This empirical paper examines the effect of stock ownership by sell-side analysts on their reporting. I hypothesize that the reports of stock-owning-analysts are more optimistic than those of non-stock-owning analysts. I identify and collect a sample of companies where the analyst disclosed that she held stock in the company on which the report was written. Since owning-analysts choose to own stock, selection bias may result. The empirical tests attempt to address this bias by incorporating a selection model and other empirical techniques. The research design tests for the presence of stock-owning-analyst reporting bias by comparing their earnings forecasts and stock recommendations to those of non-owning analysts. The results of an ordered probit find that owning-analyst recommendations are slightly less optimistic than those of control analysts. The marginal effect, however, is small. Regression analysis shows that owning-analyst earnings forecasts are slightly more optimistic than those of control analysts, on average. The selection model results provide support for the necessity of controlling for endogeneity. The selection model's second stage regression offers only weak evidence of owner optimism. Additional tests, however, do not support a change in analyst reporting, both in quality and frequency, during the period of analyst ownership, relative to periods of non-ownership. These results are robust to a separate sample of owners identified based on Securities and Exchange Commission Form 144 filings (See Appendix) and alternative research methodologies. Overall, I conclude that owning-analysts appear to be optimistic, but the results suggest that optimistic analysts choose to own stock. In addition, based on this sample, I find evidence that underwriting co-managers of Initial Public Offerings (IPOs) issue optimistic recommendations and their earnings forecasts contain the average optimistic bias. In contrast, I find IPO lead underwriters do not issue optimistic recommendations and their earnings forecasts are less biased, on average.
