan (2013). Unfortunately,
none of these studies included Indonesia in their sample. Bahmani-Oskooee
and Alse (1994), who relied upon cointegration between the trade balance
and the real exchange rate as an alternative approach to testing the Marshall-
Lerner condition, did include Indonesia among 41 countries in their sample.
However, since Indonesia’s trade balance and its real effective exchange
rate were integrated of different order, cointegration technique could not be
applied to the case of Indonesia. Lal and Lowinger (2002) carried out another
study that estimated a trade balance model for seven Asian countries includ-
ing Indonesia. Using Johansen’s cointegration technique, they showed that,
although there were two cointegrating vectors between Indonesian trade
balance with the rest of the world and its determinants, the real effective
exchange rate was not significant in the long run. In the short run, however,
the J-Curve was detected.4
1
The Marshall-Lerner condition basically claims for a small country that, as long as the sum of
import and export demand elasticities adds up to unity, devaluation will improve the trade balance.
Bahmani-Oskooee et al. (2013) provide the most comprehensive review of the literature.
2
For a review of the literature on J- and S-curves, see Bahmani-Oskooee and Hegerty (2010).
3
Bank of Indonesia.
4
Another body of the literature aims at estimating import and export demand functions separately.
Examples include King (1993), Alse and Bahmani-Oskooee (1995), Charos et al. (1996), Truett and Truett
(2000), Du and Zhu (2001), Love and Chandra (2005), Agbola and Damoense (2005), and Narayan and
Narayan (2005).
The J-Curve 105
Assessing the long-run relation between the trade balance and the real
exchange rate at the commodity level has now become a tradition. In this
article, we follow Bahmani-Oskooee and Xu (2012) and adopt the following
specification:
Since data are reported by the U.S., we define the variables and coefficient
signs from the U.S. perspective. As such, in Equation (1), TBi measures the
trade balance of industry i and it is defined as the ratio of U.S. exports of
commodity i to Indonesia over its import of commodity i from Indonesia.
5
The trading partners were: Australia, Canada, China, Hong Kong, Japan, Korea, Malaysia, New
Zealand, Philippines, Saudi Arabia, Singapore, Thailand, the U.S., and the UK.
106 M. Bahmani-Oskooee and H. Harvey
Clearly, the main determinants of industry i’s trade balance are identified to
be level of economic activity in Indonesia (YIndo ), level of economic activity
in the U.S. (YUS ), and the real bilateral exchange rate (REX). Since an increase
in the level of economic activity in Indonesia is expected to boost the U.S.
exports to that country, we expect an estimate of b to be positive. However,
since an increase in the U.S. economic activity is expected to boost U.S.
imports from Indonesia, we expect an estimate of c to be negative. From
the Appendix, we gather that the definition of the REX follows the definition
of the nominal exchange rate, which is defined to be number of dollars per
rupiah. Therefore, if a dollar depreciation or an increase in REX is to improve
the U.S. trade balance of industry i, we expect an estimate of d to be positive.
Since the J-Curve effect is a short-run concept, we need to incorporate
a short-run dynamic into Rquation (1). Bahmani-Oskooee and Xu (2012) fol-
low Pesaran et al.’s (2001) bound testing approach and specify Equation (1)
in an error-correction modeling format as in Equation (2).
n1
n2
n3
LnTBi, t = α + bk Ln TBi, t−k + ck Ln YIndo, t−k + dk Ln YUS, t−k
k=1 k=0 k=0
n4
+ ek Ln REXt−k + λ1 Ln TBi, t−1 + λ2 Ln YIndo, t−1 + λ3 Ln YUS, t−1
i=0
+ λ4 Ln REXt−1 + μ
(2)
Since our emphasis is on the effects of REX, the short-run effects are judged
by the estimates of ek ’s and the long-run effects by the estimate of λ4 normal-
ized on λ1 .6 However, for the long-run effects to be valid and not spurious,
Pesaran et al. (2001) propose applying an F-test to determine the joint sig-
nificance of lagged level variables in (2) as a sign of cointegration among
all four variables. The F-test that they propose has new critical values that
they tabulate for large samples and Narayan (2005) does it for small sam-
ples. An upper bound critical value is provided when all variables in a given
model are I(1) and a lower bound critical value is provided when they are
all I(0). They then demonstrate that the upper bound critical value could be
used even if some variables are I(1) and some are I(0). Since most macro
variables are either I(1) or I(0), under this approach there is really no need
to test for unit root and this is one of the main advantages of this approach.7
6
For details of the normalization procedure, see Bahmani-Oskooee and Tanku (2008).
7
For some other applications of this approach, see Bahmani-Oskooee et al. (2005), Bahmani-
Oskooee and Hegerty (2007), Halicioglu (2007), Narayan et al. (2007), Tang (2007), Mohammadi et al.
(2008), Wong and Tang (2008), De Vita and Kyaw (2008), Payne (2008), Bahmani-Oskooee and Gelan
(2009), Shahbaz et al. (2012), and Chen and Chen (2012).
The J-Curve 107
As the Appendix shows, annual data over the period 1973-2011 are used to
estimate specification (2) for each of the 23 industries that trades between the
U.S. and Indonesia. Following the literature, we impose a maximum of four
lags on each of the first difference variables and use Akaike’s Information
Criteria (AIC) to choose the optimum lags. The results are reported in
Tables 1 and 2.
For brevity of presentation, we only report the short-run coefficient
estimates of the real bilateral exchange rate in Table 1. However, long-run
coefficient estimates are reported for all variables. From the short-run esti-
mates, we gather that there are 18 industries in which there is at least one
significant short-run coefficient (at the 10% level of significance), implying
that most industries’ trade balances are affected by exchange rate changes
in the short run. If we follow the traditional definition of the J-Curve (i.e.,
negative short-run coefficients at the early lags followed by positive ones at
higher lags), we find no J-Curve effects. However, if we follow an alterna-
tive definition by Rose and Yellen (1989, 67)—i.e., negative short-run effects
followed by positive long-run effects—then clearly we find support for the
phenomenon in nine industries. In these industries, which are coded as 031,
541, 653, 666, 812, 821, 894, 899, and 931, the real exchange rate carries posi-
tive and significant coefficient (again, at the 10% significance level). Whereas
real depreciation of the dollar does not seem to have long-run effects on the
aggregate bilateral trade balance between the U.S. and Indonesia, disaggre-
gation by commodity proves to be useful in identifying nine commodities
that will benefit from real depreciation in the long run. Note that all of these
industries happen to be small and the two largest industries—i.e., industries
coded 231 (crude rubber) and 841 (clothing except fur clothing)—do not
seem to react to exchange rate changes in the long run.
As for the other long-run coefficient estimates, the results reveal that the
level of economic activity in Indonesia has significant impact on the trade
balance of six industries. While in three industries coded 031, 292, 332 the
effect is expectedly positive, in three industries coded 653, 656, and 666 it
is negative. The negative coefficient obtained for YIND implies that as the
Indonesian economy grows, it produces more of import substitute goods in
these industries and therefore imports less of these commodities from the
U.S. The U.S. economic activity (YUS ) does not seem to have much of an
impact since it only carries significant coefficient in two industries (541 and
656).8
8
Note that we included a dummy variable in our specification to evaluate the impact of the Asian
financial crisis in 1997. The dummy carried a significant coefficient in eight industries coded 031, 231, 332,
653, 666, 729, 821, and 894, suggesting that the Asian financial crisis had an impact on U.S.-Indonesia’s
trade balances for these industries.
TABLE 1 Short-Run and Long-Run Coefficient Estimates
Short-Run Estimates Long Run-Estimates
SITC Industry T. Share In REXt In REXt-1 In REXt-2 In REXt-3 Constant In YINDO Ln YUS Ln REX
Aggregate −0.04 (0.14) 0.74 (2.63) 0.83 (2.72) −0.43 (1.45) 20.23 (0.93) 1.98 (1.74) −5.97 (1.41) −0.04 (0.14)
031 Fish, fresh & simply preserved 3.26% 2.77 (3.69) −0.30 (0.34) 0.94 (1.07) −0.07 (1.25) 47.74 (1.51) 3.14 (2.06) −9.45 (1.58) 1.37 (1.81)
121 Tobacco, unmanufactured 0.26% 2.21 (3.77) 42.89 (1.29) 1.88 (1.27) −6.29 (1.10) 1.23 (1.35)
231 Crude rubber incl. synthetic & recl. 11.46% 0.38 (1.68) −7.24 (0.54) −0.48 (0.44) 1.67 (0.61) 0.68 (1.27)
292 Crude vegetable materials, n.e.s. 0.05% −0.06 (0.20 −0.43 (1.11) −0.81 (2.35) −0.82 (2.49) 11.21 (0.80) 1.86 (3.23) −3.07 (1.42) 0.52 (0.80)
332 Petroleum products 0.74% 2.59 (3.29) 0.73 (0.86) 2.49 (2.79) −8.12 (0.45) 1.37 (1.70) −0.46 (0.14) 0.92 (1.61)
422 Other fixed vegetable oils 1.39% −1.23 (1.30) −85.98 (1.18) −7.29 (1.47) 21.40 (1.37) 1.09 (0.66)
541 Medicinal & pharmaceutical 0.23% −0.09 (0.15) −1.88 (2.36) 0.07 (0.09) −2.03 (2.73) −22.29 (1.42) −0.99 (0.90) 6.34 (1.69) 2.03 (2.24)
products
551 Essential oils, perfume and flavor 0.28% 1.55 (3.66) 1.08 (2.07) −36.44 (1.50) 0.34 (0.26) 3.53 (0.79) −0.93 (1.38)
632 Wood manufactures, n.e.s. 0.39% −1.98 (1.49) 168.35 (1.23) −6.08 (1.00) 29.56 (1.25) 1.84 (0.81)
653 Text fabrics woven 0.08% 3.39 (3.09) −0.12 (0.09) −2.07 (1.55) −1.23 (1.22) −7.21 (0.11) −5.70 (1.77) 14.08 (1.12) 4.91 (2.71)
656 Made up articles, wholly or chiefly 0.18% 1.54 (2.12) −19.49 (0.87) −4.53 (3.66) 8.28 (2.06) −0.18 (0.23)
657 Floor coverings, tapestries, etc. 0.07% 2.55 (1.67) 4.16 (2.42) 1.01 (0.51) −4.10 (2.41) −11.24 (0.03) −14.96 (0.49) 7.33 (0.09) −16.51 (0.88)
666 Pottery 0.22% 1.98 (2.03) 43.29 (1.38) −4.23 (2.73) 6.13 (1.04) 5.12 (5.31)
108
729 Other electrical machinery and 1.50% −0.36 (0.82) −1.06 (2.08) −0.86 (1.83) −0.79 (1.81) −12.21 (0.72) −0.26 (0.38) 1.98 (0.64) 0.23 (0.33)
apparatus
812 Sanitary, plumbing, heating & light 0.10% 0.42 (0.37) 95.81 (2.33) 0.38 (0.18) −6.32 (0.83) 4.98 (3.79)
821 Furniture 2.32% 1.40 (2.72) 70.45 (3.55) 0.15 (0.12) −4.13 (1.13) 4.47 (6.38)
841 Clothing except fur clothing 19.97% 0.38 (0.56) −23.77 (0.39) −1.29 (0.40) −0.51 (0.04) −4.08 (1.31)
892 Printed matter 0.08% 0.83 (0.89) 132.67 (1.31) 6.72 (1.24) −26.63 (1.32) 1.66 (0.72)
894 Perambulatorsm toys, games and 0.88% 1.91 (2.23) 0.35 (0.32) 0.36 (0.39) −1.44 (1.64) 68.45 (6.99) −1.39 (0.42) 1.26 (0.09) 7.30 (3.23)
sport
896 Works of art, collectors pieces 0.03% 1.68 (1.21) 7.47 (4.16) 5.58 (3.23) 2.30 (1.57) 80.16 (1.37) 1.26 (0.65) −15.92 (1.55) −5.33 (2.09)
897 Jewellery and gold/silver smiths 0.51% −0.73 (0.78) 2.28 (1.97) 1.42 (1.01) −1.47 (1.12) −799.66 (0.55) −21.91 (0.54) 122.49 (0.55) 1.71 (0.14)
watches
899 Manufactured articles, n.e.s. 0.96% 1.83 (2.93) 56.19 (2.01) 0.79 (0.52) −5.10 (0.97) 2.64 (3.05)
931 Special transactions not classd. 4.77% 1.06 (1.78) 97.35 (1.46) −3.88 (1.27) −13.88 (1.27) 5.54 (2.32)
according to kind
Notes:
1. Number inside the parenthesis next to each coefficient is absolute value of the t-ratio.
2. T. Share (Trade share) of each industry is calculated as sum of exports and imports of that industry as percent of sum of total U.S. exports and imports to Indonesia, which
includes industries for which no data were available. These shares are only for 2011.
TABLE 2 Diagnostic Statistics
SITC Industry F at opt. lags ECMt-1 LM RESET CUSUM CUSUMSQ Adj R2
Aggregate 0.61 −0.39 (2.62) 0.12 1.46 S S 0.59
031 Fish, fresh & simply preserved 1.34 −0.89 (3.39) 0.02 2.05 S US 0.41
121 Tobacco, unmanufactured 1.33 −0.51 (2.40) 0.56 0.04 S S 0.45
231 Crude rubber incl. synthetic & recl. 1.77 −0.55 (2.99) 0.98 2.26 S S 0.07
292 Crude vegetable materials, n.e.s. 6.92 −0.58 (4.25) 2.59 1.11 S S −0.01
332 Petroleum products 1.24 −1.48 (3.98) 0.01 0.17 S S 0.55
422 Other fixed vegetable oils 2.85 −0.46 (2.09) 2.04 0.25 S S 0.46
541 Medicinal & pharmaceutical products 0.76 −0.76 (2.82) 0.83 2.19 S S 0.37
551 Essential oils, perfume and flavor 3.19 −0.45 (2.44) 0.04 0.83 S S 0.58
632 Wood manufactures, n.e.s. 1.67 0.39 (2.01) 5.29 0.33 S S 0.33
653 Text fabrics woven 1.29 −0.57 (3.47) 2.95 16.02 S S 0.32
656 Made up articles, wholly or chiefly 6.02 −1.18 (4.97) 0.29 0.15 S S 0.66
657 Floor coverings, tapestries, etc. 3.88 0.15 (0.71) 0.37 0.99 S S 0.62
666 Pottery 3.47 −0.73 (4.95) 0.62 0.05 S S 0.16
729 Other electrical machinery and apparatus 1.23 (4.23) 0.01 0.26 S S
109
−0.67 −0.21
812 Sanitary, plumbing, heating & light 1.98 −0.62 (4.00) 5.30 1.38 S US 0.15
821 Furniture 1.29 −0.51 (3.37) 3.45 0.58 S S 0.32
841 Clothing except fur clothing 9.89 −0.35 (2.75) 9.64 2.97 S S 0.19
892 Printed matter 2.06 −0.34 (2.28) 3.49 7.05 S US 0.05
894 Perambulators, toys, games and sports 1.37 −0.38 (2.99) 0.04 0.13 S S 0.37
896 Works of art, collectors pieces 9.44 −0.71 (3.03) 10.79 0.27 S S 0.31
897 Jewellery and gold/silver smiths watches 1.29 −0.09 (0.62) 1.13 5.87 S US 0.26
899 Manufactured articles, n.e.s. 3.48 −0.69 (4.34) 3.65 0.55 S US −0.01
931 Special transactions not classd. according to kind 3.67 −0.34 (2.26) 5.24 0.88 S S 0.10
Notes:
1. LM: Lagrange multiplier test of residual serial correlation. It is distributed as χ 2 (1).
2. RESET: Ramsey’s test for functional form. It is distributed as χ 2 (1).
3. CUSUM: Cumulative Sum of Recursive Residuals.
4. CUSUMSQ: Cumulative Sum of Squares of Recursive Residuals.
5. Number inside the parenthesis next to a coefficient is absolute value of the t-ratio.
110 M. Bahmani-Oskooee and H. Harvey
Research on the relation between the trade balance and the real exchange
rate still continues by considering each country’s experience separately.
Studies that examined the Marshall-Lerner condition did not include
Indonesia in their sample. Those that relied upon establishing a direct link
between the trade balance and the real exchange rate using cointegration
analysis did not find any significant relationship either using aggregate trade
flows between Indonesia and rest of the world or between Indonesia and
most of her trading partners.
9
For the origin of the CUSUM and CUSUMSQ tests, see Brown et al. (1975) and for a sample that
represents industry 841, see Figure 1.
The J-Curve 111
20
10
–10
–20
1978 1987 1996 2005 2011
The straight lines represent critical bounds at 5% significance level
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
–0.2
–0.4
1978 1987 1996 2005 2011
The straight lines represent critical bounds at 5% significance level
REFERENCES
Chen, S.-W., and Chen, T-C. 2012. Untangling the Non-Linear Causal Nexus between
Exchange Rates and Stock Prices: New Evidence from the OECD Countries.
Journal of Economic Studies 39(2): 231–259.
De Vita, G., and Kyaw, K. S. 2008. Determinants of Capital Flows to Developing
Countries: A Structural VAR Analysis. Journal of Economic Studies 35(4):
304–322.
Du, H., and Zhu, Z. 2001. The Effect of Exchange-Rate Risk on Exports: Some
Additional Empirical Evidence. Journal of Economic Studies 28(2): 106–121.
Gylfason, T., and Risager, O. 1984. Does Devaluation Improve the Current Account?
European Economic Review 25(1): 37–64.
Halicioglu, F. 2007. The J-Curve Dynamics of Turkish Bilateral Trade: A Cointegration
Approach. Journal of Economic Studies 34(2): 103–119.
Houthakker, H. S., and Magee, S. P. 1969. Income and Price Elasticities in World
Trade. Review of Economics and Statistics 51(2): 111–25.
Khan, M. S. 1974. Import and Export Demand in Developing Countries. IMF Staff
Papers 21(3): 678–693.
King, A. 1993. The Functional Form of Import Demand: The Case of U.K. Motor
Vehicle Imports, 1980-90. Journal of Economic Studies 20(3): 36–50.
Lal, A. K., and Lowinger, T. C. 2002. The J-Curve: Evidence from East Asia. Journal
of Economic Integration 17(2): 397–415.
Love, J., and Chandra, R. 2005. Testing Export-Led Growth in South Asia. Journal of
Economic Studies 32(2): 132–145.
Magee, S. P. 1973. Currency Contracts, Pass Through and Devaluation. Brookings
Papers on Economic Activity 1: 303–25.
Mohammadi, H., Cak, M., and Cak, D. 2008. Wagner’s Hypothesis: New Evidence
from Turkey Using the Bounds Testing Approach. Journal of Economic Studies
35(1): 94–106.
Narayan, P. K. 2005. The Saving and Investment Nexus for China: Evidence from
Cointegration Test. Applied Economics 37(17): 1979–1990.
Narayan, P. K., Narayan, S., Prasad, B. C., and Prasad, A. 2007. Export-Led Growth
Hypothesis: Evidence from Papua New Guinea and Fiji. Journal of Economic
Studies 34(4): 341–351.
Narayan, S., and Narayan, P. K. 2005. An Empirical Analysis of Fiji’s Import Demand
Function. Journal of Economic Studies 32(2): 158–168.
Payne, J. E. 2008. Inflation and Inflation Uncertainty: Evidence from the Caribbean
Region. Journal of Economic Studies 35(6): 501–511.
Pesaran, M. H., Shin, Y., and Smith, R. J. 2001. Bound Testing Approaches to the
Analysis of Level Relationship. Journal of Applied Econometrics 16(3): 289–326.
Rose, A. K. 1991. The Role of Exchange Rates in a Popular Model of International
Trade: Does the “Marshall-Lerner” Condition Hold? Journal of International
Economics 30(3–4): 301–316.
Rose, A. K., and Yellen, J. L. 1989. Is There a J-Curve? Journal of Monetary Economics
24(1): 53–68.
Shahbaz, M., Jalil, A., and Islam, F. 2012. Real Exchange Rate Changes and the Trade
Balance: The Evidence from Pakistan. The International Trade Journal 26(2):
139–153.
Sweidan, O. D. 2013. The Effect of Exchange Rate on Exports and Imports: The Case
of Jordan. The International Trade Journal 27(2): 156–172.
114 M. Bahmani-Oskooee and H. Harvey
APPENDIX
Data Definition and Sources
Empirical analysis is based on annual data over the period 1973–2011. The
following sources are used to collect the necessary data:
1. World Bank
2. International Financial Statistics of the IMF
Variables
TBi = measure of the trade balance for industry i. It is defined as the ratio of
U.S. exports of commodity i to Indonesia over its imports of commodity i
from Indonesia. Industry-level trade flows come from source 1.
YINDO = Indonesia’s real GDP. Data come from source 2.
YUS = US real GDP. Data come from source 2.
REX = Real bilateral exchange rate between U.S. dollar and Indonesian
rupiah. It is defined as (PINDO • NEX/PUS ), where PUS is US CPI, PINDO is
Indonesia’s CPI, and NEX is the nominal bilateral exchange rate defined
as the number of U.S. dollars per rupiah. Thus, an increase in REX is a
reflection of real depreciation of the U.S. dollar. All CPI data and nominal
exchange rate data come from source 2.
Copyright of International Trade Journal is the property of Routledge and its content may not
be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.