A different approach to estimating betas of securities subject to thin trading and serial correlation

This paper enquires whether the parameters of asset pricing models can be better represented by cointegration analysis to correct the bias in β estimates. Due to the existence of correlation in lagged series, cointegration analysis, or regression in levels, would produce better estimates of asset pricing model parameters than the regressional analysis of rates of return if the series are cointegrated. In addition, the estimation is empirically simpler.