| Summary: | <p>This thesis studies the role of contrast effects and biased expectations in financial decisionmaking and financial markets. The first study explores the role of skewness preferences in dynamic decision-making at the hands of salience theory. Previous research suggests that otherwise risk-averse people are willing to take risks if the outcome distribution is positively skewed. Salience theory can explain this by assuming that states with high contrasts between the outcomes attract attention and their probability is overestimated. Skewness preferences are particularly important in dynamic setups because these allow agents to endogenously create skewness through the choice of their stopping strategy. I extend salience theory to a dynamic setup and show that it predicts that agents will take gambles if the expected value is not too negative. Moreover, if they gamble they choose stopping strategies that yield positively skewed outcome distributions. These predictions differ both from expected utility theory and other behavioural models. I test the predictions experimentally and find broad support.</p>
<p>In the second study, I examine whether the earnings forecasts of analysts after a firm announces its earnings depend on the earnings surprises of companies that announced shortly before the firm. Evidence from a plethora of domains suggests that the interpretation of information depends on how it compares to contrasting information. Thus, the earnings of a given firm might look worse the better other firms perform. I find that positive earnings surprises of other firms make analysts revise their forecast of a firm’s earnings upwards but, at the same time, make their forecast more pessimistic relative to the true earnings. This result is in line with a positive news channel in combination with a contrast effect channel of the other firms’ earnings surprise on the analysts’ forecasts.</p>
<p>In the third study, I develop a method to test if a given return predictor reflects mispricing rather than risk. Asset pricing research has uncovered hundreds of characteristics that can predict the cross-section of returns, but the nature of many of these remains elusive. If a predictor is linked to returns through risk, it should be unrelated to changes in the market’s expectations about firm profits. Alternatively, return predictably can be explained by biased expectations. If a return predictor captures this form of mispricing, it should predict changes in expectations in addition to returns. I use the earnings forecasts of professional analysts as a proxy to the market’s expectations and test for 173 return predictors if they can also predict forecast revisions. I find that around 40% of predictors can do so and, thus, reflect mispricing.</p>
|
|---|