Three essays on information flows into equity markets

Resumen

How information is absorbed into the prices of financial assets is one of the central questions of the pricing system – which in turn is key to the whole free market system. In this thesis, I study three distinct ways through which the informational setting surrounding a given security might affect its pricing. In the first chapter, I focus on the effect of coverage by sell-side analysts on stock returns and co-movement across different stocks. Using a differences-in-differences approach, I explore situations coverage terminations due to mergers among brokerage houses. I first estimate that returns on affected stocks tend to co-move up to 65% percent more with returns on stocks of industry peers. Secondly, there is a return premium for stocks covered by fewer analysts even after controlling for betas, size, book-to-market, momentum and liquidity. These results suggest that (1) analysts reduce stock returns co-movement by facilitating absorption of firm-specific information into prices and (2) investors demand a premium to invest in less intensively covered firms, so that analysts do create value. Companies covered by fewer analysts seem to experience a shortage of firm-specific information, which makes investors more reluctant to buy their stock and increases the relative importance of industry and market-wide information for their stock pricing. In the second chapter, I explore the effects of social networks among portfolio managers of equity mutual funds on their trading decisions. I use biographical data on university attendance to detect social connections and find that funds are especially keen on trading in the same direction as socially connected money managers. Funds that are more prone to trade along connected peers display improved performance (as measured by gross returns and using Carhart's four-factor alphas) both before and after expenses. The trend to trade in the same direction as connected peers is uncorrelated to other measures of mutual fund skills and to fund characteristics such as age, size, expenses ratio, turnover and flows of funds. The return gap between funds more prone to herd with connected peers and funds less inclined to do so is highly persistent -- leading to statistically significant differences in fund performance that last over 20 quarters. Results suggest that social connections with other managers represent a significant source of information used by stock pickers. Such ties seem to work as channels through which information is spread over equity markets and absorbed into security prices. In the third and last chapter, I present evidence of persistence in mutual funds' ability to predict short-term returns on individual stocks. I develop a measure for the level of skill with which a fund anticipates the short-term performance of the stock. In turn, I define a stock's set of Best Predictors as the top decile of funds with the greatest predictive power over its abnormal returns. The aggregate behavior of a stock's Best Predictors predicts (1) returns on the stock over the next quarter and (2) institutional demand for the stock in the future. At the fund level, I identify a new dimension of mutual fund skill: its ability to predictive the behavior of the very stocks it holds. This measure predicts differences in future fund performance (that last over 14 quarters) and future fund flows. Evidence suggests that funds tend to specialize in (and become particularly good at) trading some stocks. The community of money managers seems to be (at least partially) aware of such cross-sectional differences - indeed portfolio managers mimic the trading behavior of each stock's own best predictors.