Measuring event impacts in thinly traded stocks

The purpose of this paper is to suggest simple procedures designed to cope with the effects of thin trading on event study tests. The procedures are directed at two central problems: (i) missing individual stock returns (i.e., days on which no trading is observed in a security), and (ii) the effect of a bid-ask spread on the time series behavior of daily stock return data. We attack these problems by explicitly incorporating them in the construction of a generating process for observed security returns. First, we develop a procedure for "filling in" missing returns. Then, we model a return-generating process of observed security returns that allows estimation of the variance of unobserved true security returns for use in hypothesis testing.