| I examine analytically and empirically how security analysts' incentives to generate trades affect earnings forecasts and stock recommendations. Analysts interested in maximizing the trading volume generated from their reports will look for information that causes investors to change the allocation of securities in their portfolios. For stocks that investors will buy, more precise information increases the number of shares bought by reducing the investor's risk premium. However, for stocks that investors will sell, more precise information may decrease the number of shares sold. Hence, my model highlights a fundamental conflict of interest between analysts and investors regarding provision of information on poorly performing stocks. In addition, the model makes the following empirical predictions: (1) An analyst is more likely to issue a report about a stock when he is optimistic, leading to a positive selection bias in analyst reports. This positive selection bias is reduced when there are opportunities to capture commissions from negative information, and (2) Earnings forecasts will be more accurate when an analyst is optimistic about the stock's performance. These predictions are tested using analysts' earnings forecasts. The selection bias tests consider the effect of firm size on the availability of pessimistic analyst forecasts. The tests of forecast accuracy use several proxies for analyst optimism, including upward forecast revisions, positive earnings changes, and longer-term earnings expectations. The results of the tests are consistent with the hypothesis that expected stock performance affects analysts' production and release of information about a stock. |
