An investigation of the effects of exchange rate volatility on exports in East Asia

ABSTRACT

This study investigates the effects of the exchange rate volatility on the export flows of Indonesia, Malaysia, Republic of Korea, Singapore, Thailand, and the Philippines during 1974–2011. Towards this goal a trade weighted real effective (rather than the bilateral) exchange rate and three different measures of volatility, i.e. obtained from an ARCH model, a GARCH model and a moving-average standard deviation measure are used in this study. Specifically, the export flows between six Asian countries and the rest of the world are investigated rather than focusing on trade with only one country. Our findings reveal that the exchange rate volatility has a significant impact on export flows in the short run as well as in the long run for all the countries in the sample. The impact in the long run is predominantly negative with the exception of Singapore, but in the short run the impact varies across countries. Moreover, our results are robust to the alternative measures of volatility used and most of the findings in the long run and short run are also robust to the crisis period.

KEYWORDS:

• Foreign exchange rate volatility
• ARCH and GARCH volatility measures
• exports
• East Asian economies

JEL CLASSIFICATIONS:

• F31
• F10

Acknowledgements

The authors are grateful to two anonymous referees whose comments have greatly contributed towards improving this paper.

Disclosure statement

The views expressed in this article are the personal views of the authors and are not necessarily reflective of views of the City of New York or its Office of Management and Budget.

Notes

¹ The lag length is chosen by using the Akaike information criterion. The Schwarz information criterion was also used without significant change in the results.

² Specifically, these countries are Australia, Austria, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, Netherlands, New Zealand, Norway, Spain, Sweden, Switzerland, United Kingdom and United States.

³ For Indonesia, four observations from 1999:Q4 to 2000:Q3 are missing and are interpolated using a Catmull and Rom (1974) Spline method.

⁴ The seasonally adjusted industrial production index for Austria, Denmark and Switzerland is not available. For these countries, the X-12 ARIMA method is used to seasonally adjust the indices.

⁵ Given a considerable missing data in the export price of the Philippines, we use its main commodity export prices provided by the IMF (Coconut oil and Sugar) to estimate the export prices base year 2000.

⁶ The details of the trade policies and exporting goods of these countries can be found in Trade Policy Review-Indonesia (2013), Trade Policy Review-Malaysia (2014), Trade Policy Review-Republic of Korea (2012), Trade Policy Review-Singapore (2012), Trade Policy Review-Thailand (2011), and Trade Policy Review-The Philippines (2012).

⁷ Following Pesaran and Shin (1999), we use the Akaike criterion measure $LL-s$, where $LL$ is the maximized log-likelihood value and $s$ is the number of freely estimated parameters.

⁸ The intuition is straightforward; if there are two specifications that present autocorrelation, the best approach is to choose the one with less number of parameters.

⁹ There are few specific exceptions: For the case of Republic of Korea, only one volatility measure is significant and has negative effect. For Thailand and Singapore, only one of the volatility measures is not significant. Finally, the relative price in Singapore is significant for only one volatility measure.

¹⁰ For instance, specific aspects of the economy can be driving this finding.

¹¹ Note that in the short run, this is the first difference of the volatility measure.

¹² The moving-average measure seems to contradict the findings of other two volatility measures, but all measures are significant. Therefore, the discussion should focus on which measure is more adequate.

¹³ To save space, graphs of CUSUM tests and their respective confidence intervals are not included but are available from authors.

¹⁴ We also add a dummy variable for the Financial crisis in 2008, which is insignificant at the conventional levels for all countries. These results are available from the authors.

¹⁵ The F-statistic of this test does not follow a regular F-distribution. Significance is determined by using the approximation of the asymptotic p-values proposed by Hansen (1997).

¹⁶ The test statistics to test for the autocorrelation are not reported. However, we observe the same pattern in residuals as in the full sample, i.e. autocorrelation of order one is not significant and autocorrelation of order twelve is significant in the same countries.