Testing the univariate conditional CAPM in thinly traded markets

Traditional tests of asset pricing undertaken within the CAPM framework have to control for nonsynchronous trading and non-trading as well as volatility clustering in especially thinly traded financial markets. This investigation therefore set out to control for nonsynchronous trading and non-trading effects and volatility clustering in the Norwegian equity market. The problem is approached by applying a linear ARMA-GARCH-in-mean lag specification. The ARMA lag specification controls for nonsynchronous trading and non-trading effects in the mean equation. The GARCH lag specification controls for conditional heteroscedasticity and volatility clustering in the latent conditional volatility equation. All lags are Schwarz efficient. The results suggest that the conditional CAPM cannot be rejected but the in-mean parameter in ARMA-GARCH-in-mean specifications show very low statistical significance except for daily data. The result therefore suggests a compensation for risk only for short time-horizons and the in-mean parameter in ARMA-GARCH-in-mean lag specifications is a poor proxy for risk in the conditional CAPM sense. Conditional heteroscedasticity and volatility clustering need to be controlled for in daily and weekly time intervals while nonsynchronous trading needs to be controlled for in daily, weekly and monthly time intervals.