Stock market return and time-varying volatility in an emerging market: evidence from sub-saharan Africa (ssa)

Résumé du document

This paper offers a comprehensive view of four time properties that emerge from the empirical time series literature on asset returns. It examines: (1) the predictability of returns from past observations; (2) the auto-regressive behavior of conditional volatility; (3) the asymmetric response of conditional volatility to innovations; (4) and the conditional variance risk premium. Three emerging markets previously under-researched in this respect are considered: South Africa (ALSI, IND, GOLD indexes), Kenya (NSE index) and Nigeria (LSE index).
The paper employs exponential GARCH (EGARCH) framework for analysis. The results indicate that asymmetric volatility found in the US and other developed markets does not appear to be a universal phenomenon. Significant asymmetric volatility is found in both South Africa and Nigerian stock markets. However, in one market, NSE (Kenya), the asymmetric volatility coefficient is significant but positive, suggesting that positive shocks increase volatility more than negative shocks of an equal magnitude. LSE (Nigeria), GOLD (the post-announcement sample) and IND (the pre-announcement sample) return series exhibit a significant and positive time-varying risk premium. NSE (Kenya) and ALSI (South Africa) return series report negative but insignificant risk-premium parameters.

Extraits

[...] Under the CAPM mean-variance hypothesis, large variances (or standard deviations) are expected to be associated with large returns, thus, is expected to be greater than zero. And, since Rt is the total market return, the term Ø0 is analogous to the risk-free rate in the CAPM. The Ø1Rt-1 and Ø2Rt-2 components are included in the mean equation to account for the auto-correlation potentially induced by non- synchronous trading in the assets that make up a market index. This problem can be particularly severe in emerging markets given their low level of liquidity (DeSantis and Imrohoroglu, 1997). The parameterization we use follows Lo and MacKinlay (1988). [...]

[...] Likewise, the issue of asymmetric volatility is very important given the degree of interdependence of the various national stock markets. Evidence on first and second moment interdependence across national stock markets is provided by Hamao, Masulis and Ngo (1990), for the U. S. the Japanese and the U. K. markets. Thus, if all major stock markets exhibit characteristics of asymmetric volatility, similar to the U. S. market, one should expect market declines worldwide to be associated with high levels of volatility. A significant amount of research on the volatility of the U. [...]

[...] However, we follow Nelson (1991) to allow for the asymmetric response of volatility to innovations and Engle et al. (1987) for Mean' effects. A technique for incorporating asymmetries in the modeling of volatility is the EGARCH model developed by Nelson (1991). Nelson argues that returns may exhibit asymmetric conditional variance behavior in the sense that negative shocks generate volatility more than positive shocks of equal magnitude. This phenomenon has been attributed to the “leverage effect” (Black, 1976) and “volatility feedback” hypothesis (Koutmos, 1997). [...]

[...] The conditional volatility of stock returns in the U. S. has also been extensively examined, most notably by French et al. (1987), Nelson (1991) and Baillie and DeGennaro (1990). Couhray and Rad (1994) investigate the time series properties of five developed markets (U. K., France, Italy, Germany and Netherlands). More recently, Koutmous (1998) models the major stock indices of nine industrialized nations using a threshold GARCH methodology. The GARCH type models have also been employed to explain the behavior of smaller European as well as Emerging Stock Markets. [...]

[...] He documents changes in the ARCH parameters before and after the 1987 crash, but finds no evidence of risk premium and volatility persistence. In passing, we note that SSA Stock Markets have received little attention in the (international) time-varying literature. Indeed, to the best of our knowledge [apart from Brooks, et al., (1997) and Appiah-Kusi and Pescetto (1998)] the time varying volatility aspects of these markets have not been previously examined. Accordingly, this paper extends this line of research to the SSA markets. Bekaert and Harvey (1997) examine volatility of 20 emerging markets. [...]