Does foreign direct investment increase exports’ productivity? Evidence from developing and emerging countries

Abstract

Raising the productivity content of exports is an important issue for developing and emerging countries. What role do foreign firms play in this process? This question has not been adequately studied. We contribute to the literature by generalizing the role of foreign direct investment (FDI) in the host country’s export productivity level. Using panel data, we present new empirical evidence suggesting that FDI boosts the overall productivity level of the developing and emerging countries’ exports.

Keywords:

• export productivity
• foreign direct investment

JEL Classifications:

• F1
• F2
• O1

Acknowledgements

We are grateful to the editor, Malcolm Sawyer, and two anonymous referees for many valuable comments and suggestions, which have significantly improved the paper. We are also grateful to Ricardo Hausmann and Sebastian Bustos for making available to us the data on export productivity. We thank Alain Desdoigts, Subrata Ghatak, Fabian Gouret, Usha Nair-Reichert, Dominique Redor and Jean Marc-Siroen for their helpful comments and suggestions on earlier versions of this paper.

Notes

1. ‘The idea is that producing high-productivity goods has greater growth benefits than producing other goods – computer chips are better than potato chips’ (Amiti and Freund 2007, 39).

2. A new advanced literature uses this measure of export productivity (Xu and Lu 2009; Minondo 2010; Xu 2010; Jarreau and Poncet 2010, 2012; Santos-Paulino 2010, 2011; Harding and Javorcik 2012).

3. See Santos-Paulino (2011, 87).

4. See also Harding and Javorcik (2013).

5. Harding and Javorcik (2012) use Wang and Wei’s (2010a) index to measure the dissimilarity between the product structure of a country-sector’s exports and that of the same sector in high-income economies.

6. A related literature focuses on the technological content of FDI and considers technology and managerial talent as the key ingredients of FDI (Root 1994; Saggi 2002; Singh and Marjit 2003; Keller 2010).

7. According to Dunning and Narula (2004, 18), the ratio of inward FDI stock to GDP is a proxy for the role of multinational firms’ activity in the host economy. Borenstein, Gregorio and Lee (1998) use this ratio to capture the technology diffusion from FDI.

8. Multicollinearity is examined using Variance inflation factors (VIF). There is evidence of the multicollinearity problem if the mean VIF is greater than 6 and the largest individual VIF is greater than 10 (Gujariti 2004; Bowerman, O’Connell, and Koehler 2005). The VIFs look fine here. The results confirm that our variables do not suffer from any multicollinearity problem.

9. ‘The original story of how comparative advantage is determined in part by a process of cost discovery by initial entrants in a new industry’. High productivity ‘discoveries’ naturally attract more emulation, and the productivity of an economy’s tradable sector tends to converge towards the productivity level of the most profitable (most productive) activities discovered to date (Rodrik 2006, 11).

10. Rodrik (2006, 11).

11. This index measures ten specific factors, including freedom in the business environment, trade, fiscal policy, government, monetary policy, investment, property rights, and corruption. The lower scores on a factor the higher the level of government interference in the economy and the lower the economic freedom.

12. Driscoll and Kraay (1998) show that erroneously ignoring cross-sectional dependence in the estimation of linear panel models can lead to severely biased statistical inference. The tests of cross-sectional dependence are implemented following De Hoyos and Sarafidis (2006). The method by Driscoll and Kraay (1998) is implemented following Hoechle (2007).

13. We report the average absolute value of the off-diagonal elements of the cross-sectional correlation matrix of residuals. This option is useful to identify cases of cross-sectional dependence where the sign of the correlations is alternating, with the likely result of making Pesaran’s test unreliable. When Pesaran’s test fails to reject the null hypothesis of cross-sectional independence of the errors, whereas Frees’ test does not, inferences should be based on the latter (De Hoyos and Sarafidis 2006, 494).

14. We re-run regressions (not reported for the sake of space saving) with RHS variables lagged one year. Our findings are robust to lagging the key RHS variable-FDI-by one year.

15. To take into account heterogeneity across the sector-country combinations, Harding and Javorcik (2012) include country-sector fixed effects in their specification. They seek to capture time invariant characteristics differentiating sectors chosen for targeting from other sectors and specific to a particular country-sector combination that may affect EXPY. In other words, their analysis focuses on within-country-sector variation in EXPY.

16. We thank an anonymous referee for emphasizing this point.