| This study attempts to control for the effects of investors' use of information other than accounting earnings in a returns-earnings regression setting by conditioning the analysis on trading volume. This investigation is motivated by the fact that while other information is unobservable, it would induce trading if investors use it for pricing securities.; Using trading volume as an additional explanatory variable in the Liu and Thomas (2000) specification, I find a statistically significant association between trading volume and stock returns but with a marginal improvement in the R-squares of the model. These results suggest that the ability of trading volume per se to proxy for other value-relevant information in the returns-earnings regression setting is limited. In a longitudinal study, I also find that the value-relevance of earnings has declined over the last 20 years, but I do not observe a clear pattern in the value-relevance of other information proxies, i.e., analysts' forecast revisions and trading volume, during that period.; I also investigate whether the difference in the magnitudes of earnings response coefficients between domestic and foreign components of earnings reported by Bondar and Weintrop (1997) is due to the failure to control for proxies for other information since the omitted variables (i.e., proxies for other information) could have biased their results. The evidence supports my conjecture and suggests that the difference in the earnings response coefficients is no longer observed when proxies for other information are included in the regression.; Overall, these results validate Liu and Thomas' claim (2000) that failure to control for other information proxies in the returns-earnings regression can lead to misleading inferences and that future research should control for those proxies when using the regression to address a research question. |
