Trade, finance, specialization and synchronization in the Asia-Pacific

Abstract

In this paper, we examine the relationship between trade, finance, specialization and output synchronization of 12 Asia-Pacific economies by studying the direct and indirect effects of increasing trade and financial integration on output synchronization. Using cross-sectional data for the periods of 1984–1996 and 1999–2007, we estimate a system of equations accounting for both endogeneity and simultaneity. Our main findings suggest that: (1) trade and financial integration has direct positive effects, while specialization has direct negative effects on output correlations. An increase in the coefficient of trade intensity and the significance of financial integration is observed in the post-Crisis period. (2) Most estimated coefficients have signs consistent with the existing literature and the results remain robust under different measures of output correlation, but a notable difference is that trade and finance have sizable positive effects on specialization; however, specialization is not a driving force of trade. (3) Countries with more variance in exchange rates have less synchronized cycles.

Keywords:

• business cycle synchronization
• regional trade
• financial cooperation

JEL codes:

• C01
• C32
• F10

Acknowledgements

The authors would like to thank Professor Noel Gaston and the anonymous referee for very helpful comments that greatly improved the paper. The authors also thank participants at Bond-NTU first annual workshop for useful comments. The authors are grateful for Mr Liyu Dou's assistance in the collection of data.

Notes

1. Inclusion of the Crisis year tends to distort the estimation for both periods. The time series graph with 9-year moving output correlations excluding AFC displays considerable increase in the post-Crisis period, indicating a clear structure break at the AFC.

2. We fit each country's log (GDP) series over a quadratic trend and compute the pairwise correlation of residuals as proxies for business cycle synchronization.

3. The Hodrick–Prescott filter decomposes a series into a trend series and its corresponding cyclical component. The cyclical components are filtered by finding a trend that satisfies: , where in our case, y_t is the log (GDP). Since our data are annual series, we set λ=100. The pairwise correlations of cyclical components are used as the proxy for business cycle synchronization.

4. The density plots for first-difference correlations and quadratic correlations show similar patterns and the plots are available upon request from the authors.

5. The categorization of seven sectors is as follows: (1) Agriculture, hunting, forestry, fishing (ISIC A and B); (2) Mining, Utilities (ISIC C and E); (3) Manufacturing (ISIC D); (4) Construction (ISIC D); (5) Wholesale, retail trade, restaurants and hotels (ISIC G, H); (6) Transport, storage and communication (ISIC I); and (7) Other Activities (ISIC J-P).

6. Some of the gravity variables are taken from Andrew Rose's website: http://faculty.haas.berkeley.edu/arose/

7. China's data on government budget balance expressed in RMB are available on the CEIC database. We normalized it by GDP at current price in RMB collected from WDI to obtain the percentage of government balance in GDP for China.

8. Data for all countries except Hong Kong and Thailand are available since 1984. For Hong Kong and Thailand, we use data for shorter periods as 1988–2007 and 1989–2007, respectively.

9. The ranking of currencies-based currency distribution of global foreign exchange market turnover is as following (descending): for the period 1984–1996: US dollar, Japanese yen, Australian dollar, Singapore dollar, Hong Kong dollar, Korean Won, New Zealand dollar, Indonesian Rupiah, Thai Baht, Chinese Renminbi, Philippine Peso and Malaysian Ringgit; for the period 1999–2007: US dollar, Japanese Yen, Australian dollar, Hong Kong dollar, New Zealand dollar, Korean Won, Singapore dollar, Chinese Renminbi, Malaysian Ringgit, Indonesian Rupiah, Thai Baht and Philippine Peso.