On heterogeneous investor confidence: Theory and evidence

Investors become heterogeneously confident if they have information of different quality or disagree about the quality of the same information. Although the general idea of heterogeneous beliefs is extensively investigated, heterogeneous confidence receives no special treatment yet. This dissertation develops theoretical models based on heterogeneous confidence and short sale constraints. The models produce predictions about trading behavior, volatility, volume and skewness. Empirical evidence supports these predictions. This dissertation calls for more attention to beliefs of higher order moments.; The first chapter develops a static model in which investors who are prohibited from short selling agree to disagree on the precision of a common public signal. The model implies that the equilibrium asset price is a convex function of the signal. The model predicts that (1) short sale constraints increase rather than decrease skewness and (2) disagreement increases skewness. Confirmative evidence is provided. The model also implies that market confidence increases with asset prices and tends to be higher than the average confidence of the investor pool, suggesting that financial markets can endogenously become overconfident.; The second chapter extends the first chapter into a dynamic model and shows that market confidence fluctuates with new information: higher after price gains than after price losses. Market confidence fluctuation provides a unified explanation for asymmetric volatility, volatility clustering and volatility-volume co-movements, through the negative correlation between market confidence and its sensitivity to new information. The model predicts asymmetric volume: heavier after price losses than after price gains. Supportive evidence is found.; The third chapter develops a model in which investors are heterogeneously confident in their priors. Because more (less) confident investors are less (more) sensitive to new information, more confident investors' asset valuation increases (decreases) less with good (bad) news. Thus more (less) confident investors tend to buy with bad (good) news. Using the spectrum dataset, supportive evidence is found. A "trading momentum" is documented: institutional investors tend to continue their buying or selling of a stock, regardless of the direction of price changes. This momentum is robust to time period, institution type and stock size.