What drives the dependence between the Chinese and global stock markets?
The economic and trade exchanges between China and major economies, as well as their financial markets, have become increasingly close since China joined the WTO. The nature and origins of dependence between stock markets have attracted considerable attention from academics, investment professionals, and government regulators. To date, the presence of stock market dependence and its concrete form has been the focus of numerous studies (Wongswan, 2006; Quinn and Voth, 2008), but few studies provide the theoretical and empirical underpinning of possible driving forces behind such dependence. In this paper, we apply time-varying copulas to measure the dependence between the Chinese and eleven international stock markets while investigating its determinants through panel regression analysis.
This study contributes to the literature in the following aspects. First, we use a unique sample focusing on the stock market dependence between China and major economies, which extends the current literature in developed countries. For example, Quin and Voth (2008) examine a dataset of 16 developed countries, which do not include China. However, with the rapid development of the Chinese economy, the Chinese stock market has been the second largest one around the world, whose dependence and impact on other international markets cannot be neglected.
Second, this study diverges from previous studies in employing time-varying copulas proposed by Patton (2006) to measure market dependence, which provides a better fit of nonlinear and time-varying dependence than traditional methods such as the linear correlation used by Tavares (2009) and the dynamic conditional correlation of Engle (2002). Further, by modeling flexibly the appropriate marginal distributions and copulas, we avoid choosing incorrect model specifications. The dependence magnitude between the Chinese stock market and major international markets is found to vary with regions obviously.
Finally, and most importantly, our panel regression results indicate that economic policy uncertainty differentials are negatively related to stock market dependence between China and major economies, which extends the work of Baker et al. (2016) and others by linking economic policy uncertainty with market dependence. Consistent with Johnson and Soenen (2002), our results suggest a negative effect of interest rate differentials on dependence. Moreover, the positive effects of the global financial crisis and trade interdependence on stock market dependence are in line with Pretorius (2002) and Paramati et al. (2016) but in contrast with Vithessonthi and Kumarasinghe (2016) that report insignificant impacts.
The remainder of this paper is organized as follows. Section 2 describes the data, variables, and methods. Section 3 reports the empirical results. Section 4 concludes the study.
2. Data, Variables, and Methods
The market returns are computed as the log-returns multiplied by 100 for our sample: China’s SSEC, Hong Kong’s HSI, the United States’s SP500, Japan's N225, Korea’s KOSPI, Australia’s SP200, the United Kingdom's FTSE100, France's CAC40, Germany’s DAX 30, Brazil's BVSP, Russia's RTS and India's SSEX30 after China joined the WTO (December 11, 2001). Table A.1 in the Online Appendix reports the summary statistics of the market returns. To obtain pairwise dependence coefficients of market returns for eleven country pairs, we implement a two-stage method to perform time-varying copulas (see Appendix B in the Online Appendix for details).
We take monthly averages of daily pairwise dependence coefficients as the dependent variable (denoted as DEP) to investigate underlying determinants using a panel regression framework. The description of explanatory variables is presented in Table 1.
|Panel A: Economic policy uncertainty differentials|
|Economic policy uncertainty differentials||EPU||The absolute difference of log of economic policy uncertainty indices between country pairs|
|Panel B: Financial crisis|
|Global financial crisis||GFC||A dummy variable that takes the value of one during the global financial crisis periods (March 2007-May 2008) and zero otherwise|
|Panel C: Macroeconomic conditions|
|Interest rate differentials||IR||The absolute difference of one-year treasury bill rates|
|Trade interdependence||TRADE||The sum of the value of imports and exports from China to other economies as a proportion of China’s GDP|
|Economic growth differentials||EG||The absolute difference in GDP growth rates|
|Inflation rate differentials||INF||The absolute difference in the inflation rates|
|Exchange rate growth||ERG||Percentage change in the bilateral exchange rates of foreign currencies in terms of the Chinese Yuan|
|Monetary policy growth differentials||MPG||The absolute difference in M2 to GDP ratios|
|Panel D: Market performance|
|Market development growth differentials||MDG||The absolute difference in the total value of stock traded to GDP|
|Market volatility differentials||VOL||The absolute difference of standard deviations of market returns|
The descriptive statistics are presented in Table A.2 in the Online Appendix. Following Carrieri et al. (2007), a series of tests is conducted to determine the appropriate specification for the panel models (see Table A.3 in the Online Appendix). We find that the two-way fixed effects model is the appropriate model for our empirical analysis, which is defined as:
DEPit= β0 +β1EPUit + β2GFCit +β3IRit +β4TRADEit +β5EGit +β6INFit +β1ERGit +β8MPGit β9MDGit β10VOLit +µi +γt +εit ,(1)
where i represents cross-sections, t represents time periods, µi denotes the cross-section, γt captures the period fixed effects, and εit is a random disturbance effect.
3. Empirical Results
Since the pairwise dependence coefficients are confirmed to be time-varying, three types of time-varying copulas are employed to characterize them. The optimal copulas and summary statistics for dependence coefficients are displayed in Table 2.
|Country pair||Optimal copula||Mean||Max.||Min.||Std. dev.|
|SSEC-HSI||time-varying Student’s t||0.4194||0.6858||-0.2584||0.1261|
|SSEC-SP500||time-varying rotated Gumbel||1.0549 [0.0956]||1.2787 [0.1282]||1.0001 [0.0437]||0.0290 [0.0090]|
|SSEC-N225||time-varying rotated Gumbel||1.1543 [0.2315]||1.4659 [0.4546]||1.0108 [0.0540]||0.0434 [0.0476]|
|SSEC-KOSPI||time-varying Student’s t||0.2644||0.5689||-0.1058||0.0987|
|SSEC-SP200||time-varying Student’s t||0.2166||0.4188||-0.1047||0.0775|
|SSEC-FTSE100||time-varying rotated Gumbel||1.0965 [0.1424]||1.2223 [0.2481]||1.0323 [0.0032]||0.0265 [0.0535]|
|SSEC-CAC40||time-varying rotated Gumbel||1.0845 [0.1286]||1.1644 [0.2132]||1.0477 [0.0171]||0.0161 [0.0384]|
|SSEC-DAX30||time-varying rotated Gumbel||1.0814 [0.1173]||1.1855 [0.1573]||1.0383 [0.0869]||0.0201 [0.0063]|
|SSEC-BVSP||time-varying Student’s t||0.1418||0.1801||0.0910||0.0084|
|SSEC-RTS||time-varying Student’s t||0.1709||0.3005||-0.0076||0.0574|
|SSEC-SSEX30||time-varying normal||0.1968 [0.1929]||0.3635 [0.2784]||-0.0229 [0.1174]||0.0611 [0.0265]|
As we observe, time-varying Student’s t copula is the optimal model for describing the dependence between China’s SSEC and most indices, including Hong Kong’s HSI, Korea’s KOSPI, Australia’s SP200, Brazil's BVSP, and Russia's RTS, while time-varying rotated Gumbel copula for other pairs except SSEC-SSEX30. Our results indicate that the dependence magnitude between Chinese and major international stock markets varies with regions obviously, from large to small: Asia-Pacific (Hong Kong, Korea, Japan, Australia), BRIC (India, Russia, Brazil), and Euramerican (the United Kingdom, France, Germany, the United States).
Table 3 presents the panel regression results, which demonstrate that the larger absolute difference in economic policy uncertainty between China and other economies would decrease their stock market dependence. A possible channel works in the following way. One country with heightened economic policy uncertainty experiences greater stock market volatility and lower investment rates, thereby triggering a cycle of falling asset prices (Baker et al., 2016). Our results also point towards a negative effect of interest rates differentials on market dependence, providing evidence consistent with Johnson and Soenen (2002). They argue that a greater differential in interest rates would reduce comovements between the Japanese and twelve Asian equity markets.
|Variable||Mod. 1||Mod. 2||Mod. 3||Mod. 4||Mod. 5||Mod. 6||Mod. 7||Mod. 8|
|EPU||-0.014 (0.005)||-0.012 (0.004)||-0.013 (0.005)||-0.013 (0.004)||-0.013 (0.004)|
|GFC||0.030 (0.009)||0.026 (0.007)||0.028 (0.008)||0.026 (0.007)||0.026 (0.008)|
|IR||-0.004 (0.001)||-0.004 (0.001)||-0.005 (0.001)||-0.004 (0.001)|
|TRADE||0.006 (0.002)||0.004** (0.002)||0.004 (0.002)||0.003 (0.002)|
|EG||0.001 (0.001)||0.001 (0.001)|
|INF||0.005 (0.005)||0.005 (0.005)|
|ERG||-0.001 (0.002)||-0.001 (0.001)|
|MPG||0.001 (0.001)||0.001 (0.001)|
Using a panel sample of China and eleven developed and emerging economies over the period of January 2002 to December 2018, this paper employs three types of time-varying copulas and the two-way fixed effects model to investigate the evolvement of pairwise stock market dependence and the potential factors that can determine it. We demonstrate that the dependence magnitude between the Chinese stock market and major international markets varies with regions obviously, from large to small in order: Asia-Pacific, BRIC, and Euramerican markets. Furthermore, panel regression analysis demonstrates that economic policy uncertainty differentials, global financial crisis, interest rate differentials, and trade interdependence are significant determinants. Our findings are of great significance to international portfolio construction and risk management.
Supplementary Materials: Online Appendix is available from the authors.
Author Contributions: Conceptualization, Huaigang Long, and Lingling Qian; methodology, Lingling Qian; software, Lingling Qian; validation, Lingling Qian, Huaigang Long; formal analysis, Lingling Qian; investigation, Lingling Qian; resources, Huaigang Long; data curation, Lingling Qian.; writing—original draft preparation, Lingling Qian; supervision, Yuexiang Jiang; project administration, Yuexiang Jiang; funding acquisition, Yuexiang Jiang. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the National Natural Science Foundation of China, Grant number , the Zhejiang Provincial Natural Science Foundation, Grant [LY21G030014].
Data Availability Statement: The processed data from this study is available upon request.
Conflicts of Interest: The authors declare no conflict of interest.