Announcement and Publication Decisions and the Use of Prediction Errors in Cross‐Sectional Analysis: Evidence on Seasoned Equity Issues

Abstract

AbstractAn important concern for empirical analysis of market responses to public releases of information is whether the voluntary nature of announcement and publication decisions imparts a selectivity bias and, if so, how this bias affects empirical results. Using a database of SEC filings of equity issues, all related Wall Street Journal published announcements are identified. The published announcements are used to classify proposed issues into subsamples with homogeneous information structures. Announcement and WSJ publication decisions are found to be non‐random but, for industrial firms, not significantly related to firm size or exchange listing. Coefficients estimated without controlling for selectivity are found not to be significantly biased for either industrial or utility issues. However, for industrial issues, significance levels are considerably lower when selectivity bias is controlled. The hypothesis common to other studies of new issue announcement effects that observations from different information structures can be pooled, with coefficient estimates restricted to be equal across information structures, is rejected. The results have implications for the methods used to study cross‐sectional market responses