Non‐linear effect of exchange rate volatility on exports: the role of financial sector development in emerging East Asian economies

Abstract

This paper empirically examines the role of financial sector development in influencing the impact of exchange rate volatility on the exports of five emerging East Asian countries – China, Indonesia, Malaysia, the Philippines and Thailand – using a GMM‐IV estimation method. The results indicate that the effect of exchange rate volatility on exports is conditional on the level of financial sector development. The less financially developed an economy, the more its exports are adversely affected by exchange rate volatility. In addition, a stable exchange rate seems to be a necessary condition to achieve export promotion via a currency depreciation in these economies.

Keywords:

• East Asia
• exports
• exchange rate volatility
• financial sector development

JEL Classifications:

• F13
• F14
• F31
• O16

Acknowledgements

The authors would like to thank the editor and in particularly an anonymous referee for very useful comments and suggestions. The authors are grateful to the participants of 18th Chinese Economic Association Conference and the Campus for Finance 2009 Research Conference, Philip Arestis, John Grahl, David Kernohan and Marian Rizov for their helpful comments. Any errors, however, are entirely our own.

Notes

1. This point of view has been reflected in the creation of the European Monetary Union (EMU), as one of the stated purposes of the EMU is to reduce exchange rate uncertainty in order to promote intra‐EU trade and investment (EEC Commission 1990).

2. In a related but different stream of literature, Aghion et al. (2009) examine the role of financial sector development in the linkage between exchange rate volatility and productivity growth.

3. In a related stream of literature, there are a number of studies which look at the nonlinear exposure of exchange rate movements on exports (e.g., Koutmos and Martin 2003; Berman and Berthou 2006; Priestley and Ødegaard 2007).

4. The gravity model has been widely used in the majority of empirical studies that examine the relationship between exchange rate volatility and bilateral exports in the panel data context (e.g., Dell’Ariccia 1999; Rose 2000; Clark, Tamirisa, and Wei 2004).

5. See McKenzie (1999) for the survey of the different measures of exchange rate volatility employed in the empirical literature.

6. The temporal window for the measure is eight quarters in order to stress the importance of medium‐run uncertainty. The current volatility is calculated on the exchange rate movements during the previous eight quarters in order to reflect the backward‐looking nature of risk.

7. For China, the data for quarterly CPI is not readily available for the whole sample period and the missing data are constructed by using Otani‐Riechel method to transform the annual data obtained from WDI (World Development Indicators) and various Chinese Statistical Yearbooks into quarterly data.

8. The first principal component of a set of variables is a weighted average of the variables in which the weights are chosen to make the composite variable reflects the maximum possible proportion of the total variation in the set. For example, the financial sector development for China is estimated as:

and this linear combination reflect the 93.5% of the total variation.

9. The justification is that relative money supply and bilateral exchange rates are highly correlated, but monetary policies are less affected by trade considerations than exchange rate policies (Frankel and Wei 1993; Clark, Tamirisa, and Wei 2004).

10. For our model, the presence of groupwise heteroskedasticity in the panel OLS residuals is tested; the modified Wald statistics reject the null hypothesis of homoskedasticity.

11. Baum, Schaffer, and Stillman (2003) state that if error terms are heteroskedastic, the GMM estimator is more efficient than the simple IV estimator.

12. Baltagi (2008, 273) notes that for macro‐panels with large N (numbers of cross‐sectional observations) and larger T (length of time series), nonstationarity deserves more attention.

13. System GMM approach suggested by Arellano and Bover (1995) can estimate the impact of time‐invariant variables. However, according to Roodman (2006), this approach is more suitable for a panel with small T and large N as the number of instruments is increasing with time dimension T. Since the time dimension of the panel data set of our study is considerably longer (68 quarters), the System GMM estimation method is not suitable.

14. We conduct the Sargan‐Hansen test and a weak ID test suggested by Stock and Yogo (2005) to verify the validity of our instrument. The diagnostic test statistics suggest that our instrument is valid and we can also reject the null hypothesis that the instrument is weak.

15. The results are available upon request.

16. The authors owe thanks to an anonymous referee for pointing out the issue of a possible multiplicative relationship between exchange rate volatility and the relative prices.