Detecting Abnormal Returns of Infrequently Traded Stocks in an Event Study

Abstract

Most emerging financial markets have a liquidity problem. Brown and Warner (1985) showed that the problem causes bias in measuring abnormal returns by means of a traditional event study. A number of studies detect abnormal returns of thinly traded stocks by using either different models to estimate an expected return or different statistical tests. This study aims to find the best combination of both approaches by means of simulation. The result supports the rank test in most cases. The selection of the return model is trivial except when the event creates negative abnormal returns and increases variance.