Interaction of volatility and autocorrelation: evidence from major stock exchanges

Compares stock market returns behaviour for six stock markets in order to find out whether nonlinearities are a result of conditional heteroscedasticity or of previous performance. Uses LeBaron’s exponential generalized autoregressive conditional heteroscedasticity model to link conditional variance with first order correlation. Applies it to daily stock market indexes from 1986 to 1991 in Canada, France, Germany, Italy, Japan and the United Kingdom. Finds that the links exist in all the markets, with high autocorrelation during stable periods, and none under high volatility, for daily but not weekly returns. Concludes that nonsynchronous trading leads to an inverse relationship between volatility and autocorrelation.