The Distribution of Investor Beliefs, Stock Ownership and Stock Returns

We study theoretically and empirically the relationship between investor beliefs, ownership dispersion and stock returns. We find that high dispersion, measured by high breadth or low Herfindahl index, forecasts returns positively for large stocks, as in Chen, Hong and Stein (2002), but negatively for small stocks. We explain that relationship in a difference-of-opinion model in which stocks differ in the size of investor disagreements and the extent of belief polarization. These differences are characterized by range and kurtosis, respectively. Proxying investor beliefs by analyst forecasts, we find that range and kurtosis affect ownership dispersion in the way that our model predicts