saving-investment connection: evidence from sub ...

International Business & Economics Research Journal – October 2010

Volume 9, Number 10

Regional Integration, Trade Openness, And Economic Growth: Causality Evidence From UEMOA Countries Douglas K. Agbetsiafa, Indiana University South Bend, USA

ABSTRACT The paper examines whether openness of the economy to international trade Granger causes growth in per capita real GDP or growth itself brings about associated increase in openness to trade in member countries of Union Économique et Monetaire Ouest-Africaine (UEMOA). Three measures of trade openness are employed and causality tests were applied using Johansen cointegration and vector error correction methodology, and the estimated results show that the direction of causality between trade openness and economic growth in these countries is sensitive to the choice of the indicator used for trade openness. Causality tests show a differing pattern of causality among the eight member countries. These results lend support to the hypothesis that openness of the economy to international trade promotes economic development and growth while growth itself brings about associated increase in openness to trade. These findings suggest that promoting exports and imports of intermediate goods, capital, in addition to improving the infrastructure, human capital, and institutional quality, will help to overcome deficiencies in the regional financial system and thereby affecting trade-growth dynamics favorably to contribute to sustainable economic growth and development in the UEMOA member countries. Keywords: trade openness, growth, causality, co-integration, UEMOA

I.

INTRODUCTION

T

he paper examines the relationship between trade openness and economic growth in eight member countries of West African Monetary Union (UEMOA) for which complete data were available using co-integration tests proposed by Johansen and Juselius and causality tests based on error-correction model as being country-specific time series data to examine the long-run equilibrium relationship and causality between five financial development indicators and economic growth. The study applies recent advances in cointegration techniques and country-specific time series data to examine the long-run equilibrium relationship between trade and economic growth in these eight member states of UEMOA. As a contribution to the literature, the current study extends the trade openness and economic development debate and its implications for regional integration. The monetary and financial dynamics that govern the relationship between trade openness and economic development in these countries could arguably shed some light on the efficacy of monetary union among these countries. The UEMOA(1) is a monetary union which encompasses most of France’s former colonies in the area. The current member states are Benin, Burkina Faso, Cote d’Ivoire, Mali, Niger, Senegal Togo, and Guinea Bissau who joined later in 1997. It forms part of the Franc Zone, the other main component of which is a second monetary union, the Economic and Monetary Community of Central Africa (CEMAC). The cornerstone of the Franc Zone is the use of currencies that the French Treasury guarantees to exchange for Euros at a fixed rate. All member states of the Union have in common the use of a common currency, the CFA franc. The enduring institutional link with the former colonial power gives the UEMOA countries an unusually high level of financial stability, compared to other African countries with similar levels of economic development.

55


Volume 9, Number 10

The Union has established a common accounting system, periodic reviews of member countries’ macroeconomic policies based on convergence criteria, a regional stock exchange, and regulatory framework for a regional banking system Six of the eight member countries are eligible for trade benefits under African Growth and Opportunity Act (AGOA), and four of these countries – Benin, Burkina Faso, Mali, and Senegal – are also eligible to receive AGOA’s textile and apparel. It is interesting to note that United States had 1.9 billion bilateral trade with UEMOA in 2009 making it the United State’s 74th largest goods and exports market. Although the structure of the financial sector and its institutional arrangements indicate that financial integration may be well advanced in some aspects, the disparity in the stages of economic growth among the member countries is reflected in the following growth statistics. While the regional average growth rate was 3.78 percent between 2004 and 2008, it was 4.67% for Benin, 5.43% for Burkina Faso, 5.1% for Mali, 4.83% for Niger, 4.83% for Senegal, 1.85% for Cote d’Ivoire, 3.10% for Guinea Bissau, and 2.63% for Togo. Substantial external debt of individual states remains one of the region's greatest challenges and, in addition, internal strife has adversely affected economic performance in some of the countries. Although existing studies in the literature have provided some important insights to the trade-growth relationship, a number of concerns about the conceptual and methodological approaches remain. First, many of the studies employed single equation ordinary least squares regression methodology to examine the relationship between openness to trade and economic growth and are therefore likely to suffer from simultaneous equation bias. Secondly, those studies that employed ordinary least squares regression analysis did so without examining the time series properties of trade and growth series. Indeed, Nelson and Plosser (1982) have shown that most macroeconomic time series data are nonstationary in their levels, but stationary when differenced. Thirdly, a number of these studies used cross-section data, thereby making it difficult to apply their findings to individual countries. Fourthly, most of the studies in extant literature concentrated on developed and developing regions of the world with little focus on the West African sub-region with its unique characteristics of low growth, high external indebtedness, and macroeconomic and structural imbalances. One of the objectives of the present study is to fill this gap in the empirical literature by providing evidence on the impact of outward-trade policies and economic development in the UEMOA zone. The research evidence would be useful to policy makers in trade reforms designed to revive and sustain economic growth and development both within the zone and across Africa. This is particularly important as development economists, policy makers, and governments of the region have embarked on serious efforts in integrating their economies. The paper employs recent advances in co-integration techniques and country-specific time series data to examine both short-run dynamics and the long-run equilibrium relationship between trade and economic growth for member countries of UEMOA. It is worth noting that in 2003, African central bankers announced a plan for a single currency and common central bank for the entire continent and regional economic unions, including the UEMOA, as an intermediate step toward a single African central bank and currency. A plan with such potential widespread economic and political consequences deserves a careful examination based on good research intelligence and analyses of the underlying growth dynamics and processes in each of the member countries in order to harness them for economic growth and development of the region. However, to date and to the present author’s knowledge, very little research has been done on these aspects of the plan toward a single economic and currency union of Africa. The rest of the paper is organized as follows: Section II reviews the empirical literature on trade-growth nexus, Section III specifies the VEC model after examining the time series properties of the trade and growth variables, Section IV discusses the empirical results, and Section V presents summary and policy implications. II.

LITERATURE SURVEY

International trade may induce economic growth in several ways, including by increasing a country’s level of specialization, and positively affecting innovation and technological diffusion (Harrison, 1996). Conversely, economic development may also trigger a country’s level of trade openness, e.g. with shifts in production and demand patterns as well as increased levels of international integration that accompany national industrialization experiences. Empirically, Edwards (1998) provides some evidence for the hypothesis that trade openness leads economic growth, finding that more open economies experience greater productivity growth. In contrast, Rodriguez and Rodrik (2001) find only limited support for a strong and positive link between openness and economic development. 56


Volume 9, Number 10

Various theories have been proffered to explain the link between international trade and economic growth. The classical and neoclassical theory seeks to explain the benefits from trade by contrasting the likely outcomes that might prevail with free trade than with autarky. Central to the classical and neoclassical theory is the notion of comparative advantage that would lead to an efficient reallocation of international resources, largely due to specialization and division of labor. In the conventional theory, the gains from trade come from the import side. However, exports have an indirect but pivotal role as “exports allow the country to “buy” imports of intermediate goods on more favorable terms than if produced at home” (Meier, 1995, p. 459). The neoclassical theory is essentially supply oriented (Federici and Marconi, 2002), though the link between exports and growth can be traced from some earlier works. In his demand-oriented theory of growth, Kaldor (1970) identifies foreign demand as the ultimate constraint on growth in an open economy. In the staple theory of growth, extensive growth of export of primary or staple product having comparative advantage is regarded as a source of higher growth rates of output. The trade-led growth hypothesis has received its latest boost from the endogenous growth theory which identifies four major sources of growth, namely, (i) increases in accumulation of capital goods; (ii) improvements in the quality of the labor force; (iii) reallocation of resources from low to high-productivity sectors; and (iv) technical change (Durlauf et al., 1996). The two basic differences between the new growth theory and the neoclassical theory are: (a) various social and economic policies are tipped to affect the growth rate in the new growth theory, while in the neoclassical theory, these policies affect only the level of real income not the growth rate; and (b) in the neoclassical theory, physical capital is deemed as a transitory source of growth as it is subject to diminishing returns, while in the new growth theory, both physical and human capital are assumed to exhibit increasing returns to scale. The source of increasing returns is knowledge, which is created by acquisition of skills through education and training, learning by doing and innovation or research and development (Grossman and Helpman, 1991). Aghion and Howitt (1992) hold that industrial innovations resulting in new and improved intermediate products have positive implications for changes in technology that in turn stimulate growth. Further, the secondary effects of learning by doing or investment in education may even exceed the direct effects on growth (Coe and Helpman, 1995). Grossman and Helpman (1991) describe interactions with the outside world as the most important mechanism for promoting innovation and growth in a small economy. The exchange of ideas can occur through personal contacts and use of imported intermediate products. Romer (1993) also holds a similar view. Roemer explains the poorness of a developing country in terms of its lack of physical objects such as factories, and roads (`Object gaps’) and lack of ideas or knowledge (`Idea gaps’) to create values compared to a developed country. An object gap is manifested in savings and capital accumulation while an idea gap includes insights about packaging, marketing, distribution, payment systems, quality control as well as the technology gap among others. An outward oriented trade policy inevitably involves interactions with foreign firms and agents and, hence, can help reduce the idea gaps. The conventional view of the trade-growth nexus argues that unidirectional causality runs from trade to economic growth. Proponents of this view include Helpman and Krugman (1985), Krueger (1985), Rodrik (1988), Vorvodas (1973), Afxentiou and Serletis (1992), and Yaghman (1994). A second view hypothesizes that unidirectional causality runs from economic growth to trade. According to this view, higher income growth facilitated by higher productivity would result in lower unit costs which would translate to higher exports. In the event that domestic production rises at a faster pace than domestic demand, producers will sell their goods in foreign markets. Proponents of this view include Kaldor (1967); Ghartey, (1993), Sharma and Dhakal, (1994). A third view argues that there can be bidirectional causality between trade and economic growth. Proponents of this view include Helpman and Krugman (1985); Kunst and Mann (1989); Ghartey (1993); Sharma and Dhakal (1994). According to a fourth view, there is no relationship between trade and economic growth. Pack (1992) contends that exports and economic growth are the outcome of many complex processes of development and structural change that occur in the economy. Support for the trade-led growth hypothesis is not, however, universal. Young (1991) argues that learning by doing effects may slow down at later stages of economic development and can even stop eventually if not reinforced by new technical progress. This echoes the general sentiment that in an uncertain world market, reliance on exports alone may not necessarily lead to a sustained long-term growth in a developing country and that the markets in the industrialized world may not be large enough to absorb these additional exports from the developing 57


Volume 9, Number 10

countries (Adelman, 1984; Cline, 1984). Furthermore, export promotion and import substitution policies may be complementary with the latter being a springboard for export-based growth (Hamilton and Thomson, 1994; Grabowski, 1994). The body of empirical literature investigating the relationship between a country’s openness to trade and its economic growth continues to grow. A number of cross-country studies find that the ratio of exports to GDP or some other measure of openness is a significant determinant of growth. Yet, as Giles and Williams (2000) correctly points out, “the focus of the debate is on whether or not a country is better served by orienting trade policies to export promotion or import substitution” (p. 262). However, export promoting policies would cause an outflow of resources from import competing sectors of the economy to the export sector, and a resulting increase in imports. On the other hand, if import barriers are lowered, resources would flow out of the import competing sector into the export sector, and thereby promoting the export sector. The impressive growth performances of countries like Korea, Hong Kong, Singapore, Taiwan, Malaysia, Thailand, India and China have prompted many to place trade policy as a fundamental element of economic development planning (See, Krueger, 1998; Sachs and Warner, 1995). Rodrik (1998) examined the direct link between trade policy and income growth to determine how much of the variation in Africa’s trade performance was attributable to terms of trade, geography, and economic policy. He finds geography, fiscal policy, and export taxation as important factors in explaining Africa’s trade performance. He also tested for the impact of trade on the long-term growth of output in Africa, and found that growth could be explained by human resources, demography, and fiscal policy, but not trade policy. Based on this, evidence he concluded that within Africa, trade policies have had the expected strong effect on the volume and growth of trade, but have played a significantly smaller role in stimulating income growth than earlier results suggest. Hoeffler (1999) also fails to find any convincing evidence that trade influences the long-term growth rate in Africa. Applying panel data of African countries to an augmented Solow growth model over the period 1970 to 1995, Hoeffler finds that investment rate, education levels, population growth, and initial output almost fully account for the observed growth of output. This result implies that the main impact of trade on long-term income growth can only occur indirectly through its impact on accumulation of human and physical capital. A study by Onafowora and Owoye (1998) also explores the relationship between trade and growth using a vector-error correction model that includes output, exports, investment and an indicator for trade policy for twelve African countries. Their goal was to determine whether export growth stimulated investment, thereby raising the rate of income growth. Results of their study show wide variations of statistical significance of the included variables as well as the explanatory power of the model across the twelve countries. The authors concluded that the export-led growth framework based on an explicit commitment to outward-looking policies and export expansion has had some success in some of the African countries. The preceding studies have shown that most empirical analyses of the relationship between trade and growth do not make allowance for possible indirect channels through which trade may be linked to economic growth. Recognizing the indirect channels may be accomplished by treating many of the system variables as endogenous. Recent research by Ndulu and Ndung’u (1998) has moved in this direction. For example, they specify a simultaneous equations model that examines the effects of trade policies on export and import shares and how these variables are linked to income growth. They used panel data for selected African countries, a random effects estimation of single equations for export share, import share, and income growth, with explanatory variables as export and import prices; trade taxes; a measure for terms-of-trade shocks; indicators for the quality of institutions and civil unrest; changes in external debt; foreign direct investment; inflation; the ratio of investment to GDP; the real exchange rate; and lags of the trade shares. From each equation, they dropped the variables with insignificant coefficients in the random effects estimation. Using Full Information Maximum Likelihood (FIML), they estimated three equations as a system. Their results show that trade policies affect exports and imports, and thus output, indirectly through the real exchange rate. The co-authors also find that the influence of trade reforms is transmitted through the growth of investment, and that the growth of output and its lag do significantly affect the export and import shares. 58


Volume 9, Number 10

Other studies like Jung and Marshall (1985) reviewed a total of eleven empirical studies published between 1967 and 1982, and found them supporting the trade–led growth hypothesis. In their review of the findings of 14 empirical studies published between 1977 and 1983, Greenaway and Sapsford (1994) found twelve of the studies supported the export-led growth hypothesis. Similarly, Giles and Williams (2000) compile more than 150 studies that include cross-country studies between 1963 and 1999 and time series studies between 1972 and 1999. Their results show that only four out of a total of fifty-seven cross-country studies show evidence of no significant causal relationship between trade and growth. Similarly, in only 10 of the 102 time-series studies, the findings indicate no clear causal relationship between exports and growth. In about as many cases, the evidence supports only the growth-led trade hypothesis. While other researchers like Chang et al. (2000) find no support for the trade-led growth theory for Taiwan, Hatemi-J and Irandoust (2000) find positive relationship between trade and growth for Ireland, Mexico and Portugal, and no causal relationship for Greece and Turkey. Other empirical findings that support the trade-led hypothesis include Ghirmay et al. (2001) with a finding of strong positive relationship between trade and growth for a number of developing countries, Greenway et al. (2002) who found positive impact of trade liberalization on the economic growth of the developing countries. The empirical evidence on the trade-growth relationship is clearly mixed, an indication that such a relationship is significantly more complicated than revealed in the standard single equation analyses. This condition provides an opportunity for more research to identify and understand both direct and indirect associations between trade and growth, inter-temporal effects, and the channels through which these effects are transmitted. III.

MODEL, DATA AND METHODOLOGY

Various measures of openness have been used in the literature. For example, while Romer et al (1989) used ratio of exports as a measure of openness to trade in testing the relationship between exports and economic growth for ex-post industrialized countries, Ram (1990) employed ratio of imports and Liu, Song, and Romilly (1997) used ratio of the sum of exports and imports as the indicator of openness. Romer (1993) argues that for timeseries analysis, the imports/GDP ratio is generally acknowledged in the literature to be the best measure currently available. The present study employs all three measures of trade. Annual data for exports, imports and per capita GDP growth rate covering the period 1970 through 2007 were derived from various monthly issues of IMF’s International Financial Statistics and the World Bank’s Africa Development Indicators database (2008). A prerequisite in applying the co-integration procedure is to test the unit root properties of the series. As a next step, a unit root test is employed to check if the considered time series are stationary, i.e. I(0), or first difference-stationary, i.e. I(1). The existence of unit roots in the considered series may contaminate the findings of the causality analyses because of the properties of non-stationary time series. To this end, the paper uses both the Augmented Dickey-Fuller and Phillips-Perron unit root tests to determine the order of integration of the series. The PP unit root test is also used based on Choi and Chung (1995) who argue that this test is more powerful when low sampling frequency data, i.e. annual data is used, compared to the standard unit root tests developed by Dickey and Fuller (1979, 1981) on which the Phillips and Perron (1988) builds. As co-integration is one of the important questions in the Vector Error Correction model (VEC), the next step is to estimate the long-run equilibrium relationship between each of the openness measures and per capita GDP growth rate utilizing the co-integration tests proposed by Johansen and Juselius (1988). It is known that cointegration relationships are unstable in small samples (Berg and Borensztein 200a), but as Granger (1987) stresses, models estimated in first differences while the data are actually cointegrated will be misspecified (which will produce an omitted–variable bias). Here the VEC model uses Maximum Likelihood (ML) method in the Johansen procedure to find cointegrating relationship(s) becomes essential, because in small samples ordinary estimation of cointegrating regressions becomes sensitive to the choice of the dependent variable. With the Maximum Likelihood method, this problem does not arise (Maddala 1992). Third, vector error-correction model based causality tests are implemented to ascertain the direction of causality between trade and economic growth. In conducting cointegration tests, the series are required to be non-stationary in their levels, and the cointegrating equation must have the same order of integration. Thus, before estimating the VEC model, the paper carried out diagnostic tests for unit roots by employing both the ADF and Phillip’s-Perron tests for stationarity. The estimated equation takes the form: 59


ΔXt=α +γXt-1 + δt +i=1 θΔXt-1 + εt p

Volume 9, Number 10 (1)

0

where, Δ is the difference operator, t is the time trend, and εt is the stationary random error, and the maximum lag length is p. The parameter of interest is γ . If γ = 0, the series contains a unit root. The co-integration procedure (Johansen, 1988; Johansen and Juselius, 1990) was applied in order to ascertain whether trade and economic growth are cointegrated for each of the eleven countries. The co-integration tests provide two likelihood ratio (LR) test statistics, the maximal eigenvalue (λ-max) of the stochastic matrix, and the trace test statistics. For (λ-max) and trace statistics, the null hypothesis is that H0: rk (П) = r against H1: rk (П) = r +1 and H0: rk (П) = r against H1: rk (П) ≥ r +1, respectively. Note that rk is rank, П is a matrix of long-run responses, and the matrix П has rank r, rk (П) =r. If the data cointegrated, П must be of reduced rank, r < N, where N is the number of variables. The paper employs both the trace and eigenvalue tests to determine whether r ≤ 1 cointegrating vectors are present in the system against the alternative hypothesis that the system is already stationary. The existence of at least one cointegrating vector in the system indicates the presence of causality between trade and economic growth. In order to test the causal relationship between trade and economic growth, it is common to apply the Granger causality test (see Granger 1969, and Sims 1972). Moreover, the co-integration technique pioneered by Engle and Granger (1987) and Granger (1986) makes a significant contribution towards testing causality. In this study, Granger causality test based on vector error-correction model is implemented. This procedure is preferred to the standard vector autoregressive model because it permits temporary causality to emerge from (a) the sum of the lagged coefficients of the explanatory differenced variables and (b) the coefficient of the lagged error-correction term. In addition, the error-correction model allows causality to emerge even if the lagged differences of the explanatory variables are not jointly significant (see Granger 1988, Miller and Russek 1990, Miller 1991, and Garcia and Zapata 1991). The error-correction model reintroduces the long-run information lost through differencing of time series data with unit roots as well as providing additional channel of Granger causality so far ignored by the standard causality tests of economic time series data. Following Granger (1969), variable (X) is said it Granger cause variable (Y) if and only if (Y) is predicted well using the past history of (X), together with the past history of (Y) itself, rather than by using just the past history of (Y). Accordingly, a bivariate vector error-correction model comprising trade and economic growth is represented in equations 2;

Δ(Trd)=αZt-1 + i=1 βi Δ(Trd)t-1 + a

b i=1

πi

Δ(Grw) t-1 + εt

Δ(Grw)=φZt-1 + j=1 1Δ(Grw)t-1+ λjΔ(Trd) t-1 + εt c

d

j=1

(2) (3)

where, Zt-1 represents the error correction term, the measures of openness to trade (Trd) are exports/GDP (Trx), imports/GDP (Trm), and (exports + imports)/GDP (Trt) respectively, and (Grw) is the growth of GDP per capita, a, b, c, d represent the optimal lags lengths obtained from the Akaike Information Criterion (AIC). In equation (2), the rejection of the null hypothesis that growth does not Granger cause trade requires that (i) the π i's co-jointly be statistically significant and/or (ii) the error correction term Zt-1 be statistically significant. Similarly, in equation (3), the null hypothesis that trade does not Granger cause growth is rejected if the i’s are jointly statistically significant, and/or the error-correction term Z t-1 is statistically significant. IV.

RESULTS AND DISCUSSION

Unit root results are summarized in Table 1. The null hypothesis of nonstationarity of the measures of trade and economic growth is tested against the alternative hypothesis of stationarity. Results of Dickey-Fuller (DF), Augmented Dickey–Fuller (ADF), and Phillips-Perron unit root tests, for the eight member countries of the Union, show that the levels of the variables are non-stationary, while their first differences are stationary, with or without a deterministic trend. According to these results, the null hypothesis of unit root cannot be rejected for the trade and growth series at 5% significance level or better. On the other hand, the null hypothesis of non-stationarity is strongly rejected for all variables in their first differences at better than 5% significance levels. 60


Volume 9, Number 10

Table 1: Dickey Fuller and Phillips-Perron Unit Root Statistics Levels First Difference Country Variable DF ADF P-P DF ADF P-P Benin Grw -2.43 -2.33 -1.91 -6.56 -4.75 b -5.26a Trx -2.64 -2.72 -2.49 -5.38 -4.93 b -4.20b b Trm -2.21 -2.76 -3.06 -5.98 -4.51 -6.10 a Trt -2.89 -2.95 -2.89 -5.00 -4.76 b -4.96 b Burkina Faso Grw -2.60 -2.55 -3.54 -3.77 -3.69 b -7.41 a b Trx -2.07 -2.33 -2.58 -4.60 -4.49 -8.35 a Trm -2.64 -3.09 -2.96 -5.38 -5.21 a -9.35 a b Trt -2.34 -2.78 -2.60 -5.11 -4.97 -9.3.7 a Guinea Bissau Grw -1.19 -1.95 -2.89 -6.50 -4.81 -13.96 Trx -2.77 -1.16 -2.67 -7.59 -7.69 -8.60 Trm -1.20 -1.49 -2.38 -5.75 -6.43 -7.24 Trt -1.72 -1.11 -3.03 -3,29 -6.74 -7.40 Ivory Coast Grw -1.43 -2.40 -2.42 -3.72 -3.67 b -3.71 b Trx -2.56 -2.68 -2.51 -5.02 -4.84 b -8.16* a b Trm -2.21 -2.83 -1.87 -4.40 -4.30 -4.20 b a Trt -1.57 -1.94 -1.90 -6.39 -6.23 -6.23 a a Mali Grw -1.96 -2.10 -2.11 -5.27 -5.15 -5.11 a Trx -2.77 -2.76 -3.22 -5.85 -5.65 a -9.24 a a Trm -2.80 -2.45 -2.96 -6.67 -6.55 -9.57 a b Trt -2.45 -2.54 -3.45 -4.27 -4.51 -9.26 a Niger Grw -2.35 -2.73 -2.46 -5.41 -4.53 b -5.40a Trx -3.50 -2.19 -3.57 -8.67 -6.16 a -9.16 a Trm -1.54 -1.08 -1.51 -7.13 -3.57 b -7.02a Trt -2.14 -1.27 -2.07 8.40 -4.19 b -8.08 a Senegal Grw -0.55 -2.03 -1.53 -5.41 -5.36 a -3.94a a Trx -1.43 -1.88 -1.65 -7.54 -8.25 -9.81a Trm -0.39 -0.71 -0.58 -6.65 -7.05 -7.57a Trt -0.69 -1.28 -0.74 -7.27 -8.04 -13.48a Togo Grw -1.30 -2.98 -1.14 -7.26 -1.07 -6.76 a Trx -2.99 -2.44 -3.04 -6.25 -2.81 -9.83a Trm -2.20 -2.38 -2.25 -5.22 -2.10 -8.83 a Trt -2.47 -2.71 -2.53 -5.53 -2.33 -9.56 a Note: The MacKinnon (1996) critical values of the ADF at the 5% significant level is -3.53 and the critical value for the Phillips Perron statistics is -3.55. The letter a, b indicates the 1%, and 5%t significance levels respectively.

Table 2 shows the co-integration results from the bivariate VEC model. Starting with the trace test statistic, the null hypothesis of no co-integration is rejected in all 11 countries at the 5 percent significance level or better. The calculated trace test statistics range from a low of 16.12 in Ivory Coast (Trm) to a high of 43.98 in Niger (Trm). The λ-max test results also reject the null hypothesis of no co-integration in favor of co-integration in all countries at the 5 percent significance level. The calculated test statistics range from a low of 12.98 in Ivory Coast to a high of 36.63 in Niger. The critical level is 12.98 at the 10 percent significance level for all countries. When a cointegration relationship is present, growth and trade openness share a common trend and a long-run equilibrium path. Accordingly, the co-integration results in all eight countries suggest a long-run equilibrium relationship between trade openness and economic growth. Additionally, evidence of co-integration implies that Granger causality must exist between each of the three measures of trade and growth series in at least one direction. The next section presents causality test statistics and results.

61


Volume 9, Number 10

Table 2: Johansen Co-integration Statistics Null: r =0 Null: r  1 Trace Maximum Trace Maximum Country Variable Eigenvalue Eigenvalue Benin Trx 23.14a 21.58a 1.55 1.55 a Trm 25.34 17.45b 7.89 7.89 Trt 25.72a 15.89b 9.83 9.83 Burkina Faso Trx 27.89a 22.43a 5.47 5.47 Trm 35.18a 27.39a 7.79 7.79 Trt 29.82a 20.89a 8.43 8.43 Guinea Bissau Trx 32.81a 29.85a 5.26 5.22 Trm 15.93b 15.53b 0.38 0.38 Trt 18.95b 17.41b 1.54 1.54 Ivory Coast Trx 31.00a 23.00a 7.99 7.99 Trm 16.12b 12.98 c 3.13 3.13 Trt 22.73a 13.63 c 9.09 9.09 Mali Trx 22.10b 17.36b 4.74 4.74 Trm 23.63b 17.05b 6.58 6.58 Trt 27.60a 20.92a 6.69 6.69 Niger Trx 21.78b 13.87 7.90 7.90 Trm 43.98a 36.63a 7.38 7.38 Trt 32.89a 24.73a 8.16 8.16 Senegal Trx 30.59a 16.57b 1.42 1.42 Trm 23.01b 18.25b 4.75 4.75 Trt 26.01a 16.04b 9.96 9.96 Togo Trx 26.69a 26.42a 0.27 0.27 Trm 22.53b 19.26b 3.27 3.27 Trt 24.73a 17.25b 7.47 7.47 Note: All calculations are carried out by Eviews 5.1. The letter a, (b or (c) indicates the 1%, 5%, and 10 % significance levels respectively. The lag orders of the underlying VAR were chosen with the AIC, where maximum lag was 4.

Table 3 below presents the causality results. According to Granger’s (1969) definition of non-causality, if one is able to better predict a series xt when including information from a series yt instead of only employing lagged values of xt , then yt Granger causes xt, and bidirectional causality or feed is present when xt also causes yt. Interestingly, differing patterns of causality emerge from the causality test results. When exports serve as the measure of trade openness, three countries – namely, Burkina Faso, Ivory Coast, and, Niger - show evidence of causality emanating from exports to growth and vice versa. In Benin, Guinea Bissau, Mali, Senegal and Togo, the evidence shows that growth leads trade openness at the 5 per cent significance level or better with the error correction term and the lagged differences as the channels of causation. On the other hand, when imports serve as a proxy for trade openness, the evidence shows bidirectional causality in Ivory Coast, Guinea Mali and Niger, and uni-directional causality from trade to growth in Guinea Bissau, with the lagged differences as the main channel of causality. In Benin, Burkina Faso, Niger, and Senegal, causality runs from growth to trade, with the error-correction term as the major channel of causality. When total trade is used, results show bidirectional causality between trade and growth in Ivory Coast, Mali, and Senegal, and causality running from growth to trade in Benin, Burkina Faso, Guinea Bissau, Niger, and Togo at the 5 percent significance level or better.

62


Volume 9, Number 10

Table 3: Causality Test Results Openness does not cause growth Country

Series



b i=1

π

i

Δ(Grw)

Growth does not cause openness 

Zt-1

t

d j=1

λΔ(Trd) j

Zt-1

c

2.89 (.09)c

Benin

Trx

1.89 (.20)

2.12 (.16)

2.80 (.08)

Burkina Faso

Trm Trt Trx

0.70 (.51) 1.24 (.30) 0.52 (.47)

0.38 (.89) .000 (.94) 3.95 (.01)a

0.12 (.88) 0.27 (.76) 3.07 (.08)b

2.68 (.05)b 3.23 (.05)b 0.14 (.93)

Trm Trt

1.28 (.33) 0.13 (.87)

0.18 (.67) 0.28 (.59)

2.75 (.05)b 0.85 (.43

5.83 (.02)b 4.15 (.05)b

Trx

1.03 (0.36)

1.88 (0.17)

3.37 (0.05)a

1.35 (0.27)

Trm

0.88 (0.42)

2.59 (0.09)

c

1.20 (0.31)

0.09 (0.90

Trt

1.08(0.35)

0.49 (0.61)

1.88 (0.17)

3.91 (.0.03)a

Trx

3.53 (.06)b

1.65 (.20)

0.20 (.65)

4.33 (.04)b

Trm Trt Trx Trm Trt Trx

a

4.09 (.02) 2.69 (.05)b 0.74 (.48) 0.60 (.55) 0.79) (.46) 0.61 (.55)

2.33 (.11) 1.38 (.26) 0.19 (.97) 2.56 (.06)b 2.76 (.05)b 3.85 (.03)b

b

3.47 (.04) 2.76 (.05)b 3.24 (.05)b 2.59 (.09)c 2.99 (.06)b 4.22 (.02)a

4.57 (.01)a 2.58 (.07)c 2.38 (.07)b 0.63 (.72) 1.84 (.16) 0.53 (.58)

Trm Trt

0.26 (.85) 0.75 (.53)

1.26 (.42) 0.51 (.53)

4.99 (.05)b 3.31 (.03)b

3.22 (.02)a 2.58 (.06)b

Trx Trm

2.45 (.10)c 0.37 (.69)

0.69 (.50) 0.18 (.90)

6.91 (.00)a 3.53 (.00)a

0.57 (.56) 3.91 (.01)a

Trt Trx Trm Trt

2.30 (.00)a 0.28 (.75) 0.23 (.79) 0.25 (.77)

0.09 (.91) 0.37 (.68) 0.58 (.56) 1.99 (.12)

7.50 (.00)a 0.03 (.96) 1.19 (.31) 0.22 (.80)

0.09 (.91) 4.08 (.02)a 0.11 (.89) 2.82 (.04)b

Guinea Bissau

Ivory Coast

Mali

Niger

Senegal

Togo

Note: 1). Zt-1 is the coefficient of the error correction term, and  dj=1 λjΔ(Trd) ,  i=1 π Δ(Grw) are the F statistics of the lagged independent variables. 2 The letter a, b, c indicates 1%, 5%, and 10% significance levels respectively. The p-ratios are in parenthesis. b

t-1

i

Before presenting summary and conclusions of the study, model adequacy is discussed. Tests of residual serial correlation uniformly indicate that the errors are white noise. The column labeled “LM” in Table 4 contains the marginal significance levels obtained in testing for the second order serial correlation using a Lagrange Multiplier procedure.1 The results of a test of constant residual variance against the alternative of autoregressive conditional heteroscedasticity are reported under the column “HET” in Table 4. This null hypothesis similarly could not be rejected. Also, the RESET tests reveal some evidence of model form misspecification bias in parameter estimates, except in Benin, Mali, Mauritania, Niger, Niger, Senegal and Togo.

63


Country Benin

Volume 9, Number 10

Table 4: Tests of Model Adequacy Variables in Equations HET RESET Grw & Trx 9.53 (.00) 2.52 (.60)

Burkina Faso

Guinea Bissau

Ivory Coast

Mali

Niger

Senegal

Togo

LM S 3.39 (.49)

Grw & Trm

9.52 (.54)

2.25 (.50)

2.17 (.70)

Grw & Trt

9.14 (.36)

0.89 (.32)

5.94 (.20)

Grw & Trx

2.61 (.08)

0.81(.45)

1.40 (.82)

Grw & Trm

0.75 (.47)

3.89 (.03)

2.90 (.57)

Grw & Trt

1.03 (.36)

3.60 (.04)

2.27 (.68)

Grw & Trx

1.75 (.19)

0.18 (.82)

1.28 (.67)

Grw & Trm

0.36 (.70)

0.42 (.65)

5.2 (0.05)

Grw & Trt

0.40 (.67)

0.45 (.63)

5.63 (.05)

Grw & Trx

0.83 (.65)

9.25 (.00)

8.09 (.06)

Grw & Trm

0.89 (.60)

9.78 (.00)

2.58 (.62)

Grw & Trt

1.22 (.34)

9.96 (.00)

2.23 (.69)

Grw & Trx

0.56 (.86)

3.85 (.03)

2.99 (.56)

Grw & Trm

1.36 (.30)

0.97 (.41)

2.49 (.64)

Grw & Trt

1.27 (.31)

1.34 (.27)

2.10 (.71)

Grw & Trx

2.14 (.06)

1.32 (.27)

5.80 (.21)

Grw & Trm

0.66 (.74)

9.39 (.00)

3.53 (.47)

Grw & Trt

217 (.06)

0.82 (.49)

5.36 (.25)

Grw & Trx

1.80 (.15)

3.01 (.06)

5.90 (.20)

Grw & Trm

1.61 (.20)

3.25 (.05)

3.32 (.50)

Grw & Trt

0.49 (.91)

1.50 (.22)

4.27 (.36)

Grw & Trx

1.11 (.46)

2.44 (.05)

2.10(.71)

Grw & Trm

1.49 (.23)

3.83 (.01)

1.62 (.80)

Grw & Trt

1.38 (.28)

2.33 (.07)

3.88 (.42)

Notes: 1. White HET: Test for heteroscedasticity based on squared residual. 2. Ramsey RESET: tests for functional form misspecification. 3. LMS: Lagrange multiplier test for residual serial correlation 4. The marginal significance levels are shown in parentheses

V.

CONCLUSIONS

The paper applies a vector error-correction model to West African data in order to examine the long-run equilibrium relationship between trade openness and economic growth and their causal relationship in UEMOA member countries. Among the research findings are first, existence of a steady-state, long-run relation between trade and growth in the majority of the countries based on the Johansen co-integration results. Both the trace and the maximal eigenvalue (λ-max) test statistics reject the null hypothesis of no co-integration in all eight countries at the 5 percent significance level or better. Second, Granger non causality tests suggest differing patterns of causality. Causality tests suggest a bidirectional causality between trade (Trx) and growth in Burkina Faso, Ivory Coast, and Niger, and a unidirectional causality from growth to trade (Trx) in Benin, Guinea Bissau, Mali, Senegal and Togo. When imports (Trm) is the measure of trade openness, the evidence shows bidirectional causality between trade (Trm) and growth in Ivory Coast, Mali, and Niger, and uni-directional causality from growth to trade (Trm) in Benin, Burkina Faso, Niger, and Senegal. With total trade (Trt) as a measure of openness, the evidence supports 64


Volume 9, Number 10

bidirectional causality in Ivory Coast, Mali, and Senegal and a one-way causality running from growth to trade (Trt) in Benin, Burkina Faso, Guinea Bissau, Niger, and Togo. In the other three countries - namely, Ivory Coast, Mali, and Senegal - unidirectional causality flows from trade to growth. In sum, trade openness appears to have a positive impact on growth in all member countries with the exception of Benin, Guinea Bissau, and Togo; while growth has a positive causal impact on trade openness in all countries, except Ivory Coast. These different patterns of causality are consistent with findings by other researchers, such as Grossman and Helpman, 1991; Bhagwati, 1988; Sachs and Warner, 1997; Onafowora and Owoye, 1998; Fosu, 1990, and Ghartey (1993). Results are also consistent with the notion that openness influences a developing country’s growth of per capita gross domestic product by impacting the level of export and import activities and economic growth. Furthermore, trade may lead to improvement of each country’s growth rate by allowing importation of capital and intermediate goods and facilitating the transmission of knowledge which can be used to adapt and imitate developed country products. From a policy perspective, the findings of the present study suggest export promotion, including manufactured exports and importation of intermediate goods, including capital and innovation all of which may contribute significantly to a sustained economic growth projectile of the countries and the monetary zone. Therefore, trade policy must remain one of the policy pillars of development plans designed to accelerate the speed of integration of domestic markets within the region and with the developed world economy. In addition, improvements in infrastructure, human capital development, institutional quality and regional legal systems are needed to overcome some of the deficiencies in the regional financial system, thereby affecting the trade-growth dynamics favorably. It is heartening to note that many of the countries in the Union have already embarked on these changes and other policies to attract foreign investment, improve the business climate, and reform the commercial, legal, and the financial systems throughout the sub-region. The need for cooperation and support by international development institutions and governments is now more urgent than ever in several areas, including access to developed world markets for manufactured and traditional exports. AUTHOR INFORMATION Douglas Agbetsiafa, Professor and Chair of Economics, a University of Notre Dame Ph. D. was a University of Ghana scholar, the Bank of Ghana scholar, and an International Development Agency scholar. He is the recipient of several teaching awards including: the IU South Bend Distinguished Teaching Award, FACET (Faculty Colloquium on Excellence in Teaching), Trustees’ Teaching Award, and several other teaching and research awards. His research interests include international finance, international trade and development, financial intermediation, and teaching pedagogy. In addition, he has made numerous professional research presentations at national and international conferences, including a recent invited presentation at Oxford University, Oxford, England. REFERENCES 1. 2. 3. 4.

5. 6. 7.

Adelman, I., 1984. Beyond Export-Led Growth. World Development 12:937-49. Afxentiou, P.C. and Serletis, A, 1991. “Exports and GNP Causality in the Industrial Countries: 19501985,” Kyklos, 44, 2 (May): 167-79. Balassa, B., 1985. “Exports, Policy Choices, and Economic Growth in Developing Countries after the 1973 Oil Shock.” Journal of Development Economics 18 (1):23-35. Baumgartner, Josef, Raman Ramaswamy, and Goran Zettergren, 1997. “Monetary Policy and Leading Indicators of Inflation in Sweden.” IMF Working Papers No. 34. Washington D.C.: International Monetary Fund. Bereg, Andrew, Eduardo Borensztein, and Zhaohui Chen., 1997, “Dollarization, Exchange rates, and Monetary Policy.” Washington, D.C.: International Monetary Fund. Bhagwati, J.N, 1988. “Export-promoting Trade Strategies: Issues and Evidence.” The World Bank Research Observer 3:27-57. Brown, R. L., J. Durbin, and J. M. Evan (1975). “Techniques for Testing the Constancy of Regression Relationships Over Time,” Journal of Royal Statistical Society, Series B, 37, 149-192. 65

International Business & Economics Research Journal – October 2010 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.

31. 32. 33. 34.

35.

Volume 9, Number 10

Chang, T., Fang, W., Liu, W. 2000. “Exports, Imports and Income in Taiwan: An Examination of the Export-Led Growth Hypothesis.” International Economic Journal 14 (2):151-60. Choi, I. and Chung, B.S. (1995) “Sampling frequency and power of tests for a unit root: A simulation study.” Economic Letters 49, 131-136. Cline, W., 1984. Exports of Manufactures from Developing Countries, Washington, Brooking Institution. Dickey, D. A., and W. A. Fuller (1979). "Distribution of Estimates of Autoregressive Time Series with Unit Root." Journal of the American Statistical Association, 427-431. Durlauf, S., Raj, B., Helliwell, J., 1996. “Long Run Economic Growth.” Empirical Economics 21(1):1-10. Engle, Robert F., and C. Granger. (1987). “Co-Integration and Error Correction Representation, Estimation, and Testing.” Econometrica 55, no. 2 (March): 251-276. Fosu, A.K. 1990. “Exports and Economic Growth: The African Case,” World Development, 18, 6 (June): 831-35. Ghirmay, T., Richard Grabowski, and Subhash C. Sharma. 2001. “Exports, Investment, Efficiency and Economic Growth in LDC: An Empirical Investigation.” Applied Economics, 33:689-700. Giles, J.A., Williams, C.L., 2000. “Export-Led Growth: A Survey of Empirical Literature and Some NonCausality Results Part-I.” Journal of International Trade and Economic Development, 9 (3):261-337. Grabowski, R., 1994. “Import Substitution, Export Promotion and the State in Economic Development.” Journal of Developing Areas, 28:535-54. Greenaway, D., Morgan, W., Wright, P., 2002. “Trade Liberalization and Growth in Developing Countries.” Journal of Development Economics, 67:229-244. Greenaway, D., Sapsford, D., 1994. “What Does Liberalization Do for Exports and Growth?” Weltwirtschaftliches, 130 (1):152-74. Grossman, Gene, and Elhanan, Helpman1991a. Innovation and Growth in the Global Economy, the MIT Press, Cambridge, MA. Grossman, Gene, and Elhanan, Helpman 1991b. “Trade Knowledge Spillovers and Growth.” European Economic Review 35, no. 2-3, April. 517-526. Hamilton, N., Thomson, C., 1994.”Export Promotion in Regional Context: Central America and Southern Africa.” World Development, 22:1379-92. Harrison, A., (1996) “Openness and Growth: A Time –series, cross-country analysis for developing countries.” Journal of Development Economics 48, 419-447 Hoeffler, A. E. (1999) "The Augmented Solow Model and the African Growth Debate." University of Oxford: Centre for the Study of African Economies, March, 1999. Johansen, S. (1991) “Estimation and Hypothesis Testing of Co-integration Vectors in Gaussian Vector Autoregressive Models." Econometrica, 59, 1551-1580. ----- and K. Juselius (1990) "Maximum Likelihood Estimation and Inference on Co-integration--with Applications to the Demand for Money," Oxford Bulletin of Economics and Statistics, 52, 169-210. Jung, S.W., Marshall, P.J., 1985.”Exports, Growth and Causality in Developing Countries.” Journal of Development Economics, 18:1-12. Kaldor, N., 1970. “The Case for Regional Policies,” Scottish Journal of Political Economy 17:337-348. Kohli, I., Singh, N., 1989. “Exports and Growth: Critical Minimum Effort and Diminishing Returns.”Journal of Development Economics, 30: 391-400. Krugman, P.R., 1984. “Import Protection as Export Promotion: International Competition in the Presence of Oligopoly and Economies of Scale,” in Kierzkowski (ed.): Monopolistic Competition in International Trade, Oxford University Press: Oxford, pp. 180-93. Liu, X, Song, H. and Romilly H. (1997). “An Empirical Investigation of the Causal Relationship between Openness and Economic Growth in China.” Applied Economics, 29, 1679-86. Maizels, A., 1963. Industrial Growth and World Trade. Cambridge, Cambridge University Press. Michaely, M., 1977. “Export and Economic Growth: An Empirical Investigation.” Journal of Development Economics, 4:49-53. Ndulu, B. and Njuguna Ndung’u (1998). "Trade and Growth in Sub-Saharan Africa." Ch.9 in (eds.) Zubair Iqbal and Mohsin Khan Trade Reform and Regional Integration in Africa IMF Institute, International Monetary Fund, Washington D.C. Onafowora, O. A. and O. Owoye (1998). “Can Trade Liberalization Stimulate Economic Growth in Africa?" World Development, Vol.26, No.3, March, pp.497-506. 66

International Business & Economics Research Journal – October 2010 36. 37. 38.

39.

40. 41. 42. 43. 44.

Volume 9, Number 10

Pack, H., 1992. “Learning and Productivity Changes in Developing Countries,” in G.K. Helleiner (ed.): Trade Policy, Industrialization and Development: New Perspectives, Oxford, Clarendon Press. Ram, Rati (1990). “Imports and Economic Growth: A Cross-Country Study,” in Economia Internazionale. 43, 45-66. Rodriguez, F. and Rodrik, D. (2001). “ Trade Policy and Economic Growth: A Skeptic’s Guide to CrossNational Evidence. In Bernanke, B. and Rogoff K. (Eds.), NBER Macroeconomics Annual. Cambridge, National Bureau of Economic Research, 261-225 Rodrik, D., 1994. “King Kong Meets Godzilla: The World Bank and the East Asian Miracle,” in A.Fishlow et al. (ed.): Miracle or Design? Lessons from East Asian Experience, Overseas Development Council, Washington D.C. Rodrik, D. (1998). "Trade Policy and Economic Performance in Sub-Saharan Africa." NBER Working Paper. No.6562, May, 1998. Romer, P.M., 1989. “Capital Accumulation in the Theory of Long Run Growth,” in Barro, Robert (Ed) Modern Business Cycle Theory (Cambridge, MM: Harvard University Press). Romer, P.M., 1993. “Idea Gaps and Object Gaps in Economic Development.” Journal of Monetary Economics, 32 (3): 543-73. Sachs, J.D. and Warner, A., 1997. “Sources of Growth in African Economies.” Journal of African Economies, Vol. 6, No. 3, pp. 335-376. Young, A., 1991. “Learning by Doing and the Dynamic Effects of International Trade.” Quarterly Journal of Economics, 106:369-405.

67

International Business & Economics Research Journal – October 2010 NOTES

68

Volume 9, Number 10