More on the volatility-trading volume relationship in emerging markets: The Chinese stock market

Abstract

This paper empirically investigates the characteristics in terms of volatility and trading volume relationships of the Chinese stock markets, and specifically of the stocks comprising the SSE180 index. Our results show that, contrary to previous evidence, both volatility and trading volume appear to be multi-fractal and highly intermittent, suggesting a common long-run behaviour in addition to the common short-term behaviour underlined by former studies. Moreover, the trading volume seems to have no explanatory power for volatility persistence when introduced in the conditional variance equation. Finally, the sign of the trading volume coefficients is mainly negative, hence showing a negative correlation between the two variables.

Keywords:

• volatility persistence
• long-memory
• trading volume

Notes

Baillie et al. 6 and Bollerslev and Mikkelsen 8 manage to develop the GARCH framework to allow for persistence in the conditional variance by introducing the Fractionally Integrated GARCH process. Despite the success of this model in capturing the long memory characteristics of the volatility, the FIGARCH framework still suffers of some stationarity and ergodicity problems. Moreover, it is difficult to discriminate between different specifications of the FIGARCH (see for example Caporin, 13 for further details).

A widely known and commonly used semi-parametric estimator for d is the so-called GPH logperiodogram regression estimator introduced by Geweke and Porter-Hudak 21. However, as the absolute returns and the trading volume violate the Gaussian assumption (as most financial series) they thus invalidate the asymptotic theory for the GPH estimator which explains our choice.

In our estimations we follow Andersen and Bollerslev 3 and use m=T ^1/2 and the scalar τ=0.25. However, we checked for robustness and obtained comparable results for other values of these parameters.

For more details on the existence of positive autocorrelations in the trading variables, please refer to Admati and Pfeiderer 1 among others.

As reported by Los and Yu 38, individual investors comprise a large part of market participants in China, currently about 60 million strong.

We thank an anonymous referee for suggesting this point.