China’s Exchange-Rate Regime Reform and Trade Between China and the Eurozone

ABSTRACT

This article investigates the effects of China’s exchange-rate regime reform on trade between China and the eurozone. Both the exchange rate between the euro (EUR) and the renminbi (RMB) and exchange-rate volatility are included in the autoregressive distributed lag (ARDL) model, and our empirical work also considers the third-country effect. Our findings show that, during the reform period, China’s exports to the eurozone are affected only by the EUR–RMB exchange rate per se and not by its volatility. However, neither the exchange rate nor its volatility significantly influences the eurozone’s exports to China during the reform period. Such asymmetry might be attributed to the discrepancy between Chinese exporters and their eurozone counterparts in the knowledge and ability to manage exchange-rate risk.

KEY WORDS:

• ARDL model
• China’s exchange-rate regime reform
• China–eurozone trade
• exchange rate

JEL CLASSIFICATION:

• F14
• F31

Funding

This research is financially supported by the NSFC Research Fund (71601091), to which we are really grateful.

Notes

1. In this article, “China” or “Chinese” refers to mainland China, excluding Hong Kong, Taiwan, and Macau.

2. The contribution of household consumption to the gross domestic product (GDP) is much smaller in China than in the United States. In 2012, the share in China was only 41.3 percent, while in the United States, it was 68.2 percent. These figures are calculated by the authors based on the household consumption and GDP data of China (US) from the Bloomberg database.

3. Data source: General Administration of Customs of the People’s Republic of China (http://www.customs.gov.cn/).

4. Review articles on the relationship between international trade and exchange-rate volatility include Bahmani-Oskooee and Hegerty (2007), Cote (1994), and McKenzie (1999).

5. According to Chit and Judge (2011, page 107), “the less financially developed an economy, the more its exports are adversely affected by exchange rate volatility.”

6. Among the many studies that examine the international trade between China and other countries, Chou (2000) analyzes China’s total exports and imports in four sectors (foodstuffs, industrial materials, manufactured goods, and mineral fuels). Bahmani-Oskooee and Wang (2007) examine data on 88 specific commodities traded between China and the United States. Wesseh and Niu (2012) study the relationship between South Africa’s exports to China and volatility in the exchange rate between the RMB and the South African rand. Nishimura and Hirayama (2013) examine the relationship between Chinese exports to Japan and volatility in the exchange rate between the RMB and the yen. They find that Japan’s exports to China are not affected by exchange-rate volatility, but China’s exports to Japan are negatively affected during the reform period.

7. An alternative approach adopts EUR–USD or USD–RMB exchange rates to conduct this empirical study, because most trade between China and the eurozone is billed in US dollars. However, the relevant price computed by a Chinese importer is the euro price multiplied by the EUR–RMB exchange rate, which obviously implies that one should use the EUR–RMB rate as an important factor in modeling trade between China and the eurozone.

8. These model specifications are summarized in detail in McKenzie (1999, 80–83, Table 2); Bahmani-Oskooee and Hegerty (2007, pp. 226–230, Table 2, pp. 238–241, Table 3, pp. 247–249, Table 4). If panel data are used to analyze multiple countries, the researcher tends to adopt a gravity model that considers two countries’ proximity either to each other or to common borders.

9. Bahmani-Oskooee and Xu (2012) show the significant third-country effect on bilateral trade between China and the United States.

10. We thank an anonymous referee for suggesting consideration of a third-country effect on bilateral trade.

11. The People’s Bank of China set the USD–RMB exchange rate to fluctuate within a certain band as its policy target, and therefore the exchange rate amplitude of the RMB with respect to the USD is smaller than that of the RMB with other currencies. Actually, from January 2002 to December 2013, the daily return volatility in the USD–RMB exchange rate is only 0.087, but for the EUR–RMB exchange rate, it is 0.622.

12. In September, October, and November 2008, respectively, holdings of US Treasury bonds by the Chinese government reportedly increased by $44.5 billion, $65.9 billion, and $29.1 billion. Data source: Treasury International Capital System of US Department of the Treasury. Website: https://www.treasury.gov/resource-center/data-chart-center/tic/Pages/ticpress.aspx.

13. Paul Krugman considers this crisis “the beginning of a second Great Depression” (“Fighting Off Depression,” New York Times, January 4, 2009).

14. Chinese exports to these countries consist mainly of manufactured goods but with different trade structures. According to UN Comtrade data in 2013, the major Chinese export to Germany and the Netherlands are machinery and transport equipment, respectively, comprising 45.9 and 63.9 percent of the total to the two countries. Chinese exports to France are mainly miscellaneous manufactured goods, comprising 41.2 percent of total Chinese exports to France. However, exports from these eurozone countries to China vary considerably. Exports from France and Germany to China are mainly machinery and transport equipment, making up 68.1 and 51.7 percent of their exports to China, respectively. However, the major Dutch exports to China are primary goods, comprising as much as 46.6 percent. France also exports primary goods to China at a share of 16.1 percent of French exports to China.

15. In the absence of a price level for each country, as a second-best deflator, we use the aggregate export and import price index to deflate the nominal exports and imports of each country.

16. Only year-on-year rates of change in the industrial production index used to be released, but starting in 2011, the NBS now publishes monthly rates. Using those rates, we have recomputed the level data. For unknown reasons, January data are missing beginning in 2006. January values are estimated by averaging the December and February values.

17. The Chinese CPI data are released monthly, so the level data of CPI are derived using these rates with 2010 = 100.

18. With respect to detailed measures on exchange-rate volatility, see Bahmani-Oskooee and Hegerty (2007) and McKenzie (1999).

19. McKenzie (1999) compares various types of exchange-rate volatility measures and noted that “the use of ARCH-based measures of volatility in an OLS framework creates a generated regressor problem in that whilst the model generates consistent parameter estimates, they are not necessarily efficient. However, the use of other techniques may lead to a measurement error problem in which the estimates are inconsistent” (p. 98).

20. Both Engle–Granger’s and Johansen’s co-integration tests require that all the series in a system be integrated with the same nonzero order. However, one advantage of the ARDL method is that it can be used to test the existence of co-integration when some of the series are stationary and others are integrated with order 1.

21. To find the maximum lag orders of m1, m2, m3, m4, and m5, we tested all possible combinations of these variables, allowing them to vary within certain ranges: m1 = 1–2, m2 = 0–2, m3 = 0–2, m4 = 0–2, and m5 = 0–2. The optimized lag orders were chosen to minimize the Akaike’s information criterion.

22. In economics, the J curve denotes the trend of a country’s trade balance following its currency depreciation under a certain set of assumptions. After the currency depreciation, in the short run, the volume of imports and exports may not change much, due in part to existing trade contracts. But in the long run, currency depreciation can improve the trade account balance.

23. Nishimura (2013) examines the influence of the RMB exchange rate on China’s total exports during the reform period and finds that the elasticity is as high as 2.45 percent, as in in this article. However, the period of China’s exchange-rate regime reform is not long enough to yield reliable empirical results. The question is still open and needs further investigation.

24. A significantly negative coefficient on the error correction term means that, in the long term, the deviation from the equilibrium will eventually converge to zero. A large coefficient indicates a rapid convergence speed. For example, the corresponding coefficient in Panel A of Table 6 is −0.3123, implying that 31.23 percent of the departure from the long-run equilibrium is corrected. Accordingly, the length of time that it takes to achieve the equilibrium is approximately 3.2 months (the reciprocal of 0.3123). For the eurozone countries’ exports to China, it is about 1.81 months. Our analysis is robust with respect to different volatility measures.

25. The ARCH-type models are explained in detail by Bollerslev, Engle, and Nelson (1994) and Xekalaki and Degiannakis (2010).

26. Because the standardized residuals of Equation (3) do not follow a normal distribution, we used the quasi-maximum likelihood (QML) method to estimate the parameters of the AR–EGARCH model. For details on the QML method, see Xekalaki and Degiannakis (2010, sec. 2.2).

27. We also conduct the Lagrange multiplier (LM) test on the simple AR(k) model (1), that is, without considering the GARCH components. The result shows that LM test statistics are significant at the 1-percent level ($L{M}_{10}^{\mathrm{A}\mathrm{R}\mathrm{C}\mathrm{H}}=244.78$). This result suggests that the return series of the JPY-RMB exchange rate has a high-order ARCH effect.

28. Whether one should use volatility in the nominal exchange rate or the real rate is discussed in detail in McKenzie (1999, sec. 3.1). He argues that “In general, it would appear that whilst the distinction between real and nominal exchange rate volatility has generated a lot of debate in the literature, the empirical results suggest that this distinction does not impact significantly on the results achieved” (p. 85).