Estimating betas on daily data for a small stock market

This paper examines the properties of different market risk (beta) measures computed on daily data for a thin security market i.e. the Helsinki Stock Exchange in Finland. In accordance with the results by Dimson and Marsh (1983) the paper shows that differences in trading frequency between different stocks produce a serious bias towards what appears to be stability in estimated betas. Furthermore the paper shows that when betas are computed the exclusive use of stock prices based on actual trades will not solve the problem of a thin trading bias in measured stability of these beta estimates. In fact other methods proposed to cope with the thin trading problem seem to be at least as efficient as use of trade-to-trade returns. Finally betas corrected for differences in trading frequency are still shown to be statistically related for firm size.