Impact of unrealized gains or losses on stock returns: theories and tests in an alternative utility theory framework.

Summary This thesis applies prospect theory and regret theory to the study of stock performance and to explain a market anomaly known as the momentum effect. This thesis proposes a theoretical model that links behavioral factors to stock performance and the momentum effect, and then conducts empirical tests to examine the theoretical model. In Chapter 2, the model is established on a concept of potential gains/losses, which indicate whether an investor is currently in a winning or losing situation. Then, the model shows that investors are very reluctant to sell their stocks in a big gain/loss situation. Chapters 3 and 4 perform empirical tests on the potential gain/loss model. The test sample includes all NYSE and AMEX stocks from the year 1982 to 2012. The tests are able to confirm the influence of potential gains/losses on stock returns. In addition, a zero-cost Extremity Minus Mean (EMM) strategy, based on the theoretical model, is documented to be profitable after controlled for risks. In Chapter 5, the potential gains/losses model is developed in a dynamic version. It suggests that the influence of potential gains/losses might persist over a long intermediate period, and generates an upward trend in performance for stocks with a large potential gain/loss. The empirical tests in this chapter focus on the time series evolution of returns. The tests show that stocks with a large potential gain/loss have a stronger upward trend. Chapter 6 applies the results of the previous chapter to explain the momentum effect. The upward trend corresponds to a positive autocorrelation of returns, which is one of the sources that contribute to the momentum profit. The empirical tests in this chapter look at the similarity between the momentum strategy and the potential gains/losses, and also examine the correlation between momentum profit and EMM strategy profit. The tests show that potential gains/losses could contribute to the momentum effect, but are not the only source. The momentum effect may be the result of a combination of several complex factors.