| Popis: |
This dissertation consists of three essays on empirical asset pricing. In the first paper, we examine the impact of test criteria in identifying true asset pricing factors. We focus on the Sharpe ratio and pricing performance improvement. While both criteria are exposed to model misspecification, we find that pricing performance criteria are inferior as their performance is driven by estimation bias. Through an empirical application, we demonstrate the impact of the criteria on the subset of selected factors. In the second paper, we study the cross-section of corporate bonds by utilizing a large set of financial statements, equity and bond characteristics. We use a predictive regression framework and the adaptive Lasso to choose the most relevant characteristics. Applying the adaptive Lasso, we find a ten-factor model, with value, bond reversal, and equity momentum spillover being the dominant factors. We evaluate the economic benefits of investing according to the predictions of the adaptive Lasso and find significant benefits in terms of absolute and risk-adjusted returns. In the third paper, we evaluate the ability of U.S. corporate bond fund managers to generate alpha. We apply the False Discovery Rate (FDR) to distinguish between 'skill' and 'luck'. We find that long-term out-performance remains elusive, with only 1% of the funds able to generate significant alpha over their life. However, fund managers can generate alpha over the short-term, with the proportion of skilled funds increasing to 13.5% when we examine three-year sub-periods. |
