Scholarly article on topic 'Trade and economic growth in developing countries: Evidence from sub-Saharan Africa'

Trade and economic growth in developing countries: Evidence from sub-Saharan Africa Academic research paper on "Economics and business"

CC BY-NC-ND
0
0
Share paper
Academic journal
Journal of African Trade
OECD Field of science
Keywords
{"Trade openness" / "Economic growth" / "Sub-Saharan Africa"}

Abstract of research paper on Economics and business, author of scientific article — Pam Zahonogo

Abstract This study investigates how trade openness affects economic growth in developing countries, with a focus on sub-Saharan Africa (SSA). We use a dynamic growth model with data from 42 SSA countries covering 1980 to 2012. We employ the Pooled Mean Group estimation technique, which is appropriate for drawing conclusions from dynamic heterogeneous panels by considering long-run equilibrium relations. The empirical evidence indicates that a trade threshold exists below which greater trade openness has beneficial effects on economic growth and above which the trade effect on growth declines. The evidence also indicates an inverted U-curve (Laffer Curve of Trade) response, robust to changes in trade openness measures and to alternative model specifications, suggesting the non-fragility of the linkage between economic growth and trade openness for sub-Saharan countries. Our findings are promising and support the view that the relation between trade openness and economic growth is not linear for SSA. Accordingly, SSA countries must have more effective trade openness, particularly by productively controlling import levels, in order to boost their economic growth through international trade.

Academic research paper on topic "Trade and economic growth in developing countries: Evidence from sub-Saharan Africa"

ARTICLE IN PRESS

JAT-00012; No of Pages 16

Available online at www.sciencedirect.com

ScienceDirect

Journal of African Trade xx (2017) xxx - xxx

Trade and economic growth in developing countries: 2

Evidence from sub-Saharan Africa 3

Pam Zahonogo qi

UniversiteOuaga II, Burkina Faso, 03 BP 7164, Ouagadougou 03, Burkina Faso 5

Received 4 September 2015; received in revised form 13 February 2017; accepted 13 February 2017 6

Abstract 8

This study investigates how trade openness affects economic growth in developing countries, with a focus on 9

sub-Saharan Africa (SSA). We use a dynamic growth model with data from 42 SSA countries covering 1980 to 10

2012. We employ the Pooled Mean Group estimation technique, which is appropriate for drawing conclusions 11

from dynamic heterogeneous panels by considering long-run equilibrium relations. The empirical evidence 12

indicates that a trade threshold exists below which greater trade openness has beneficial effects on economic 13

growth and above which the trade effect on growth declines. The evidence also indicates an inverted U-curve 14

(Laffer Curve of Trade) response, robust to changes in trade openness measures and to alternative model 15

specifications, suggesting the non-fragility of the linkage between economic growth and trade openness for sub- 16

Saharan countries. Our findings are promising and support the view that the relation between trade openness and 17

economic growth is not linear for SSA. Accordingly, SSA countries must have more effective trade openness, 18

particularly by productively controlling import levels, in order to boost their economic growth through 19

international trade. 20

© 2017 Afreximbank. Production and hosting by Elsevier B.V. All rights reserved. This is an open access article 21

under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 22

JEL classification: F14; 040 24 Keywords: Trade openness; Economic growth; Sub-Saharan Africa 25 __26

1. Introduction 28

Trade liberalization has become widespread over the past three decades, particularly among 29 developing and transition economies, as a result of the perceived limitation of import substitution- 30

E-mail address: pzahonogo@gmail.com. Peer review under responsibility of Afreximbank.

http://dx.doi.org/10.1016/jjoat.2017.02.001

2214-8515/© 2017 Afreximbank. Production and hosting by Elsevier B.V. All rights reserved. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

ARTICLE IN PRESS

2 P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx

based development strategies and the influence of international financial institutions, such as the 31 International Monetary Fund and the World Bank, which have often made their support conditional 32 on trade liberalization. The fundamental rationale for this degree of commitment to a program of trade 33 reform is the obvious belief that liberalization is a prerequisite to a transition from relatively closed to 34 relatively open economies. Economists generally agree that open economies grow faster than their 35 counterparts do (Grossman and Helpman, 1991; Edwards, 1993). If openness is indeed positively 36 related to growth, it then follows that liberalization is a requirement for growth. Despite their early 37 promise, recent experience suggests that not all trade reforms have been as successful as anticipated 38 (Singh, 2010). 39

The relationship between trade openness and economic growth has been theoretically controversial. 40 While conventional wisdom predicts a growth-enhancing effect of trade, recent developments suggest 41 that trade openness is not always beneficial to economic growth. Increased international trade can 42 generate economic growth by facilitating the diffusion of knowledge and technology from the direct 43 import of high-tech goods (Barro and Sala-i-Martin, 1997; Baldwin et al., 2005; Almeida and 44 Fernades, 2008). Trade facilitates integration with the sources of innovation and enhances gains from Q2 foreign direct investment. By increasing the size of the market, trade openness allows economies to 46 better capture the potential benefits of increasing returns to scale and economies of specialization 47 (Alesina et al., 2000; Bond et al., 2005). In their theoretical models, Grossman and Helpman (1991) 48 show that trade openness improves the transfer of new technologies, facilitating technological progress 49 and productivity improvement, and that these benefits depend on the degree of economic openness. 50 This consensus rests on the assumption that trade creates economic incentives that boost productivity 51 through two dynamics: in the short-run, trade reduces resource use misallocation; in the long run, it 52 facilitates the transfer of technological development. Trade liberalization can also force governments 53 to commit to reform programs under the pressure of international competition, thus enhancing 54 economic growth (Sachs and Warner, 1995; Rajan and Zingales, 2003). Trade liberalization in 55 developing countries has therefore often been implemented with the expectation of growth stimulation. 56 However, endogenous growth models postulate that the contribution of trade to economic growth 57 varies depending on whether the force of comparative advantage orientates the economy's resources 58 toward activities that generate long-run growth or away from such activities. Moreover, theories 59 suggest that, due to technological or financial constraints, less-developed countries may lack the social 60 capability required to adopt technologies developed in more advanced economies. Thus, the growth 61 effect of trade may differ according to the level of economic development. Despite its potential positive 62 effect on growth, some theoretical studies claim that trade openness may hamper growth. For Redding 63 (1999), Young (1991), and Lucas (1988), opening up to trade might actually reduce long-run growth if 64 an economy specializes in sectors with dynamic comparative disadvantage in terms of potential 65 productivity growth or where technological innovations or learning by doing are largely exhausted. 66 For such economies, selective protection may foster faster technological advances. 67

The empirical analyses are as inconclusive as the theoretical perspectives. Some studies have 68 identified a positive association between trade openness and economic growth (Chang et al., 2009; 69 Kim, 2011; Jouini, 2015), while others have found no association, or even a negative association 70 (Musila and Yiheyis, 2015; Ula^an, 2015). The literature is inconclusive partly because different 71 analysts use different proxies for liberalization or trade openness and rely on different methodologies. 72 The evidence for growth enhancements through trade liberalization displays mixed effects because of 73 problems with misspecification and the diversity among the liberalization indices used. 74

Using cross-country data and initial real income per capita as the threshold variable, Kim and Lin 75 (2009) found significant threshold effects in the relationship between trade and growth. Greater 76 openness to international trade has positive impacts on economic growth for high-income economies. 77

ARTICLE IN PRESS

P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx 3

For low-income economies, however, higher trade openness has negative impacts on economic 78 growth. The beneficial effects of trade liberalization thus seem to increase as economies develop, 79 confirming the arguments about the adoptive capacity of a country in determining knowledge 80 accumulation and technology implementation. 81

This study contributes to research by assessing whether the relationship between trade and growth 82 differs between more open and less open countries in SSA. This distinction is important, as various 83 theoretical models and empirical results have suggested that the effect of trade on economic growth 84 may vary according to the level of trade openness and level of income as a measure of economic 85 development. Particularly for developing countries, the lack of investment in human capital and of a 86 well-functioning financial system may hamper the growth expected from trade liberalization through 87 technological innovation. For such countries, Kim and Lin (2009) suggest selective protection. If such 88 a nonlinear relationship exists, we should be able to estimate the threshold at which the sign of the 89 relation between trade and growth switches. This study reexamines the role of trade and contributes to 90 the literature by empirically analyzing the threshold effects of trade on economic growth based on 91 panel data for sub-Saharan African countries. The empirical evidence is based on a dynamic growth 92 model using data from 42 sub-Saharan countries covering 1980 to 2012. We employ the Pooled Mean 93 Group estimation technique, which is appropriate for drawing conclusions from dynamic heterogeneous 94 panels by considering long-run equilibrium relations. Our findings support the view that the relation 95 between trade globalization and economic growth is not linear for sub-Saharan Africa and point to an 96 inverted U curve-type response. 97

The remainder of the paper is organized as follows. Section 2 reviews the theoretical and empirical 98 literature linking trade and economic growth. Section 3 presents the study's model and discusses the 99 relevant econometric issues. Section 4 presents the data used to implement the model. Section 5 100 summarizes and analyses the empirical results. Finally, Section 6 concludes the paper and outlines its 101 main economic policy implications. 102

2. Trade openness and growth in the literature 103

Traditional trade theory predicts growth gains from openness at the country level through 104 specialization, investment in innovation, productivity improvement, or enhanced resource allocation. 105 The role of trade policy in economic development has been a key matter of debate in the development 106 literature. Ricardo's theory suggests that openness abroad allows a country to reorient its scarce 107 resources to more efficient sectors. The neoclassical growth models drawn from Solow's (1957) 108 model consider technological change as exogenous and suggest that, consequently, trade policies do 109 not impact economic growth. However, new economic growth theories assume that technological 110 change is an endogenous variable and that trade policies can be combined with those on international 111 trade. The existence and nature of the link between trade openness and economic growth have been 112 the subject of considerable debate. However, neither the existing theoretical models nor empirical 113 analyses have produced a definite conclusion. 114

The potential growth effects of trade liberalization are well known. While the intermediate impact 115 is likely to be negative, as resources become redundant in areas of comparative disadvantage, their 116 eventual reallocation into areas of comparative advantage will increase the growth rate; the evidence 117 points to a J curve-type response (Greenaway et al., 2002; Falvey et al., 2012). Longer growth gains 118 must be obtained through improvements in factor productivity (Kim and Lin, 2009), which can occur 119 through a variety of channels such as technology diffusion and innovation. While trade openness 120 facilitates the diffusion of technology and innovations (Krueger and Berg, 2003; Lucas, 1998), Q3 technology adoption depends on a country's absorptive capacity, which is determined by human 122

ARTICLE IN PRESS

4 P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx

capital and R&D (Verpagen, 1991; Fagerberg, 1994), financial development (Aghion et al., 2005), Q4 governance, and national institutional settings (Haltiwanger, 2011; McMillan and Verduzco, 2011). 124 Thus, developing countries—characterized by a lack of human capital, R&D, a well-functioning 125 financial system, and a high-quality bureaucracy—may not take full advantage of technology 126 transfer. 127

The empirical results, like the theoretical analyses, are controversial. The evidence has indicated 128 that excessive regulations restrict growth because resources are prevented from moving into the most 129 productive sectors and to the most efficient firms within sectors (Bolaky and Freund, 2008) and that 130 institutions can help explain the heterogeneity in the trade-growth relationship (Sindzingre, 2005). 131 Falvey et al. (2012) employed threshold regression techniques on crisis indicators to identify the 132 relevant crisis values and the differential post-liberalization growth effects in crisis and non-crisis 133 regimes. Their findings indicate that an economic crisis at the time of liberalization does affect 134 post-liberalization growth, in a direction that depends on the nature of the crisis. An internal crisis 135 implies lower growth and an external crisis higher growth relative to a non-crisis regime. Based on an 136 augmented production function, Fosu (1990) argued that export increases improve economic growth 137 in African countries, whereas Ula^an (2015) used a dynamic panel data framework to conclude that 138 trade openness measures are not robustly significantly associated with economic growth, implying 139 that trade openness alone does not boost economic growth. Trejos and Barboza (2015) provide robust 140 empirical evidence that trade openness is not the main engine of the Asian economic growth 141 "miracle." 142

The benefits of trade openness are not automatic. Policies, such as measures aimed at fostering 143 macroeconomic stability and a favorable investment climate, must accompany trade openness Q5 (Newfarmer and Sztajerowska, 2012). Kim and Lin (2009) found that trade openness contributes to 145 long-run economic growth, with effects varying according to the level of economic development. 146 Herzer (2013) found that the impact of trade openness is positive for developed countries and 147 negative for developing ones. The effect of trade liberalization on growth depends on the 148 liberalization level. An income threshold exists above which greater trade openness has beneficial 149 effects on economic growth and below which increased trade has detrimental consequences (Agénor, 150 2004; Liang, 2006). 151

Empirical studies have found a possible two-way causality in the trade-growth link, whereby 152 countries that trade more may have higher income, while countries with higher income may be better 153 able to afford the infrastructure conducive to trade, may have more resources with which to overcome 154 the information search costs associated with trade, or may demand more traded goods (Kim and Lin, 155 2009). Zeren and Ari (2013) revealed positive bidirectional causal links between openness and 156 economic growth for G7 countries. 157

Among the reasons for the inconclusive results in the literature on the trade-growth link is the fact 158 that different studies use different proxies for trade openness and rely on different methodologies. 159 Most empirical studies based on cross-country growth regressions suggest a significant growth- 160 promoting effect of trade openness, although these have been criticized for poor data quality and 161 inadequate control of endogeneity (Edwards, 1998; Le Goff and Singh, 2014). The inconclusive 162 results may occur because trade liberalization must almost certainly be combined with other 163 appropriate policies, and linear regression models cannot capture such complementary dynamics 164 (Winters, 2004). Greenaway et al. (2002) provided evidence that misspecification and the diversity of 165 liberalization indices are partly responsible for the inconclusiveness of the research. Using a dynamic 166 panel framework and three different indicators of liberalization, their results indicate that 167 liberalization does appear to impact growth, albeit with a lag. These results suggest that working in 168 a panel context is more effective than working in a cross-country context. Such a technique extracts 169

ARTICLE IN PRESS

P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx 5

more information and produces more reliable estimates than the time series and cross-section 170 regressions do. 171

This study complements the literature on trade and growth by providing new cross-country 172 empirical evidence that considers the threshold effects of trade across sub-Saharan Africa. Moreover, 173 rather than just focusing on the direct effect of trade on growth, this study goes further and explores 174 other channels through which trade can affect economic progress, such as governance, financial 175 development, and education. 176

3. Model and econometric issues 177

The neoclassical augmented growth model developed by Mankiw et al. (1992) is utilized to 178 estimate the effect of trade openness on economic growth. Two main motivations underlie the 179 specification of this model. First, the model considers human capital, which enhances labor 180 productivity and can boost growth. Second, as the objective is to see how growth is influenced by 181 trade and the economic policy environment, several policy-related variables are used in the equation. 182 Taking into account the variable of interest (trade) and the heterogeneity of the coefficients and other 183 control variables, the model can be expressed as follows: 184

Yu = ai + XiYit-i + Tp=iPpiXpit + yutradeu + y^tradel + zit (1)

where Yit is GDP per capita for country i at time t, X is the vector of control variables, including 185 education, rate of population growth, investment rate, financial development, institutions, crisis, and 187 debt. Trade is a trade openness variable, £it is an error term, and ai reflects country-specific effects. 188 The logarithm of the initial GDP is included to control for convergence. However, it can also be 189 interpreted as a proxy for a country's stock of capital. Under this assumption, economic growth in the 190 poorest countries is more rapid than in the richest countries. The coefficient of this variable should be 191 negative. 192

The transformation of Eq. (1) as an error correction equation gives 193

A Yit = $i(Yit-1-9oi-Y?p=l9piXpU-1-^1itradeit-1-&2itradelU-^j-Y^=1^piAXpit-y^-Atia detit-y i,Atra det2u + &it

with doi = ai h-K, dpi = Ppi/ 1-Xi, ¿1; = Vii/1^, Sli = h-Xi, & = - (1 - Ai).

Economic growth is captured by the log-difference of real GDP per capita (AYit). The trade 196 variable is captured by three indicators: the ratio of the sum of imports and exports to GDP, exports to 197 GDP, and the ratio of imports to GDP. To address the possible causality between growth and trade, 198 the Hodrick Prescott filter (HP) has been used in the denominator (filtered GDP) of the three measures 199 to smooth the series by following the method of Arnone and Presbitero (2010). 200

Furthermore, 0oi is introduced for country-specific effects, sit represents the term of error, 0pi, 81i 201 and 82i capture the dynamic of long-run effects, while p>pi, Y1i, and Y2i capture the short-run dynamics. 202 Finally, the quadratic form is introduced to capture the nonlinear relationship or threshold effect 203 between trade and economic growth. ^ represents adjustment speed toward the long-run state; this 204 should be negative and significant to confirm the long-run relationship between trade and economic 205 growth. 206

Following Pesaran et al. (1999) and Jouini (2015), we apply the maximum-likelihood method to 207 estimate Eq. (2) by initially assuming that the error terms are normally distributed. The Pooled Mean 208 Group (PMG) approach is used to estimate dynamic heterogeneous panels by considering long-run 209 equilibrium relations, contrary to other techniques, such as the dynamic panel GMM method, that 210

ARTICLE IN PRESS

6 P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx

purge any potential long-run linkage among variables. The PMG estimation approach allows identical 211 long-run coefficients without assuming homogeneous short-run parameters. By doing so, the PMG 212 estimation approach differs from techniques, such as the Mean Group (MG) developed by Pesaran 213 and Smith (1995), that estimate a regression for each group and then calculate the coefficient means 214 (Evans, 1997; Lee et al., 1996). The MG long-run estimators are consistent, but they are inefficient if 215 coefficient homogeneity holds. Under these conditions, the PMG estimation approach is useful since 216 it provides consistent and efficient long-run estimators when parameter homogeneity holds. The PMG 217 approach is preferable to the MG method since it provides estimates that are less sensitive to outlier 218 estimates. 219

We address endogeneity concerns by augmenting the PMG estimator with lags of regressors and 220 dependent variables to minimize the resultant bias and ensure that the regression residuals are serially 221 uncorrelated. 222

Eq. (2) is rewritten as follows: 223

AYU = ^^YU-i-e0i-^kp=i()pXu-1-~5itradeU-i-~52trade2U-1Sj-^p=iPpit\.XpiryXit±tradeu-y^ktradel + su (3)

The trade threshold by country trade i * is obtained as follows: trade*t = -8ii/2g2 ; with the 224

—Sii /

threshold e ' 26ii. To enhance robustness, the effect of the 2008-2010 crisis is taken into account in 227 the model through a dummy variable. We consider 2008 as the breakpoint because this year is the 228 beginning of the crisis. The crisis variable takes zero before 2008 and one after 2008. We conducted 229 three stationarity tests: the test of Levine et al. (LLC, 2002), the test of Im et al. (IPS, 2003), and the 230 test of Maddala and Wu (MW, 1999). These tests are a generalization of the Augmented Dickey- 231 Fuller test (ADF). For cointegration, we applied the Pedroni (1999) and Kao (1999) tests. 232

4. Data and variable definitions 233

This study uses annual data covering 1980 to 2012 taken from 42 sub-Saharan African countries. 234 The choice of the period of study is related to the availability of data on interest variables such as trade 235 and economic growth. The dependent variable is economic growth, measured as the log difference of 236 the gross domestic product per capita (GDP). We also include a set of control variables that are 237 commonly used in growth equations. 238

4.1. Trade openness variables 239

Following Jouini (2015), Le Goff and Singh (2014), Zeren and Ari (2013), and Ula^an (2015), we 240 chose as our trade openness variable a measure of effective trade openness and not a measure of 241 liberalization policies because the main concern of this study is the impact of actual globalization on 242 economic growth. We employ three variables for trade openness: we use the sum of exports and 243 imports as a share of GDP (TRADE); for a robustness check, we also consider exports (EXPORT) 244 and imports (IMPORT) as a share of GDP separately. 245

4.2. Control variables 246

Since macroeconomic policies affect growth performance through their impact on the rate of 247 inflation, financial development, the financial crisis, external debt, investment in human and physical 248 capital, and institutional quality, variables for these effects are used in the growth equation to capture 249

ARTICLE IN PRESS

P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx

the impacts of such policies. The effect of inflation (INF) is a controversial issue. Some studies claim 250 that inflation has a positive impact on growth (Dornbusch et al., 1996), while other studies suggest 251 that this effect is characterized by a nonlinear relationship (Fischer, 1993; Kremer et al., 2009). 252 Inflation is proxied by the rate of change in price levels. Investment (INV) has been used in empirical 253 studies because it is viewed as a direct proxy of contribution to capital accumulation, as well as an 254 indicator of efforts to develop basic economic infrastructure. It is measured in this study as gross fixed 255 capital formation. Human capital is a key determinant of technology adoption as permitted by trade 256 openness (Benhabib and Spiegel, 2005; Cohen and Levinthal, 1989). The effect of human capital 257 is captured by using two variables: the gross secondary enrollment rate (EDU) and the population 258 growth rate (POP). The lack of a well-functioning financial system may prevent less-developed 259 countries from taking full advantage of technology transfer from trade openness and may impact 260 economic growth. Financial development (FD) is measured by private credit as a share of GDP. 261 Institutional quality is included in the growth equation to capture the impact of political rights and 262 civil liberties. It is hypothesized that the absence of political rights and civil liberties lowers the 263 security of life and property, thus reducing the rate of accumulation and the efficiency of factor of 264 production. Institutional quality therefore impacts economic growth (Asiedu, 2003; Acemoglu and 265 Robinson, 2012) and may also impact trade openness (Falvey et al., 2012). Institutional quality is 266 proxied by a governance index (IGOV) through the average of the six institution measures 267 presented by Kaufmann et al. (1999): voice and accountability, political stability and absence of 268 terrorist violence, government effectiveness, regulatory quality, rule of law, and control of 269 corruption. We also add variables on external debt (DEBT) and crisis (CRISIS) to capture the 270 effect of debt and financial crisis on economic growth. External debt is deleterious to economic 271 growth (Fosu, 1999), and a financial crisis can reduce GDP growth. Fosu (2013) found that the 272 2008 crisis reduced growth by 60% in sub-Saharan Africa. Following Chang et al. (2009), we 273 introduce interacting terms to allow the economic growth-trade openness relationship to vary with 274 several country characteristics (i.e., governance, education, and financial development). Table A4 275 in Appendix 1 provides the data sources, definitions, and descriptive statistics of the variables used 276 in the model. 277

5. Empirical results 278

The existence of a unit root is tested using the tests employed in Levine et al. (2002), Im et al. 279 (2003), and Maddala and Wu (1999). These tests are performed on the variables in the model, levels, 280 and difference. The null hypothesis of the presence of a unit root is rejected if the three tests confirm 281 that hypothesis simultaneously. According to the statistics of the three types of unit root test, variables 282 such as LGDP (logarithm of GDP), LDEBEXP (logarithm of ratio of external debt to export), LINV 283 (logarithm of investment), EDU (education variable), LDEBGDP (logarithm of ratio of external debt 284 to GDP), and LFD (logarithm of financial development variable) are non-stationary in level. 285 Stationary variables in level are INF (inflation rate), LTRADE (logarithm of trade openness), IGOV 286 (governance index), POP (population growth rate), LSDEBEXP (logarithm of external debt services 287 to export), LEXPORT (logarithm of export), and LIMPORT (logarithm of import). The results of 288 these tests on the variables in first difference fail to accept the hypothesis of the presence of a unit root 289 for all variables at a 1% level (see Table A1 of Appendix 1). We also conduct cointegration tests using 290 Pedroni (1999) and Kao (1999) tests. The results suggest a cointegration relation between economic 291 growth and trade openness (see Table A2 of Appendix 1). 292

ARTICLE IN PRESS

8 P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx

t1.1 Table 1

t1.2 PMG long-run estimates of impact of trade openness ((X + M)/GDP)) on growth of per capita GDP, 1980-2012.

t1.3 Model 1 Model 2 Model 3 Model 4

t1.4 Coef. Coef. Coef. Coef.

t1.5 GDP(-1) - 0.0686*** -0.0759*** -0.0334*** -0.0968***

t1.6 (0.0124) (0.0132) (0.0137) (0.0185)

t1.7 LINV 0.4168*** 0.4430*** 0.5529*** 0.2014***

t1.8 (0.0263) (0.0243) (0.0441) (0.0179)

t1.9 LFD - 0.1425*** - 0.1227*** 2.1508*** - 0.1360***

t1.10 (0.0309) (0.0291) (0.3020) (0.0190)

t1.11 INF -0.0001*** -0.0001*** -0.0004*** -0.0003**

t1.12 (0.0000) (0.0000) (0.0001) (0.0001)

t1.13 IGOV 0.2424*** -1.0389*** 0.1398** -0.0815***

t1.14 (0.0414) (0.2442) (0.0552) (0.0288)

t1.15 POP - 0.0317* -0.0299** 0.1135*** -0.0213***

t1.16 (0.0164) (0.0146) (0.0245) (0.0068)

t1.17 EDU 0.0079*** 0.0074*** 0.0083*** -0.0229***

t1.18 (0.0012) (0.0011) (0.0023) (0.0057)

t1.19 CRISIS - 0.1063*** -0.1211*** - 0.1302** 0.0171

t1.20 (0.0381) (0.0361) (0.0522) (0.0273)

t1.21 LDEBT -0.0380** -0.0585*** -0.0119 - 0.1112***

t1.22 (0.0173) (0.0137) (0.0241) (0.0149)

t1.23 LTRADE 1.9357*** 0.3480*** 1.4348*** 0.0084

t1.24 (0.5401) (0.0515) (0.2181) (0.0520)

t1.25 LTRADE2 - 0.1975***

t1.26 (0.0644)

t1.27 IGOV*LTRADE 0.2918***

t1.28 (0.0582)

t1.29 LFD*LTRADE -0.6004***

t1.30 (0.0789)

t1.31 EDU*LTRADE 0.0094***

t1.32 (0.0013)

t1.33 Cons - 0.3316*** -0.1749*** -0.2760** 0.3521***

t1.34 (0.0667) (0.0410) (0.1350) (0.0585)

t1.35 Number of countries 42 42 42 42

t1.36 Number of observations 1302 1302 1302 1302

t1.38 Note: ***, **, and * indicate statistical significance at 1%, 5%, and 10% levels respectively.

t1.39 Standard errors in parentheses.

t1.40 Source: Author's calculations with data provided.

The PMG estimation results are summarized in Tables 1 to 3. We present only the long-run 293 coefficients for the analysis (short-run coefficients are available on request). The short-run error 294 correction term (speed of adjustment) is significantly negative for all models, confirming the 295 cointegration relationship between the variables of interest and implying that the linkage between 296 economic growth and the explanatory variables is characterized by high predictability and that the 297 spread movement is mean-reverting. 298

The estimation results indicate a nonlinear relationship between trade openness and economic 299 growth, and the evidence is robust to alternative trade openness measures (i.e., sum of exports and 300 imports as a share of GDP and exports and imports as a share of GDP separately). The results show 301 the presence of a Laffer Curve of trade (inverted U) and confirm that trade openness has a positive and 302

ARTICLE IN PRESS

P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx 9

t2.1 Table 2

t2.2 PMG long-run estimates of impact of trade openness (X/GDP) on growth of per capita GDP, 1980-2012.

t2.3 t2.4 Model 1 Model 2 Model 3 Model 4

Coef. Coef. Coef. Coef.

t2.5 GDP(— 1) - 0.0704*** -0.0782*** -0.0688*** -0.0606***

t2.6 (.0186) (0.0136) (0.0146) (0.0131)

t2.7 LINV 0.2718*** 0.4568*** 0.4393*** 0.5051***

t2.8 (0.0171) (0.0247) (0.0248) (0.0291)

t2.9 LFD - 0.1900*** -0.1056*** -0.2981** -0.0552*

t2.10 (0.0243) (0.0295) (0.1176) (0.0332)

t2.11 INF -0.0018*** -0.0001*** -0.0001*** -0.0001***

t2.12 (0.0005) (0.0000) (0.0000) (0.0000)

t2.13 IGOV 0.1450*** -0.5467*** 0.2100*** 0.3174***

t2.14 (0.0365) (0.1568) (0.0423) (0.0519)

t2.15 POP -0.0329*** -0.0135 -0.0374*** -0.0384**

t2.16 (0.0055) (0.0138) (0.0140) (0.0161)

t2.17 EDU 0.0148*** 0.0075*** 0.0077*** 0.0170***

t2.18 (0.0007) (0.0009) (0.0010) (0.0028)

t2.19 CRISIS 0.0193 -0.1095*** -0.0672* - 0.1160***

t2.20 (0.0257) (0.0376) (0.0369) (0.0383)

t2.21 LDEBT -0.0494*** -0.0764*** - 0.1041*** -0.0770***

t2.22 (0.0123) (0.0105) (0.0116) (0.0108)

t2.23 LEXPORT 1.0118*** 0.2940*** 0.1929** 0.5411***

t2.24 (0.2876) (0.0297) (0.0916) (0.0821)

t2.25 LEXPORT2 - 0.0861*

t2.26 (0.0442)

t2.27 IGOV*LEXPORT 0.2026***

t2.28 (0.0462)

t2.29 LFD*LEXPORT 0.0572*

t2.30 (0.0342)

t2.31 EDU*LEXPORT -0.0032***

t2.32 (0.0009)

t2.33 Cons 0.0040 - 0.1675*** -0.0806*** - 0.2231***

t2.34 (0.0170) (0.0361) (0.0269) (0.0536)

t2.35 Number of countries 42 42 42 42

t2.36 Number of observations 1302 1302 1302 1302

t2.37 Note: ***, **, and * indicate statistical significance at 1%, 5%, and 10% levels respectively.

t2.39 Standard errors in parentheses.

t2.40 Source: Author's calculations with data provided.

significant effect on economic growth but only up to a threshold; above this threshold, the effect 303 declines. These results indicate that the openness variables are relevant drivers of economic growth 304 for sub-Saharan countries over the long run but that openness should be controlled since the 305 associated coefficients of such variables and their quadratic terms are (respectively) positively and 306 negatively significant at conventional levels. 307

For the first measure of trade openness (sum of exports and imports as a share of GDP), the 308 threshold is estimated to be 134.21%. In other words, trade is associated with higher levels of 309 economic growth up to the threshold where the sum of exports and imports represents 134.21% of 310 GDP. Beyond this threshold, the effect of trade on growth declines. According to the second 311 measure of trade openness (exports as a share of GDP), the findings indicate a threshold of 312

ARTICLE IN PRESS

10 P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx

t3.1 Table 3

t3.2 PMG long-run estimates of impact of trade openness (M/GDP) on growth of per capita GDP, 1980-2012.

t3.3 Model 1 Model 2 Model 3 Model 4

t3.4 Coef. Coef. Coef. Coef.

t3.5 GDP(-1) - 0.0771*** -0.0696*** -0.0477*** -0.0641***

t3.6 (0.0229) (0.0118) (0.0163) (0.0129)

t3.7 LINV 0.2774*** 0.4118*** 0.5694*** 0.3909***

t3.8 (0.0165) (0.0256) (0.0363) (0.0248)

t3.9 LFD 0.0089 -0.0896*** 1.5119*** —0 1444***

t3.10 (0.0214) (0.0318) (0.1995) (0.0310)

t3.11 INF - 0.0001* -0.0001*** -0.0003*** - 0.0001***

t3.12 (0.0000) (0.0000) (0.0001) (0.0000)

t3.13 IGOV 0.0419 -0.6107** 0.1512*** 0.2754***

t3.14 (0.0288) (0.2444) (0.0481) (0.0380)

t3.15 POP -0.1999*** -0.0757*** 0.0910*** -0.0492***

t3.16 (0.0229) "(0.0185) (0.0191) (0.0144)

t3.17 EDU 0.0043*** 0.0064*** 0.0076*** -0.0095

t3.18 (0.0010) (0.0011) (0.0016) (0.0062)

t3.19 CRISIS -0.0688* - 0.1234*** - 0.1643*** - 0.1114***

t3.20 (0.0375) (0.0385) (0.0410) (0.0345)

t3.21 LDEBT -0.2558*** -0.1492*** -0.0357 -0.0848***

t3.22 (0.0231) (0.0223) (0.0255) (0.0185)

t3.23 LIMPORT 1.0690*** 0.1625*** 1.3083*** 0.0116

t3.24 (0.2338) (0.0437) (0.1820) (0.0734)

t3.25 LIMPORT2 - 0.1526***

t3.26 (0.0331)

t3.27 IGOV*LIMPORT 0.1965***

t3.28 (0.0667)

t3.29 LFD*LIMPORT -0.5152***

t3.30 (0.0605)

t3.31 EDU*LIMPORT 0.0047***

t3.32 (0.0017)

t3.33 Cons 0.1445*** -0.0184 -0.3593*** 0.0445***

t3.34 (0.0339) (0.0172) (0.1400) (0.0141)

t3.35 Number of countries 42 42 42 42

t3.36 Number of observations 1302 1302 1302 1302

t3.37 Note: ***, **, and * indicate statistical significance at 1%, 5%, and 10% levels respectively.

t3.39 Standard errors in parentheses.

t3.40 Source: Author's calculations with data provided.

355.68%, suggesting that trade affects economic growth positively until exports account for 313 355.68% of GDP. After this threshold, trade's impact on economic growth declines. For the third 314 measure (imports as share of GDP), trade is associated with a higher level of growth when imports 315 account for 33.16% of GDP; the effect declines after this threshold (see Table A3 of Appendix 1). 316 The probability of reaching these threshold proportions of trade (134.21% or 355.68% of GDP) is 317 small, practically non-existent, indicating that openness to exports may not reduce economic 318 growth for sub-Saharan African countries. For the third measure, the result suggests that imports 319 can reduce economic growth. Thus, sub-Saharan African countries must efficiently control trade 320 openness, particularly import levels, when seeking to boost their economic growth through 321

ARTICLE IN PRESS

P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx 11

international trade. Trade provides developing countries with access to the investment and 322 intermediate goods that are vital to their development and the transfer of foreign technology, but 323 such countries should productively reduce the import of consumption goods, by creating an 324 environment that is conducive to efficiently producing domestically competing products in which 325 there is dynamic comparative advantage. 326

These findings suggest that the openness of African economies to international trade should be 327 associated with growth, which is in line with other empirical works (Ismail et al., 2010; Erfakar, 328 2011). Unlike in previous studies, however, the relation is not linear, confirming the fragility of the 329 links between trade openness and economic growth for sub-Saharan African countries. This result is 330 in line with the findings in Ula^an (2012) on Organization for Economic Cooperation and Q6 Development (OECD) and non-OECD countries. 332

While the first regression considers only trade openness effects, we then examine the influence of 333 several structural country characteristics in the trade-growth relationship. The results with the interaction 334 terms are shown in Models 2, 3, and 4 (see Tables 1 to 3). We first test whether the trade-growth 335 relationship changes with the development of the financial sector (Model 3). The coefficient of the 336 interaction term with financial development is negative and significant for the first and third measures of 337 trade openness and is positive for the second measure. These results suggest that a greater openness to 338 trade via exports is associated with strong economic growth when the financial sector is more developed. 339 In other words, easier access to credit may allow the export-oriented sectors to benefit more from trade 340 openness. However, greater openness via imports is associated with lower economic growth, suggesting 341 that easier access to credit among import-oriented sectors may reduce economic growth. 342

Model 4 shows the results of the estimations testing the role of human capital in the trade-growth 343 relationship. The beneficial impact of an increase in trade openness on growth is greater when 344 investment in human capital is higher. We find that an increase in the gross secondary enrollment rate 345 is associated with a higher growth rate. This result is consistent with theoretical models suggesting 346 that the effect of trade on growth may depend on the adoption of technology determined by human 347 capital. Finally, we examine whether the relationship between openness to trade and economic growth 348 may hinge on a country's institutional environment (Model 2). The results suggest that trade openness 349 may be favorable to economic growth when institutional quality improves. In others words, an 350 environment with high-quality governance seems to be more favorable to the emergence of new 351 enterprises, allowing the economy to grow faster. 352

6. Conclusion 353

In this study, we have tested a dynamic growth model for sub-Saharan African economies using three 354 measures of trade openness. Our results suggest that trade openness may impact growth favorably in the 355 long run, but the effect is not linear. Our results show the presence of a Laffer Trade Curve (inverted U) 356 and confirm that trade openness has a positive and significant effect on economic growth only up to a 357 threshold, above which the effect declines. These results are robust to changes in trade openness 358 measures. The non-linear relation between trade openness and economic growth suggests that the 359 benefits of trade are not automatic. The growth effects of trade openness may differ according to the level 360 of trade openness. Accordingly, sub-Saharan African countries must productively control trade 361 openness, particularly the import of consumption goods, in boosting their economic growth through 362 international trade. Our results suggest that trade openness must be accompanied by complementary 363 policies aimed at encouraging the financing of new investment and enhancing the quality of institutions 364 and the ability to adjust and learn new skills. These policies would then allow resources to be reallocated 365 away from less productive activities and toward more promising ones. Trade globalization should 366

ARTICLE IN PRESS

12 P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx

therefore not be seen in isolation. Additional policies are needed to enhance its impact on economic 367 growth. Sub-Saharan countries should carry out relevant policy reform to encourage investment, allow 368

effective governance, and promote human capital accumulation. 369

Uncited reference Q7

World Bank, 2014 371

Acknowledgment 372

We are grateful to the Editor-in-Chief and the referees of the journal for their comments. Any 373 errors are, however, ours. 374

Appendix 1. Panel unit root, panel cointegration, and threshold effects tests 375

Table A1

Panel unit root test.

Variable Level First difference

LLC IPS MW LLC IPS MW

LGDP - 3.123*** -0.476 123.232*** -23.396*** -25.361*** 953.534***

(0.001) (0.316) (0.003) (0.000) (0.000) (0.000)

INF -20.200*** -21.707*** 795.128*** -54.469*** -51.631*** 2342.69***

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

LTRADE - 5.189*** -4.640*** 163.589*** -39.362*** -37.658*** 1613.635***

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

IGOV -13.660*** -10.802*** 1531.543*** -120.00*** -121.000*** 3005.901***

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

LDEBTEXP - 4.631*** 1.226 81.059 -30.446*** -28.976*** 1123.340***

(0.000) (0.889) (0.570) (0.000) (0.000) (0.000)

LINV -4.568*** -0.722 90.878 -29.067*** -29.069*** 1141.552***

(0.000) (0.234) (0.285) (0.000) (0.000) (0.000)

POP -7.162*** -2.015** 230.614*** -17.582*** - 21.896*** 814.422***

(0.000) (0.021) (0.000) (0.000) (0.000) (0.000)

EDU -4.143*** -0.692 126.890*** -26.748*** -29.656*** 1217.673***

(0.000) (0.244) (0.001) (0.000) (0.000) (0.000)

LDEBGDP -2.511*** 3.875 52.314 -23.812*** -23.843*** 848.934***

(0.006) (0.999) (0.997) (0.000) (0.000) (0.000)

LSDEBTEXP -9.731*** -6.977*** 227.586*** -38.191*** -37.062*** 1607.242***

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

LFD -2.037** 1.644 70.877 -29.279*** -26.213*** 931.556***

(0.020) (0.950) (0.845) (0.000) (0.000) (0.000)

LEXPORT - 4.2118*** -4.2080*** 163.371*** -36.683*** - 32.198*** 1235.9***

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

LIMPORT -5.6871*** -6.819*** 265.09*** -39.383*** -34.966*** 1341.3***

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

Source: Author's calculations with data provided.

Note: LLC, IPS, and MW indicate (respectively) unit root tests in Levine et al. (2002), Im et al. (2003), and Maddala and Wu (1999; Fisher-ADF). The values in parentheses represent the probabilities associated with the test statistics. ***, **, and * are the release thresholds of the null hypothesis of presence of unit root, 1%, 5%, and 10%.

ARTICLE IN PRESS

P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx

Table A2

Panel cointegration test.

Ltrade

Panel statistics

Group statistics

PEDRONI V-statistic

Rho-statistic

PP-Statistic

ADF-statistic

13.16927*** (0.000) 0.22481 (0.5889) -1.50571* (0.0661) 1.03177 (0.8489)

ADF-statistic: -7.243*** (0.000)

1.0013

(0.8417)

-2.7078***

(0.0034)

0.04023

(0.5160)

Lexport

Panel statistics

Group statistics

PEDRONI V-statistic

Rho-statistic

PP-statistic

ADF-statistic

14.08436*** (0.000) -1.2184 (0.1115)

- 3.0474*** (0.001)

- 0.4126 (0.3399)

ADF-statistic: -7.645*** (0.000)

0.2702

(0.6065)

-3.2900***

(0.0005)

- 0.6122

(0.2702)

Limport Panel statistics Group statistics

PEDRONI

V-statistic 13.5655***

(0.000)

Rho-statistic 0.38369 1.30834

(0.6494) (0.9046)

PP-statistic -1.43443* -2.396713***

(0.0757) (0.0083)

ADF-statistic 1.07839 0.31785

(0.8596) (0.6247)

KAO ADF-statistic: -7.143117***

(0.000)

Source: Author's calculations with data provided.

Note: Standard errors in parentheses; ***, **, and * specify that coefficients are statistically significant at the 1%, 5%, and 10% levels.

ARTICLE IN PRESS

14 P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx

Table A3 Test results of threshold effects.

Openness indicators

Ltrade Lexport Limport

Growth model 134.21 355.68 33.16

Notes: The threshold values are in percentage of GDP.

Table A4

Data sources, definitions, and descriptive statistics.

Variable Definition Source Average Standard deviation

Dependent variable

GDP Growth of per capita GDP WDI 2069.86 3293.117

Trade variables

(X + M)/GDP The sum of exports and imports as a WDI 70.02633 35.54374

share of GDP (TRADE)

X/GDP Exports as a share of GDP (EXPORT) WDI 29.31413 18.14106

M/GDP Imports as a share of GDP WDI 40.7122 23.04476

Human and physical capital variables

EDU Gross secondary enrollment rate WDI 30.3445 22.98946

POP Population growth rate WDI 2.505688 1.104741

INV Gross fixed capital formation WDI 2.15e + 09 6.54e + 09

Macroeconomic stability variables

INF Rate of change in price level WDI 52.19921 756.5322

DEBT External debt as share of GDP WDI 92.33425 125.7139

CRISIS Financial crisis, set at 0 before 2008 year WDI

and 1 after 2008

FD Private credit as a share of GDP WDI 17.50941 19.18839

Institutional quality variable

IGOV Governance index Kaufmann indicators (WDI) -.2723996 .5050246

Source: Author's calculations with data provided.

Appendix 2. List of countries in the sample 377 388

Country Country Country 382

Angola Gabon Niger 384

Benin Gambia, The Nigeria 386

Botswana Ghana Rwanda 388

Burkina Faso Guinea Senegal 390

Burundi Guinea-Bissau Seychelles 392

Cameroon Kenya Sierra Leone 394

Cape Verde Lesotho South Africa 396

Central African Republic Liberia Sudan 398

Chad Madagascar Swaziland 400

Comoros Malawi Tanzania 402

Congo, Dem. Rep. Mali Togo 404

Congo Rep. Mauritania Uganda 406

Cote d'Tvoire Mauritius Zambia 408

Ethiopia Mozambique Zimbabwe 410

Source: Author. 412

ARTICLE IN PRESS

P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx

References 415

Acemoglu, D., Robinson, J.A., 2012. Why Nations Fail: The Origins of Power, Prosperity, and Poverty. Crown Publisers, 416

Aghion, P., Howitt, P., Mayor-Foulkes, D., 2005. The effect of financial development on convergence: theory and evidence. 419

Q. J. Econ. 120 (1), 173-222. 420

Alesina, A., Spolaore, E., Wacziarg, R., 2000. Economic integration and political disintegration. Am. Econ. Rev. 90 (5), 421

1276-1296. 422

Almeida, R., Fernades, A., 2008. Openness and technological innovations in developing countries: evidence from firm-level 423

surveys. J. Dev. Stud. 44 (5), 701-727. 424

Arnone, M., Presbitero, A.F., 2010. Debt Relief Initiatives: Policy Design and Outcomes. In: Kirton, John (Ed.), Global Finance 425

Series. 426

Asiedu, E., 2003. Debt relief and institutional reform : a focus on heavily indebted poor countries. Q. Rev. Econ. Finance 43 (3), 427

614-626. 428

Baldwin, R.E., Braconier, H., Forslid, R., 2005. Multinationals, endogenous growth, and technological spillovers: theory and 429

evidence. Rev. Int. Econ. 13 (5), 945-963. 430

Barro, R.J., Sala-i-Martin, X., 1997. Technological diffusion, convergence, and growth. J. Econ. Growth 2 (1), 2-26. 431

Benhabib, J., Spiegel, M., 2005. Human Capital and Technology Diffusion. In: Aghion, P., Durlauf, S.N. (Eds.), Handbook of 432

Economic Growth. Elsevier, Amsterdam. 433

Bolaky, B., Freund, C., 2008. Trade, regulations, and income. J. Dev. Econ. 87 (2), 309-321. 434

Bond, E.W., Jones, R.W., Ping, W., 2005. Economic takeoffs in a dynamic process of globalization. Rev. Int. Econ. 13 (1), 435

1 -19. 436

Chang, R., Kaltani, L., Loayza, N.V., 2009. Openness can be good for growth: the role of policy complementaries. J. Dev. 437

Econ. 90 (1), 33-49. 438

Cohen, W., Levinthal, D., 1989. Innovation and learning: two faces of R&D. Econ. J. 99 (September), 569-596. 439

Dornbusch, R.S., Fischer, S., Kearney, C., 1996. Macroeconomics. The McGraw-Hill Companies, Inc., Sydney. 440

Edwards, S., 1993. Openness, trade liberalization, and growth in developing countries. J. Econ. Lit. 31 (3), 1358-1393. 441

Edwards, S., 1998. Openness, productivity and growth: what do we really know? Econ. J. 108 (447), 383-398. 442

Erfakar, M.E., 2011. Growth, foreign direct investment, trade and inflation: an empirical application on Turkey. Middle East. 443

Finance Econ. 9, 137-147. 444

Evans, P., 1997. How fast do economies converge. Rev. Econ. Stat. 79 (2), 219-225. 445

Fagerberg, J., 1994. Technology and institutional differences in growth rate. J. Econ. Lit. 32 (3), 1147-1175. 446

Falvey, R., Foster, N., Greenaway, D., 2012. Trade liberalization, economic crises, and growth. World Dev. 40 (11), 447

2177-2193. 448

Fischer, S., 1993. The role of macroeconomic factors in growth. J. Monet. Econ. 32 (3), 485-511. 449

Fosu, A.K., 1990. Exports and economic growth: the African case. World Dev. 18, 831-835. 450

Fosu, A.K., 1999. The external debt burden and economic growth in 1980s: evidence from sub-Saharan Africa. Can. J. Dev. 451

Stud. 20 (2), 307-318. 452

Fosu, A.K., 2013. Impact of global financial and economic crisis on development: whither Africa? J. Int. Dev. 25 (8), 453

1085-1104. 454

Greenaway, D., Morgan, W., Wright, P., 2002. Trade liberalization and growth in developing countries. J. Dev. Econ. 67 455

(2002), 229-244. 456

Grossman, G.M., Helpman, E., 1991. Innovation and Growth in the Global Economy. MIT Press, Cambridge, MA. 457

Haltiwanger, J., 2011. Globalization and Economic Volatility. In: Bacchetta, M., Jansen, M. (Eds.), Making Globalization 458

Socially Sustainable. ILO and WTO, Geneva, pp. 119-146. 459

Herzer, D., 2013. Cross-country heterogeneity and the trade-income relationship. World Dev. 44, 194-211. 460

Im, K.S., Pesaran, M.H., Shin, Y., 2003. Testing for unit roots in heterogeneous panels. J. Econ. 115 (1), 53-74. 461

Ismail, A., Zaman, K., Atif, R.M., Jadoon, A., Seemab, R., 2010. The role of exports, inflation and investment on economic 462

growth in Pakistan (1980-2009). Int. J. Econ. Res. 1, 1 -9. 463

Jouini, J., 2015. Linkage between international trade and economic growth in GCC countries: empirical evidence from PMG 464

estimation approach. J. Int. Trade Econ. Dev. 24 (3), 341-372. 465

Kao, C., 1999. Spurious regression and residual-based tests for cointegration in panel data. J. Econ. 90, 1-44. 466

Kaufmann, D., Kraay, A., Zoido-Lobaton, P., 1999. Aggregating Governance Indicators. Policy Research Paper 2195. The 467

World Bank. 468

Kim, D.-H., 2011. Trade growth and income. J. Int. Trade Econ. Dev. 20 (5), 677-709. 469

Crown Publishing Group, a Division of Random House, pp. 1-571. Agenor, P.R., 2004. Does globalization hurt the poor? IEEP 1 (1), 21-51.

ARTICLE IN PRESS

16 P. Zahonogo / Journal of African Trade xx (2017) xxx-xxx

Kim, D.-H., Lin, S.-C., 2009. Trade and growth at different stages of economic development. J. Dev. Stud. 45 (8), 1211-1224. 470 Kremer, S., Bick, A., Nautz, D., 2009. Ination and Growth: New Evidence from a Dynamic Panel Threshold Analysis. SFB 649 471 Discussion Paper 2009-036. 472

Krueger, A., Berg, A., 2003. Trade, Growth, and Poverty: a Selective Survey. IMF Working Paper, WP/03/30. International 473 Monetary Fund, Washington. 474

Le Goff, M., Singh, R.J., 2014. Does trade reduce poverty? A view of Africa. J. Afr. Trade 1 (2014), 5-14. 475

Lee, K., Pesaran, M.H., Smith, R., 1996. Growth and convergence in a multi-country empirical stochastic Solow model. 476 J. Appl. Econ. 12, 357-392. 477

Levine, A., Lin, C.F., Chu, C., 2002. Unit root test in panel: asymptotic and finite sample properties. J. Econ. 108 (1), 1-24. 478 Liang, Z., 2006. Threshold Estimation on the Globalization Poverty Nexus: Evidence from China. Research Paper. 479 UNU-WIDER United Nations University (No. 2006/57). 480

Lucas, R.E., 1988. On the mechanic of economic development. J. Monet. Econ. 46 (1), 167-182. 481

Maddala, G.S., Wu, S., 1999. A comparative study of unit root tests with panel data and a new simple test. Oxf. Bull. Econ. Stat. 482 61, 631-652. 483

Mankiw, N.G., Romer, D., Weil, D.N., 1992. A contribution to the empirics of economic growth. Q. J. Econ. 107 (2), 407-437. 484 McMillan, M., Verduzco, I., 2011. New Evidence on Trade and Employment: an Overview. In: Jansen, Marion, Peters, Ralf, 485 ManuelSalazar-Xirinachs, José (Eds.), Trade and Employment: from Myths to Facts. International Labor Organization, 486 Geneva, pp. 23-60. 487

Musila, J.W., Yiheyis, Z., 2015. The impact of trade openness on growth: the case of Kenya. J. Policy Model. 37 (2015), 488 342-354. 489

Newfarmer, R., Sztajerowska, M., 2012. Trade and Employment in fast-Changing World. In: Lippoldt, Douglas (Ed.), Policy 490 Priorities for International Trade and Jobs. OECD, Paris, pp. 7-73. 491

Pedroni, P., 1999. Critical values for cointegration tests in heterogenous panels with multiple regressors. Oxf. Bull. Econ. Stat. 492 S1 (61), 653-670. 493

Pesaran, M., Smith, R., 1995. Estimation of long relationships from dynamic heterogenous panels. J. Econ. 68, 79-114. 494 Pesaran, M.H., Shin, Y., Smith, R.P., 1999. Pooled mean group estimation of dynamic heterogeneous panels. J. Am. Stat. 495 Assoc. 94. 496

Rajan, R.G., Zingales, L., 2003. The great reversals: the politics of financial development in the twentieth century. J. Financ. 497 Econ. 69 (1), 5-50. 498

Redding, S., 1999. Dynamic comparative advantage and the welfare effects of trade. Oxf. Econ. Pap. 51 (1), 15-39. 499

Sachs, J.D., Warner, A., 1995. Economic reform and the process of global integration. Brook. Pap. Econ. Act. 1, 1-118. 500 Sindzingre, A., 2005. Explaining Threshold Effects of Globalization on Poverty: an Institutional Perspective. Research Paper. 501 UNU-WIDER United Nations University (No. 2005/53). 502

Singh, T., 2010. Does international trade cause economic growth? A survey. World Econ. 33 (11), 1517-1564. 503

Solow, R., 1957. Technical change and the aggregate production function. Rev. Econ. Stat. 39, 312-320. 504

Trejos, S., Barboza, B., 2015. Dynamic estimation of the relationship between trade openness and output growth in Asia. 505 J. Asian Econ. 36 (2015), 110-125. 506

Ulaçan, B., 2015. Trade openness and economic growth: panel evidence. Appl. Econ. Lett. 22 (2), 163-167. 507

Verpagen, B., 1991. A new empirical approach to catching up or falling behind. Struct. Chang. Econ. Dyn. 2 (2), 359-380. 508 Winters, L.A., 2004. Trade liberalization and economic performance: an overview. Econ. J. 114 (493), F4-F21. 509

World Bank, 2014. World Development Indicators (Washington D.C). 510

Young, A., 1991. Learning by doing and the dynamic effects of international trade. Q. J. Econ. 106 (2), 369-405. 511

Zeren, F., Ari, A., 2013. Trade openness and economic growth: a panel causality test. Int. J. Bus. Soc. Sci. 4, 317-324. 512