Please use this identifier to cite or link to this item:
|Title:||Three Essays on Predicting the Equity Premium and Asset Returns|
|Abstract:||This dissertation contains three essays on the predictability of asset returns and its implications. All essays contain tests and applications that could be implemented in real time, and a summary of economic and statistical significance. The dissertation as a whole adds to the growing literature that returns can be predictable, enough so to provide economically significant differences in returns. The same applies to predicting market returns, or market timing. The first essay modifies a popular measure of investor sentiment (Baker and Wurgler, 2006) with the idea in mind that market-wide sentiment will partially reverse itself in the next period. Economic fundamentals are removed to provide a more pure measure of sentiment, which is found to predict returns better (with sentiment having a negative relationship with market returns), resulting in a better investment strategy. The second essay implements a statistical methodology which allows many predictors (portfolio returns) to be aggregated into a composite index. This index predicts returns well, and also gives insight into why two well-observed stock market anomalies, size and value premiums, may occur: they predict future market returns and are therefore an ICAPM state variable that reflects future wealth opportunities. For this reason they carry a risk premium. The third essay provides a new forecasting approach that imposes the restrictions of the popular arbitrage pricing theory (APT) on an existing statistical approach (principal components analysis). The result is that the expected returns of asset positions hedged against systematic risk are better estimated, and average error is greatly reduced out-of-sample.|
|Appears in Collections:||Open Access Dissertations and Theses|
Files in This Item:
|Stivers_Adam_D_FinalSubmission2016August_PhD.pdf||1.73 MB||Adobe PDF||View/Open|
Items in MacSphere are protected by copyright, with all rights reserved, unless otherwise indicated.