Testing forecasting power of the conditional relationship between beta and return

The purpose of this paper is to investigate the forecasting power of the conditional relationship between beta and international stock returns.

Using the market model, the individual betas for each country in the sample are estimated by ordinary least square. The conditional relation between beta and return is tested by estimating cross‐sectional regressions for each month (i.e. 343 cross‐sectional regressions). Mean value of coefficients and t‐statistics (one tail) are computed to test whether the mean values of coefficients are significantly positive and negative. Whether there is a systematic relationship between the bull market and the bear market is also tested.

Overall, no support of the model was found. Although positive, there is insignificant relationship between beta of current period when excess market return is positive with next period stock return. Moreover, although negative, there was insignificant relationship between beta of current period when excess market return in negative with next period stock return. Similar results were found when the sample was divided into January and non‐January months.

This contribution is to test the validity of conditional relationship between beta and stock returns in international setting by considering the effect of current period up‐ and down‐markets on the next period stock return. If the conditional version of capital asset pricing model (CAPM) holds, there must be a consistent result, i.e. similar to those obtained by the studies using contemporaneous relationship. It is important to perform such tests on the conditional version of CAPM since an ex post state dependent model may not be used as a forecasting model.