However, many studies support that time-varying volatility in equity market returns exhibits an asymmetric eff ect (e.g., French et al., 1987; Nelson, 1991; Glosten et al., 1993). is means that a stock or a stock market index entails more risk following an unexpected negative return, than

it does following an unexpected positive return of the same magnitude. Hence, asymmetric reverting patterns in return dynamics cannot be captured from the conventional autoregressive model restricted by the constant serial correlation coeffi cient. In this case, a nonlinear autoregressive model that allows serial correlation to the change in response to a prior positive and negative return shock is required. erefore, we also examine the shortrun interrelationships between the stock markets by testing the overreaction hypothesis using an asymmetric nonlinear smooth-transition generalized autoregressive conditional heteroskedasticity (ANST-GARCH) model and propose the most appropriate strategy to be followed by international asset managers, traders, and investors in the Balkan stock markets.