Stock return distribution in the BRICS

Abstract

Stock returns in emerging market economies exhibit patterns that are distinctively different from developed countries: returns are noted to be highly volatile and autocorrelated, and long horizon returns are predictable. While these stylized facts are well established, the assumption underlying the distribution of returns is less understood. In particular, the empirical literature continues to rely on the normality assumption as a starting point, and most asset pricing models tend to overstretch this point. This paper questions the rationale behind this supposition and proceeds to test more formally for normality using multivariate joint test for skewness and kurtosis. Additionally, the paper extends the literature by examining a number of empirical regularities for Brazil, Russia, India, China and South Africa (the BRICS for short). Our main findings are that the distribution of stock returns for the BRICS exhibits peakedness with fatter and longer tails, and this is invariant to both the unit of measurement and the time horizon of returns. Volatility clustering is prevalent in all markets, and this decays exponentially for all but Brazil. The relationship between risk and return is found to be significant and risk premiums are prevalent in our sample.