Trade-led growth hypothesis: An empirical analysis of South Asian countries

Economic Modelling 35 (2013) 654–660

Contents lists available at ScienceDirect

Economic Modelling journal homepage: www.elsevier.com/locate/ecmod

Trade-led growth hypothesis: An empirical analysis of South Asian countries Qazi Muhammad Adnan Hye a,⁎, Shahida Wizarat b, Wee-Yeap Lau c a b c

Faculty of Economics and Administration, University of Malaya, Malaysia Economics Department, Institute of Business Management, Karachi, Pakistan Department of Applied Statistics, Faculty of Economics and Administration, University of Malaya, Malaysia

a r t i c l e Article history: Accepted 31 July 2013 JEL classification: F10 E23 Keywords: Exports Imports Economic growth

i n f o

a b s t r a c t In this paper we examine trade-growth nexus using data from six Asian countries. We apply the ADF unit root test to check for stationarity; and the autoregressive distributed lag (ARDL) approach for a long-run relationship among exports, imports and economic growth. For the direction of causality, we implement the modified Granger Causality test. We find that the export-led growth model is relevant to all countries except Pakistan, while the import-led growth model is relevant to all countries. The growth-led export model applies to all countries except Bangladesh and Nepal. The growth-led import model and export–import model are relevant for all countries in the sample. The results show that domestic and overseas demand contribute to economic growth and employment generation. Growth accruals through import-led model augur well for the countries studied. Our findings point to the potential for growth through tapping domestic demand in the event of global recession. It also appears that opportunities of joint catering for domestic demand through south–south trade expansion are possible. © 2013 Elsevier B.V. All rights reserved.

1. Introduction The theoretical and empirical link between international trade and economic growth has been investigated at length by academicians. The neoclassical economists point to the strong association between trade expansion and economic growth. They argue that export growth is the main driver of economic growth. Articulating this world view, Helpman and Krugman (1985) posit that export growth expedites economic growth through economies of scale – specialization in production and dissemination of technical knowledge. Bhagwati (1988), in supporting the neoclassical trade theory, proposed the growth led export hypothesis where he stated that economic growth stimulates both supply and demand sides of the economy. Bhagwati (1988) also noted that export growth promotes economic growth and in turn, economic growth promotes skill formation as well as technological progress. Both factors add to productive efficiency, and thus create a comparative advantage for the country. Easterly (2007) argues that exports help entry to the international market and expansion of the manufacturing sector. In addition exports boost economic efficiency through better allocation of resources and promote economic growth in the long run. According to Stiglitz (2007), the rapid growth of China and India is predominantly due to expansion of trade, mostly exports. The extant empirical literature can be divided into three strands. Some use cross-country data and rank correlation method to test the Export-Led Growth (ELG) hypothesis (see Heller and Porter, 1978; ⁎ Corresponding author. E-mail addresses: [email protected] (Q.M.A. Hye), [email protected] (S. Wizarat), [email protected] (W.-Y. Lau). 0264-9993/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.econmod.2013.07.040

Kravis, 1970; Maizels, 1963; Rana, 1986; Tyler, 1981), among others. The second group also uses cross-country data with regression technique e.g., OLS, 2SLS, 3SLS, Models; and panel data methods (SeeAlam, 1991; Amirkhalkhali and Dar, 1995; Balassa, 1985; Coppin, 1994; De Gregorio, 1992; Dodaro, 1991; Emery, 1967; Fosu, 1990, 1996; Mbaku, 1989; McNab and Moore, 1998; Michalopoulos and Jay, 1973; Otani and Villaneuva, 1990; Ram, 1985; Rana, 1986; Salvatore, 1983; Sheehey, 1992; Singer and Gray, 1988; Sprout and Weaver, 1993; Tyler, 1981; Voivodas, 1973; Williamson, 1978; Yaghmaian and Ghorashi, 1995). The last group uses time series techniques to examine the exportgrowth nexus. For example Shan and Sun (1998) study the export-led growth hypothesis for China using monthly data and found bidirectional causality between export growth and economic growth. Ramos (2001) examines the association between export, import and GDP growth for Portugal using multivariate Johansen–Juselius (JJ) approach. He found bidirectional causality between GDP and exports, GDP and imports, imports and exports growth. Narayan and Smyth (2004) used co-integration and error correction model to examine the link between real exports, human capital accumulation and economic growth. The authors found a long-run relationship only if they included real exports as the dependent variable. Mah (2005) tested the autoregressive distributed lag (ARDL) model and found a long-run relationship and bidirectional causality between real GDP growth and export growth. In addition to data on term of trade, Awokuse (2005) used change in capital and foreign output shocks to Korean quarterly data. He used vector error correction model (VECM) and the augmented vector autoregressive (VAR) procedures to test causality. The results confirm the ELG and GLE hypothesis. The results show that capital, terms of trade and foreign output shocks influence economic growth.

Q.M.A. Hye et al. / Economic Modelling 35 (2013) 654–660

Trade and Growth in Bangladesh

655

Table 1 ADF unit root test results. Statistical level

Economic Growth

τT (ADF)

τμ (ADF)

Bangladesh Ln(Y) −0.431 Ln(X) −1.515 Ln(M) −2.742

Exports Imports

Fig. 1. Trade and growth in Bangladesh.

In addition, Tang (2006) reviewed the ELG hypothesis for China using imports as an additional variable. He used both the ARDL and the JJ approaches. The results failed to find co-integration between exports, imports and real GDP. Awokuse (2007) further tested the link between exports, imports and GDP using the Granger causality approach. He finds support for the ELG and the GLE hypotheses for Bulgaria; unidirectional causality from export and import to GDP growth for the Czech Republic; the ILG model for Poland. The variance decomposition and impulse response functions support the findings. Subsequently, Herrerias and Orts (2009) explore the relationship between imports, investment, output and productivity in China. They show that in the long run both imports and investment stimulate output and labor productivity, but do not find causality between investment and imports. Katircioglu et al. (2010) employ the ARDL approach to examine a long-run relationship between trade and economic growth for Fiji Island, Papua New Guinea and Solomon Islands. They find that real income stimulates export growth in Fiji, but the ELG and ILG hypotheses cannot be validated for the Islands and the Pacific region. Hye and Siddiqui (2010) employ the rolling window bounds testing and variance decomposition methods to test foreign debt sustainability hypothesis for Pakistan. They conclude that imports cannot cause exports' but exports effectively cause imports; so foreign debt is unstable. Lee (2010) finds support for ELG, GLE, ILG and growth-led imports (GLI) in the short run only, but in the long run he does not find support for ELG and ILG for Pakistan. Shahbaz et al. (2011) employ the ARDL bounds testing approach to cointegration and error correction method (ECM) with quarterly data from Pakistan. They find that exports are positively correlated with economic growth. Hye and Siddiqui (2011) further examine the relationship between exports, terms of trade and economic growth in Pakistan using ARDL with rolling regression method. They document that exports enhance economic growth, but adverse terms of trade hurt economic growth. They argue that it is necessary to promote exports in order to improve the terms of trade.1 Hye (2012) further tests the ELG, GLE, ILG, GLI and foreign deficit sustainability hypothesis in China using annual data. The results confirm a long-run relationship between economic growth and exports, economic growth and imports, and exports and imports. They are in favor of the validity of ELG, GLE, ILG and GLI hypotheses. He also argues that foreign deficit is sustainable for China. Dar et al. (2013) apply wavelets based correlation and cross correlation methodology to Indian monthly data and find positive association between export growth and output growth. They also find that this relationship gets stronger as the time horizons expand. They also find bidirectional causality. Al-Khulaifi (2013) investigates a long-run relationship between exports and imports 1 Related research on India includes Mallick (1996), Chandra (2002, 2003), Love and Chandra (2004, 2005)

Statistics (1st difference) τ (ADF)

3.021 0.217 3.021

7.014 6.798 7.014

τT (ADF)

τμ (ADF)

τ (ADF)

−11.891⁎ −6.311⁎ −4.692⁎

−6.951⁎ −5.989⁎ −5.153⁎

−3.688⁎ −3.088⁎ −4.447⁎

Bhutan Ln(Y) Ln(X) Ln(M)

−1.989 −2.703 −1.075

−0.151 −1.178 −0.433

10.238 2.029 2.634

−3.834⁎⁎ −3.675⁎⁎ −6.143⁎

−3.912⁎ −2.766⁎⁎ −6.264⁎

−1.527⁎⁎⁎ −1.759⁎⁎⁎ −3.929⁎

India Ln(Y) Ln(X) Ln(M)

−0.193 −0.592 −1.601

3.164 1.805 1.821

11.621 5.183 4.684

−7.685⁎ −2.464⁎⁎ −2.877⁎⁎

−6.314⁎ −4.885⁎ −7.095⁎

−2.504⁎⁎ −5.491⁎ −7.619⁎

Nepal Ln(Y) Ln(X) Ln(M)

−2.171 −1.714 −2.642

1.437 −0.949 −0.424

9.114 −0.949 3.444

−6.501⁎ −3.854⁎⁎ −6.521⁎

−5.514⁎ −3.904⁎ −6.587⁎

−1.639⁎⁎⁎ −3.331⁎⁎ −3.908⁎

Pakistan Ln(Y) −0.678 Ln(X) −2.889 Ln(M) −0.661

−1.299 −0.196 −0.661

4.633 2.592 2.067

−5.524⁎ −6.488⁎ −3.386⁎⁎⁎

−5.125⁎ −6.628⁎ −3.559⁎⁎

−1.401 −5.441⁎ −3.047⁎

Sri Lanka Ln(Y) −2.399 Ln(X) −1.467 Ln(M) −1.901

0.663 −0.923 −1.222

4.029 2.668 2.094

−5.638⁎ −4.601⁎ −4.779⁎

−5.586⁎ −4.636⁎ −4.803⁎

−1.715⁎⁎⁎ −3.699⁎ −4.041⁎

Note: Y represents the real gross domestic product, X represents real exports, M represents real imports. The series are in natural logarithm form. τT presents the model with a drift and trend; τμ the model with drift but without trend, while τ is the model without drift and trend.

for Qatar using annual data. He employs JJ cointegration for long-run relationship and Granger causality test to determine the direction of causality. He finds co-integration between exports and imports, and international budget constraints in Qatar are not violated. The Granger causality results also indicate that imports cause exports in the long run. 2. Methodology This study uses annual data from 1971 to 2009 for Pakistan and Bangladesh; 1960–2009 for India and Sri Lanka; 1965–2009 for Nepal and 1981–2009 for Bhutan. These data are from the World Bank online database (http://data.worldbank.org/data-catalog/world-developmentindicators). The series, gross domestic product (GDP), export (X) and import (M) of goods and services are in constant 2000 dollars.2 The ELG and GLE hypotheses are examined using causal link between economic growth and exports (Eqs. (1) & (2)).3 Y t ¼ α 0 þ α 1 X t þ ψt

ð1Þ

X t ¼ β0 þ β1 Y t þ νt

ð2Þ

The causality between imports and economic growth is explored by the import-led growth and growth-led import hypothesis (Eqs. (3) & (4)).4 Y t ¼ θ0 þ θ1 Mt þ ψt 2

ð3Þ

The variables are transformed into natural logarithm, for estimation purposes. The export-led growth hypothesis suggests that export growth enhances economic growth through economies of scale by specialization in production and promoting the distribution of technical knowledge (Helpman and Krugman, 1985). All of these are discussed in review of literature. 4 In the same way Deme (2002), Sato and Fukushige (2007), Katircioglu et al. (2010) have estimated the ILG and GLI hypothesis. 3

656

Q.M.A. Hye et al. / Economic Modelling 35 (2013) 654–660

Table 2 Bound testing analysis and long-run coefficients. Variable

With deterministic trends FIV

Bangladesh 1) Y and X FX (InY/lnX) FX (InX/lnY). 2) Y and M FY (InY/lnM) FM (InM/lnY) 3) X and M FX (InX/lnM) FM (InM/lnX) Bhutan 1) Y and X FX (In Y/lnX) FX (In X/lnY) 2) Y and M FY (lnY/lnM) FM (lnM/lnY) 3) X and M FX (lnX/lnM) FM (lnM/lnX) India 1) Y and X FX (ln Y/lnX) FX (ln X/lnY) 2) Y and M FY (ln X/lnM) FM (ln M/lnY) 3) X and M FX (ln X/lnM) FM (ln M/lnX) Nepal 1) Y and X FX (ln Y/lnX) FX (ln X/lnY) 2) Y and M FY (ln Y/lnM) FM (ln M/lnY) 3) X and M FX (ln X/lnM) FM (ln M/lnX) Pakistan 1) Y and X FX (ln Y/lnX) FX (ln X/lnY) 2) Y and M FX (ln Y/lnM) FM (ln M/lnY) 3) X and M FX (ln X/lnM) FM (ln M/lnX) Sri Lanka 1) Y and X FM (ln M/lnX) FX (ln X/lnY) 2) Y and M FY (ln Y/lnM) FM (ln M/lnY) 3) X and M FX (ln X/lnM) FM (ln M/lnX) a b c

FV

Without deterministic trends

Conclusion

tv

FIII

tIII

(H0)

Long-run estimated equation

8.123a 2.218c

4.552b 3.325c

−1.227c −2.326c

12.635a 3.206b

−1.355c −2.361c

Rejected Inconclusive

LnY = 11.651⁎⁎⁎ + 0.605⁎⁎LnX –

11.583a 17.477a

1.654c 25.519a

1.351c −6.734a

17.812a 27.077a

2.157c −7.307a

Rejected Rejected

LnY = 22.247⁎⁎⁎ + 0.091⁎⁎⁎LnM LnM = −16.109⁎⁎⁎ + 1.595⁎⁎⁎LnY

5.527a 19.323a

6.927a 25.354a

−3.555b −6.961a

5.824a 29.863a

−3.377a −7.309a

Rejected Rejected

LnX = 3.411 + 0.793⁎⁎⁎LnM LnM = 7.249⁎ + 0.697⁎⁎⁎LnX

4.066a 7.078a

5.933a 10.598a

−2.441c −4.466a

0.621c 11.441a

1.113c −4.781a

Rejected Rejected

LnY = 16.326⁎⁎⁎ + 0.138⁎⁎LnX LnX = −10.231⁎⁎⁎ + 1.464⁎⁎⁎ LnY

2.927c 6.751a

4.116c 10.117a

−2.793c −4.411a

4.752a 5.155a

−2.993b −3.252a

Rejected Rejected

LnY =2.051⁎ + 0.944⁎⁎⁎LnM LnM = −1.711⁎⁎⁎ + 1.052⁎⁎⁎ LnY

3.381c 11.546a

4.925b 17.316a

−3.077b −5.716a

5.362a 1.228c

−3.235a −1.564c

Rejected Rejected

LnX = -5.559⁎⁎⁎ + 1.283⁎⁎⁎ LnM LnM =7.427⁎⁎ + 0.635⁎⁎⁎ LnX

7.184a 2.386c

0.942c 2.284c

−1.031c −2.125c

11.058a 3.545b

−0.453c −2.098b

Rejected Inconclusive

LnY = 9.809⁎⁎⁎ + 0.727⁎⁎⁎LnX –

5.816a 3.886b

4.342b 3.708c

−2.592b −2.677c

5.616a 5.115a

−1.329c −2.623b

Rejected Rejected

LnY = 8.734⁎⁎⁎ + 0.771⁎⁎⁎LnM LnM = −24.994⁎⁎⁎ + 1.871⁎⁎⁎ LnY

1.787c 3.073c

1.475c 2.981c

−1.708c −2.443c

2.603c 4.683a

−1.264c −2.484c

Accepted Rejected

– LnM = −0.891 + 1.048⁎⁎⁎ LnX

7.089a 3.493b

3.726c 4.831b

−2.712c −3.107b

5.617a 5.249a

−0.739c −3.107b

Rejected Rejected

LnY = 4.917 + 0.929⁎⁎⁎ LnX LnX = −4.917 + 1.231⁎⁎⁎ LnY

8.681a 9.267a

6.761a 13.246a

−2.577c −3.987a

12.509a 2.347c

−3.129b −2.159c

Rejected Rejected

LnY = 8.776⁎⁎⁎ + 0.681⁎⁎⁎ LnM LnM = −19.369⁎⁎⁎ + 1.831⁎⁎⁎ LnY

6.928a 6.702a

10.204a 9.986a

−4.218a −4.283a

4.165a 1.678c

−2.867b 1.522c

Rejected Rejected

LnX = 1.867 + 0.885⁎⁎⁎ LnM LnM = 10.794⁎⁎⁎ + 0.425⁎⁎⁎ LnX

2.444c 5.042a

1.691c 6.926a

−1.821c −3.681a

3.056c 7.445a

−1.584c −3.671a

Accepted Rejected

– LnX = −20.977 + 1.785⁎ LnY

1.952c 5.812a

1.477c 8.642a

−1.148c −4.093a

1.912c 8.172a

−1.221c −4.031a

Accepted Rejected

– LnM = 6.964⁎⁎⁎ + 0.645⁎⁎⁎ LnY

2.447c 5.581a

3.637c 8.321a

−2.535c −4.068a

0.901c 7.446a

−1.061c −3.851a

Accepted Rejected

– LnM = 12.889⁎⁎⁎ + 0.447⁎⁎⁎ LnX

9.573a 2.474c

13.791a 3.235c

−5.213a −1.533c

1.007c 2.106c

−1.093c −1.765c

Rejected Accepted

LnY = 1.742⁎⁎⁎ + 1.117⁎⁎⁎ LnX

5.871a 3.619b

7.924a 4.001c

−3.981a −1.051c

0.706c 2.137c

−0.393c −1.206c

Rejected Inconclusive

LnY = 344.001 + 0.071⁎⁎⁎ LnM

2.186c 3.809b

3.011c 5.633a

−2.379c −3.095b

2.096c 5.825a

−2.022c −3.409a

Accepted Rejected

LnM = −0.784 + 1.048⁎⁎⁎ LnX

Indicates that the statistic value lies above the upper bound, that it falls within the lower and upper bounds and, that it lies below the lower bound.

Mt ¼ φ0 þ φ1 Y t þ ν t

ð4Þ

The theoretical link between import and export; and the current budget constraint are provided by Husted (1992). C t ¼ Y t þ Bt −It ð1 þ r ÞBt

ð5Þ

were, Ct, Yt, Bt, It and r refer to current consumption, output, investment and rate of interest in the international market respectively. Husted imposes enough structures and restrictions on Eq. (5) to obtain an empirically testable model, as follows:

X t ¼ λ0 þ λ1 Mt þ ψt

ð6Þ

Q.M.A. Hye et al. / Economic Modelling 35 (2013) 654–660

Trade and Growth in Bhutan

Trade and Growth in Nepal

Economic Growth

Economic Growth

Exports

Fig. 2. Trade and growth in Bhutan.

Fig. 4. Trade and growth in Nepal.

Arize (2002) examines alternatively, we rewrite Eq. (6) as follows: Mt ¼ γ 0 þ γ 1 X t þ νt

ð7Þ

The international budget constraint is stable when co-integration between X and M, and their relevant coefficients are equal to or greater than one (See, Husted, 1992; Hye and Boubaker, 2011; Hye, 2012; Wu and Zhang, 1998). In the present study, we use the ADF unit root test to determine the order of integration. For a long-run association between each pair of series, the autoregressive distributed lag (ARDL) approach to cointegration is implemented. The approach offers several advantages over the traditional methods. It applies irrespective of the order of integration of the regressors, I(0), purely I(1) or mutually cointegrated (Pesaran et al. (2001). The ARDL approach has better small sample properties; and involves estimating the following error correction models. ρ X

θi ΔLnðRÞt−i þ

ρ X

i¼1

ΔLnðSÞt ¼ γ0 þ

ρ X i¼1

LnðRÞt ¼ β0 þ

ρ X

φ1 j LnðRÞt− j þ

λi ΔLnðSÞt−i þ σ 1 LnðRÞt−1

ð8Þ LnðSÞt ¼ σ 0 þ

ρ X

γ i ΔLnðRÞt−i þ δ1 LnðSÞt−1

ð9Þ

i¼0

The variable Ln(R) is the natural logarithm of dependent variable, while Ln(S) is the natural logarithm of the independent variable in Eq. (8) and vice versa. The v1t and v2t are the error terms. The F- and the t-statistic are used to determine the existence of a long-run

Trade and Growth in India

β1 j St− j þ η1t

ð10Þ

ρ X

φ2 j LnðSÞt− j þ

P X

β2 j Rt− j þ η2t

ð11Þ

j¼0

The modified Granger causality test is used to determine the shortrun and long-run direction of causality. In Eq. (12) the modified Granger causality test is as follows: 

ΔRt ΔSt



 ¼

 X       ρ  Γ1 ϖ1 n11i n12i n13i Δ lnRt−i ζ ½ECT t−1  þ 1 þ þ Γ2 n21i n22i n23i Δ lnSt−i ϖ2 ζ2 i¼1

ð12Þ ECTt-1 represents the error correction term, which is derived from the long-run equation. The short-run causality is tested using the statistical significance on the lagged differences of the variables for each vector (Fig. 1). 3. Empirical results

Economic Growth

Exports Imports

Fig. 3. Trade and growth in India.

P X j¼0

j¼1

ϖi ΔLnðSÞt−i þ

þ δ2 LnðRÞt−1 þ v2t

relationship. Pesaran et al. (2001) provide two sets of critical values for each of the 1%, 5% and 10% levels of significance [i.e. lower critical bound (LCB) for I(0) and upper critical bound (UCB) for I(1) series]. However, we use the Narayan (2005) critical values because of our small sample. If the calculated F-statistic exceeds the UCB, then the H0 is rejected and we conclude in favor of a long-run relationship. If the F-statistic falls below the LCB, the null of no co-integration is sustained. If the F statistics fall between the two bounds, the results are inconclusive. The null hypothesis for Eq. (8) is 〈H0 = σ1 = σ2 = 0〉 and for Eq. (9) it is 〈H0 =δ1 = δ2 = 0〉. However we also use the Pesaran et al. (2001) t-statistic to determine the long-run relationship. Most of the researchers use F statistic but we use both F and t here. The tstatistic is used to tests θi = 0 in Eq. (8) and ϖi = 0 in Eq. (9)5. Next we estimate ARDL based long-run coefficients:

j¼1

i¼0

þ σ 2 LnðSÞt−1 þ v1t

Imports

Exports

Imports

ΔLnðRÞt ¼ λ0 þ

657

The ADF unit root test results reported in Table 1 indicate that each series is in natural logarithm for economic growth (LnY), real exports (LnX) and real imports (LnM). Based on the result, these series are I(1) for South Asian countries. Table 2 reports the ARDL based long-run coefficients. As noted, three pairs of relationships are tested: (1) exports and economic growth; (2) imports and economic growth; and (3) epxorts and imports. In the case of Bangladesh the results support a long-run relationship in the first

5

The long-run decision method in t-statistic is the same as in F-statistic.

658

Q.M.A. Hye et al. / Economic Modelling 35 (2013) 654–660

Table 3 Granger Causality test.

Bangladesh 1) Y and X Ln (Y) Ln (X) 2) Y and M Ln (Y) Ln (M) 3) X and M Ln (X) Ln (M) Bhutan 1) Y and X Ln (Y) Ln (X) 2) Y and M Ln (Y) Ln (M) 3) X and M Ln (X) Ln (M) India 1) Y and X Ln (Y) Ln (X) 2) Y and M Ln (Y) Ln (M) 3) X and M Ln (X) Ln (M) Nepal 1) Y and X Ln (Y) Ln (X) 2) Y and M Ln (Y) Ln (M) 3) X and M Ln (X) Ln (M) Pakistan 1) Y and X Ln (Y) Ln (X) 2) Y and M Ln (Y) Ln (M) 3) X and M Ln (X) Ln (M) Sri Lanka 1) Y and X Ln (Y) Ln (X) 2) Y and M Ln (Y) Ln (M) 3) X and M Ln (X) Ln (M)

Short-run causality results

Long-run causality results

F-statistics

T-statistic (Pro.)

Decision of short-run causality

Decision of long-run causality

Ln (Y)

Ln (X)

Ln (M)

ECM(-1)

– 521 (0.235)

0.121 (0.887) –

– –

−5.088 (0.000) −0.249 (0.804)

No

Y←X

– 5.393 (0.010)

– –

0.399 (0.674) –

−4.986 (0.000) −3.141 (0.003)

Y→M

Y↔M

– –

– 3.891 (0.0314)

6.062 (0.006) –

−1.269 (0.213) −5.792 (0.000)

X←M

X→M

– 5.334 (0.013)

1.689 (0.209) –

– –

−1.772 (0.091) −4.185 (0.0000)

Y→X

Y↔X

– 1.375 (0.275)

– –

2.189 (0.138) –

−2.345 (0.029) −3.002 (0.007)

No

Y↔M

– –

0.849 (0.442) –

−3.272 (0.003) −3.456(0.002)

X→M

X↔M

8.221 (0.002)

– 571 (0.569)

1.881 (0.165) –

– –

−2.705 (0.009) −2.451 (0.018)

No

Y↔X

– –

1.706 (0.194) –

−3.982 (0.0000) −2.974 (0.004)

Y→M

Y↔X

2.701 (0.079)

– 0.671 (0.517)

0.512 (0.602) –

−1.724 (0.092) −2.824 (0.007)

No

X↔M

– 857 (0.432)

0.686 (0.509) –

– –

−4.452 (0.000) −1.580 (0.122)

No

Y↔X

– 519 (0.599)

– –

4.127 (0.024) –

−5131 (0.000) −2.019 (0.051)

Y←M

Y↔M

– –

– 2.331 (0.111)

1.498 (0.237) –

−2548 (0.015) −1.136 (0.263)

No

X←M

– 205493 (0.8154)

1.170554 (0.3240) –

– –

−1.13157 (0.26678) −3.45954 (0.00164)

No

Y↔X

– 3.414179 (0.0461)

– –

0.359218 (0.7012) –

−1.72446 (0.09492) −3.50866 (0.00144)

Y→M

Y↔M

– –

– 0.040842 (0.9600)

0.364047 (0.6979) –

−2.16366 (0.03858) −3.07634 (0.00444)

No

X↔M

– 534 (0.590)

7.927 (0.001) –

– –

−5787 (0.000) −2.155 (0.037)

Y←X

Y↔X

– 1.081 (0.348)

– –

2.905 (0.066) –

−4.342 (0.000) −2.160 (0.036)

Y←M

Y↔M

– –

– 0.219 (0.804)

0.017 (0.982) –

−1.107 (0.374) −2.348 (0.023)

No

X→M

combination when real GDP (LnY) is the dependent variable. A 1% increase in exports enhances economic growth by 0.61%. There exists a long-run relationship for the other two pairs viz., imports and economic growth; and exports and imports. A 1% increase in imports increases economic growth (measured by growth in real GDP) by 0.09%, and a 1% increase in economic growth increases imports by 1.60% and a 1%

increase in imports enhances exports by 0.79%. A 1% increase in exports boosts imports by 0.70% in Bangladesh on average. In addition, the sum for coefficients of exports and imports is not equal to or greater than one, as such, the international budget constraint is weakly stable. As for Bhutan, the ARDL results show that there are long-run relationships for all three pairs. The estimated long-run elasticities indicate that

Q.M.A. Hye et al. / Economic Modelling 35 (2013) 654–660

659

Trade and Growth in Pakistan

Trade and Growth in Sri Lanka

Economic Growth

Economic Growth

Exports

Exports Imports

Imports

Fig. 5. Trade and growth in Pakistan.

exports and imports are growth elastic, and exports are imports elastic as well.6 The international budget constraint is weakly stable (Fig. 2). The ARDL results for India show long-run relationship for the (Y–X) pair when economic growth is the dependent variable. However, when export is the dependent variable, long-run results are inconclusive. For the second pair (Y–M), a long-run relationship exists. For the third pair (X–M), a lon-run relationship is found only when export is a dependent variable. The long-run coefficients of the pairs 2 and 3 show that imports are economic growth elastic; and exports are import elastic. The international budget constraint is weakly stable (Fig. 3). For Nepal, we find a long-run relationship in all three pairs. The longrun coefficients indicate that exports and imports are growth elastic. In Pakistan, the results indicate that economic growth impacts exports and imports in the long run; exports are economic growth elastic, and affects imports in the long run. The international budget constraint is weakly stable (Fig. 4). In Sri Lanka, in the long run, exports and imports affect economic growth; and exports impact imports. The long-run estimates indicate that economic growth is export elastic, and imports are export elastic.7 The international budget constraint is weakly stable as the related coefficient are not equal to one. Table 3 shows the result of Granger's causality test. In Bangladesh, for the relationship between economic growth and export, it shows that only exports cause economic growth in the long run – validation of the ELG model. We find bidirectional long-run causality between imports and economic growth; and unidirectional link from growth to imports in the short run. Finally, exports Granger cause imports in the long run while in the short run, imports granger cause exports (Fig. 1). As for Bhutan and India, we find bidirectional causality between economic growth and exports; economic growth and imports; and exports and imports in the long run. Therefore, both the ELG and the ILG models are validated for the two countries (Figs. 2 and 3). For Nepal, we find unidirectional causality from exports to economic growth, and bidirectional causality between imports and economic growth in the long run. For Pakistan, in the long run we find unidirectional causality from economic growth to exports and bidirectional causality between economic growth and imports (Fig. 5). We find bidirectional long-run causality from export-growth and import-growth for Sri Lanka, and unidirectional causality from exports to imports in the long run (Fig. 6).8

6 A 1% increase in economic growth boosts exports and imports by 1.46 and 1.05% respectively, and a 1% increase in imports is associated with 1.28% increase in exports, on average. 7 A 1% increase in export enhances growth by 1.12%, and 1% increase in exports stimulates imports by 1.05%, on average, ceteris paribus. 8 The Figs. 1–6 show the only long run direction of causality.

Fig. 6. Trade and growth in Sri Lanka.

4. Policy implications and conclusion The paper investigates the ELG, GLE, ILG, GLI and foreign deficit sustainability hypotheses for six South Asian nation: Pakistan, Bangladesh, India, Sri Lanka, Nepal and Bhutan. We use the ADF unit root tests for stationarity, and the ARDL approach for a long-run relationship. The direction of long- and short-run causality is examined by using VECM. Each series is found to be integrated of order one for all six nations. Our main conclusions are as follows: ELG model is relevant for all countries except Pakistan; while the ILG model is relevant for all six Asian countries. The GLE model is relevant for all countries except Bangladesh and Nepal while GLI model is relevant for all the countries. The international budget constraints are weakly stable for the six South Asian countries. References Alam, M.S., 1991. Trade orientation and macroeconomic performance in DCs: an empirical study. Econ. Dev. Cult. Chang. 39 (4), 839–848. Al-Khulaifi, A.S., 2013. Exports and Imports in Qatar: evidence from cointegration and error correction model. Asian Econ. Financ. Rev. 3 (9), 1122–1133. Amirkhalkhali, S., Dar, A.-A., 1995. A varying-coefficients model of export expansion, factor accumulation and economic growth: evidence from cross-country, time series data. Econ. Model. 12 (4), 435–441. Arize, A., 2002. Imports and Exports in 50 countries: tests for cointegration and structural breaks. Int. Rev. Econ. Financ. 11 (1), 101–115. Awokuse, T.O., 2005. Exports, economic growth and causality in Korea. Appl. Econ. Lett. 12 (11), 693–696. Awokuse, T.O., 2007. Causality between exports, imports, and economic growth: evidence from transition economies. Econ. Lett. 94 (3), 389–395. Balassa, B., 1985. Exports, policy choices, and economic growth in developing economies after the 1973 oil shock. J. Dev. Econ. 18 (1), 23–35. Bhagwati, J.N., 1988. Protectionism. The MIT Press, Cambridge, M.A. Chandra, R., 2002. Export growth and economic growth: an investigation of causality in India. Indian Econ. J. 49, 64–73. Chandra, R., 2003. Reinvestigating export-led growth in India using a multivariate cointegration framework. J. Dev. Areas 37, 73–86. Coppin, A., 1994. Determinants of LDC output growth during the 1980s. J. Dev. Areas 28 (2), 219–228. Dar, A. b, Bhanja, N., Samantaraya, A., Tiwari, A.K., 2013. Export led growth or growth led export hypothesis in India: evidence based on time-frequency approach. Asian Econ. Fin. Rev. 3 (7), 869–880. De Gregorio, J., 1992. Economic growth in Latin America. J. Dev. Econ. 39, 59–84. Deme, M., 2002. An examination of the trade-led growth hypothesis in Nigeria: a cointegration, causality and impulse response analysis. J. Dev. Areas 36 (1), 1–15. Dodaro, S., 1991. Comparative advantage, trade and growth: export-led growth revisited. World Dev. 19 (9), 1153–1165. Easterly, W.R., 2007. Free market and economic development. International Symposium on Poverty Reduction and Beyond Development Strategies for Low Income Countries. Emery, R.F., 1967. The relation of exports and economic growth. Kyklos 20, 470–486. Fosu, A.K., 1990. Exports and economic growth: the African case. World Dev. 18 (6), 831–835. Fosu, A.K., 1996. Primary exports and economic growth in developing countries. World Econ. 19, 465–475. Heller, P.S., Porter, R.C., 1978. Exports and growth: an empirical reinvestigation. J. Dev. Econ. 5, 191–193.

660

Q.M.A. Hye et al. / Economic Modelling 35 (2013) 654–660

Helpman, Elhanan, Krugman, Paul R., 1985. Market Structure and Trade. MIT Press, Cambridge, Mass. Herrerias, M.J., Orts, V., 2009. Imports and Growth in China. available online at http://www. cerdi.org/uploads/sfCmsContent/html/317/Herrerias.pdf. Husted, S., 1992. The emerging US current account deficit in the 1980s: a cointegration analysis. Rev. Econ. Stat. 74 (1), 159–166. Hye, Q.M.A., 2012. Exports, imports and economic growth in China: an ARDL analysis. J. Chin. Econ. Foreign Trade Stud. Vol. 5 (1), 42–55. Hye, Q.M.A., Boubaker, H.B.H., 2011. Exports, imports and economic growth: an empirical analysis of Tunisia. IUP Monet. Econ. 9 (1), 6–21. Hye, Q.M.A., Siddiqui, M.M., 2010. are imports and exports cointegrated in Pakistan? A rolling window bound testing approach. World Appl. Sci. J. 9 (7), 708–711. Hye, Q.M.A., Siddiqui, M.M., 2011. Export-led growth hypothesis: multivariate rolling window analysis of Pakistan. Afr. J. Bus. Manag. 5 (2), 531–536. Katircioglu, S., Eminer, F., Aga, M., Ozyigit, A., 2010. Trade and growth in the Pacific Islands— Empirical evidence from the bound test to level relationship and Granger Causality tests. Rom. J. Econ. Forecast. (4), 88–101. Kravis, I.B., 1970. Trade as a handmaiden of growth: similarities between the nineteenth and twentieth centuries. Econ. J. 80 (320), 850–872. Lee, Chew Ging, 2010. Exports, imports and economic growth: evidence from Pakistan. Eur. J. Sci. Res. 45 (3), 422–429 (http://www.eurojournals.com/ejsr.htm, ISSN 1450-216X). Love, J., Chandra, R., 2004. Testing export-led growth in India, Pakistan and Sri Lanka using a multivariate framework. Manch. Sch. 72, 483–496. Love, J., Chandra, R., 2005. Testing export-led growth in South Asia. J. Econ. Stud. 32, 132–145. Mah, J.S., 2005. Export expansion, economic growth and causality in China. Appl. Econ. Lett. 12 (2), 105–107. Maizels, Alfred, 1963. Industrial Growth and World Trade. Cambridge University Press, Cambridge. Mallick, S.K., 1996. Causality between exports and economic growth in India: evidence from cointegration based error-correction model. Indian J. Econ. LXXVI, 307–320. Mbaku, J.M., 1989. Export growth and economic performance in developing countries: further evidence from Africa. J. Econ. Dev. 14, 127–142. McNab, R.M., Moore, R.E., 1998. Trade policy, export expansion, human capital and growth. J. Int. Trade Econ. Dev. 7 (2), 237–256. Michalopoulos, C., Jay, K., 1973. Growth of exports and income in the developing world: a neoclassical view. Discussion Paper No. 28. US Agency of International Development, Washington DC. Narayan, P.K., 2005. The saving and investment nexus for China: evidence from cointegration tests. Appl. Econ. 37 (17), 1979–1990.

Narayan, P.K., Smyth, R., 2004. Temporal causality and dynamics of exports, exports and real income in China. Int. J. Appl. Econ. 1 (1), 24–45. Otani, I., Villaneuva, D., 1990. Long-term growth in developing countries and its determinants: an empirical analysis. World Dev. 18 (6), 769–783. Pesaran, M.H., Shin, Y., Smith, R.J., 2001. Bounds testing approaches to the analysis of level relationships. J. Appl. Econ. 16 (3), 289–326. Ram, R., 1985. Exports and economic growth: some additional evidence. Econ. Dev. Cult. Chang. 33 (2), 415–425. Ramos, F.F.R., 2001. Exports, imports, and economic growth in Portugal: evidence from causality and cointegration analysis. J. Econ. Model. 18 (4), 613–623. Rana, P.B., 1986. Exports and economic growth: further evidence from Asian LDCs. Pakistan J. Appl. Econ. 5, 163–178. Salvatore, D., 1983. A simultaneous equations model of trade and development with dynamic policy simulations. Kyklos 36 (1), 66–90. Sato, S., Fukushige, M., 2007. The End of Import-Led Growth? North Korean Evidence. (online available at:) http://www2.econ.osaka-u.ac.jp/library/global/dp/0738.pdf. Shahbaz, M., Azim, P., Ahmad, K., 2011. Exports-led growth hypothesis in Pakistan: further evidence. Asian Econ. Fin. Rev. 1 (3), 182–197. Shan, J., Sun, F., 1998. On the export-led growth hypothesis: the econometric evidence from China. Appl. Econ. 30 (8), 1055–1065. Sheehey, E.J., 1992. Exports and growth: additional evidence. J. Dev. Stud. 28, 730–734. Singer, H.W., Gray, H., 1988. Trade policy and growth of developing countries: some new data. World Dev. 16 (3), 395–403. Sprout, R.V.A., Weaver, J.H., 1993. Exports and economic growth in a simultaneous equations model. J. Dev. Areas 27 (3), 289–306. Stiglitz, J.E., 2007. Making Globalization Work for Developing Countries. W.W. Norton & Company, Inc. Tang, T.C., 2006. New evidence on export expansion, economic growth and causality in China. Appl. Econ. Lett. 13 (12), 801–803. Tyler, W., 1981. Growth and export expansion in developing countries: some empirical evidence. J. Dev. Econ. 9 (1), 121–130. Voivodas, C.S., 1973. Exports, foreign capital inflow and economic growth. J. Int. Econ. 3 (1), 337–349. Williamson, R.B., 1978. The role of exports and foreign capital in Latin American economic growth. South. Econ. J. 45 (2), 410–420. Wu, Y., Zhang, J., 1998. Are the U.S. exports to and imports from Japan cointegrated? J. Econ. Integr. 13 (4), 626–643. Yaghmaian, B., Ghorashi, R., 1995. Export performance and economic development: an empirical analysis. Am. Econ. 39 (2), 37–45.