Three essays on stock market anomalies, behavioral finance, and financial econometrics

This dissertation studies two important stock market anomalies, the correlation between stock returns and inflation and the predictability of stock returns. Chapter 1 is an introduction. Chapter 2 investigates why the stock return-inflation relation changes over time. Kaul (1987) considers changes in the monetary policy regime, while Hess and Lee (1999) propose changes in the composition of structural shocks. I show in Chapter 2: (1) different from Kaul (1987) and Hess and Lee (1999), both changes in the monetary policy regime and changes in the composition of structural shocks can in principle cause changes in the stock return-inflation relation; (2) empirically, the change in the monetary policy regime is quantitatively more important in explaining the data. In Chapter 3, I propose a new test that is particularly powerful against the type of alternative proposed by the recent behavioral models. When the test is applied to the data, I find evidence supporting the behavioral models in that (1) prices of stocks with more uncertainty and slower information diffusion tend to have both short-run positive and long-run negative autocorrelations; (2) the three-factor model cannot explain all observed autocorrelation patterns. The results are not likely due to data mining, because similar autocorrelation patterns are found in different sets of portfolios, different stock markets, different sample periods, and even for using different intervals to measure autocorrelations. Motivated by the same behavioral models and the contradictory empirical evidence regarding the stock price reaction to the common factor, in Chapter 4, I propose a regression-based test that is robust to serial correlation and heteroskedasticity in stock returns. When the test is used to the data, contrary to Lewellen (2002), I find evidence in support of the behavioral models in that stock prices also short run under- and long-run overreact to market wide information