Date of Graduation
This dissertation comprises three empirical essays that tackle various issues concerning the pricing implications of analyst earningsâ€™ forecasts and stock recommendations for U.S. stocks over the period 1993-2011. In the first essay, our objective is to test the overvaluation explanation of the forecast dispersion anomaly conditioning on analyst stock recommendations. Our hypothesis is that future returns are higher (lower) for stocks that currently receive more â€œbuyâ€ (â€œsellâ€) recommendations and have a lower (higher) level of forecast dispersion. We find strong evidence that the forecast dispersion in analystsâ€™ earnings is an important factor in the explanation of returns, once we adjust for differences in risk. Moreover, we provide strong evidence that the impact of forecast dispersion is more pronounced in the group of stocks that receive the least favorable recommendations in a given period. These results are consistent with Millerâ€™s (1977) hypothesis, according to which if short-sale constraints bind, then high opinion divergence stocks will become overpriced and hence have low subsequent returns. In the second essay we study how the information contained in analyst earningsâ€™ forecasts may be related to ex-post stock return volatility for US firms over the period 1983:07-2011:01. We decompose the analyst forecast dispersion using the theoretical framework of Barron et al. (1998) into two components: uncertainty and asymmetric information. We document strong evidence that the uncertainty of analysts about the firmâ€™s future prospects contributes to the increase in the realized stock return volatility. Contrary to our expectations, the asymmetric information reflected in the lack of consensus among analysts in the current period seems to be insignificant in explaining next-period stock return volatility. We further decompose the total stock return volatility into systematic and idiosyncratic volatilities. Our empirical analysis suggests that the uncertainty embedded in the analystsâ€™ forecasts is important in explaining the idiosyncratic volatility, but the lack of consensus due to private information among analysts is important in explaining the systematic volatility. These results suggest that the impact of analyst forecast dispersion on stock prices may be due to both differences of opinion across analysts and information uncertainty about a firmâ€™s fundamentals. In the last essay we examine the effect of loss incidence on the information asymmetry, using annual data for US firms over the period 1986-2011. We use three different measures that proxy for asymmetric information: the level and the change of analystsâ€™ earnings forecast dispersion as well as its decomposed components using the bid-ask spreads. We show that the positive relationship between the loss incidence and forecast dispersion is driven by losses and not financial distress, for both the level and the decomposed measures of forecast dispersion. However, we do not find any significant relationship between reported losses and the change of analystsâ€™ earnings forecast dispersion. Our results show that the significant impact of losses on the analystsâ€™ earnings forecast dispersion is due to a combination of both higher information asymmetry and higher uncertainty regarding the future value of the firm. As such, our study provides a possible reconciliation for the different results provided by Ertimur (2004) and Ciccone (2001) about the incidence of losses.
Papakroni, Jorida, "Three Essays on Analysts' Earnings Forecast Dispersion and Stock Returns." (2013). Graduate Theses, Dissertations, and Problem Reports. 9551.