The role of exports, FDI and imports in development: evidence from Sub-Saharan African countries

Abstract

The disappointing economic performance of Sub-Saharan African (SSA) economies in the late 1980s prompted reforms in foreign trade and Foreign Direct Investment (FDI) in the early 1990s. Using the Autoregressive Distributed Lag (ARDL) approach and Pedroni panel estimation procedures that allow for heterogeneity, this study found that exports and FDI have a significant impact on economic growth. Granger-type causality tests show the interrelatedness of exports, FDI, imports and income variables. The results also provide evidence of a two-stage causal chain of exports, imports and income. This article calls for more market-oriented policy reforms in SSA countries.

Acknowledgements

We would like to thank the editor of Applied Economics and two anonymous referees for their insightful comments, which greatly helped improve the article. Any remaining errors are the responsibility of the authors.

Notes

¹ Economic growth averaged 4.5% in 1965–1969, declining to 4.2% in 1970–1979, before facing serious decline and stagnation to an average of 2.5% in 1980–1989.

² These reforms included, but were not limited to, the removal of export restrictions, lowering of import tariffs and relaxation of quantitative restrictions on imports.

³ In addition to other economic adjustments, two-thirds of SSA countries initiated trade liberalization. However, average tariffs remained high, nontariff barriers to trade existed and most of the countries also had a variety of quantitative restrictions and exchange controls (Sachs and Warner, 1997).

⁴ FDI has increased from about US$2.2 billion in 1980s to around US$20 billion in 2004 (UNCTAD, 1999, 2003; Ahmed et al., 2007).

⁵ There are also empirical works indicating a stronger relation between FDI and economic efficiency (Ghirmay et al., 2001).

⁶ See Ikiara (2003) for a detail discussion.

⁷ Others claim that the growth-enhancing effect of FDI is not obvious; which may vary from country to country (Borensztein et al., 1998).

⁸ For further discussion on the sectorial distribution and transformation in countries such as Ghana, see Ahmed et al. (2007).

⁹ Bende-Nabende (2002) notes that this is the case, even in oil-exporting countries.

¹⁰ Detailed policy analyses regarding FDI and trade for these countries are available upon request.

¹¹ Later, we utilize Pedroni's (1999) panel estimation approach to check the robustness of our findings.

¹² VAR and ARDL models are now standard technique in econometric analyses and thus, we limit our discussion on this. For an excellent discussion, see Zachariadis (2006).

¹³ Appendix gives full definitions and sources of all the variables we have used.

¹⁴ We use the principal weighted average method to compute a single index for regression analysis.

¹⁵ Some studies used quarterly data; annual data are preferred for LDCs, due to unavailability and quality of such quarterly data.

¹⁶ To conserve space, we do not provide the unit root test results here, but they are available on request.

¹⁷ We have not included foreign income (y ^us) as we assume it to be exogenous. We limit our examination to this number of relations even though FDI-imp could be considered.

¹⁸ The lags used here were identified using the AIC.

¹⁹ To examine the consistency of our findings, we have converted the FDI, y and Expo variables into per capita and re-conducted unit root and causality tests. The results generally remain the same and are available upon request.

²⁰ Although not reported due to space constraint, a test of cointegration between variables indicated that there exists at least a unique cointegarting vector, and therefore the existence of a long-run relationship among the variables involved.

 ²¹In a different specification, we have included terms of trade and the results are available upon request. To generate enough degrees of freedom for estimation, this variable has not been included here.

²² Asafu-Adjaye and Chakraborty (1999) for a detailed analysis on this.

²³ Prior to the cointegration examination, we also perform a panel unit root to test the presence or absence of a unit root. With the exception of a few cases where we observe FDI to show stationarity and given various weaknesses of panel unit root technique (Karlsson and Lothgren, 2000), the null hypothesis of nonstationarity cannot be rejected.