Using Psychology to Understand and Decode Financial Behavior

The first chapter of my dissertation is titled "The Information Content of Realized Losses". In this chapter, I show that we can use ideas from behavioral finance to address a long-standing financial question -- namely, do the trades of informed investors, like company insiders, have predictive power for subsequent stock returns. A large literature shows that the purchases of company insiders have strong positive predictive power. However, much of the literature suggests that company insider sales are, at best, a weak signal for future stock returns. Using an insight from the behavioral finance literature, I show that company insider sales do have predictive power -- we just have to look in the right place. Namely, I show that company insider sales at a loss are a strong negative signal for future stock returns while company insider sales at a gain have no predictive power. This conditioning variable, that is whether one sells at a gain, or sells at a loss, was motivated by the disposition effect -- or the well-documented propensity of investors to avoid sales at a loss and to sell at a gain. In other words, investors are averse to selling at a loss. If a company insider is averse to selling at a loss, and this is the key insight of Chapter 1, he should require an especially strong negative signal to do so, and so when he does sell at a loss, this should predict future negative returns. I examine this insight even further to explore the possible mechanisms behind his aversion to selling at a loss -- I uncover evidence that the insider is averse to selling at a loss because he does not want to admit that his previous purchase decision was a mistake.;The second chapter of my dissertation is titled "Dividends and Trust". In this chapter I provide new evidence on the role of trust in financial decision-making. Using a Dutch individual-level data set, I show that less-trusting investors allocate more of their portfolio to dividend-paying firms than their more-trusting counterparts. I also show that this has relevance for asset prices -- I find that the valuation of dividend payers relative to non-dividend payers is higher in less trusting regions than the relative valuation of dividend payers to non-dividend payers in more trusting regions.;The final chapter of my dissertation, a co-authored work with Aytekin Ertan (London Business School), Stephen Karolyi (Carnegie Mellon University), and Robert Stoumbos (Yale), titled "Pre-earnings Announcement Overreaction", offers evidence on extrapolation, in particular over-extrapolation, in financial markets. We predict that investors will extrapolate from past earnings announcement returns. So, if a firm has recently had very strong earnings announcement returns, investors will expect the firm's next earnings announcement to also be a success. Expecting a good earnings announcement return, these investors will purchase stock in the firm shortly before the earnings announcement. We find evidence that this is the case. Using data from a large discount brokerage, we analyze individuals' purchase decisions in the period right before earnings announcements. We find that previous earnings announcement returns are a strong determinant of individuals' purchase decisions in this period. Additionally, we uncover evidence of return predictability consistent with our story. If a firm is in the top decile of our extrapolated return measure, we expect the firm to have a 5-day pre-earnings announcement return 27 basis points higher than all firms not in the top decile. Additionally, we see evidence of a reversal once the information comes out (i.e. after the earnings announcement).