Three Essays on Stock Market Anomalies and Investments

Access & Terms of Use
embargoed access
Embargoed until 2025-05-15
Copyright: Wang, Haoxu
My thesis consists of three essays. My first essay is on relative strength anomalies. After long being one of the main puzzles in asset pricing, momentum has ironically become a case of observational equivalence. It can now be explained both by behavioral factors capturing mispricing and by the neoclassical-inspired investment q-factors. Besides, q-factors explain the related 52-week-high anomaly. We note that recent tests subsuming both anomalies are unconditional exercises while the bulk of momentum profits are predictable and occur in bull markets and after periods of low volatility. Comparing asset pricing models conditionally, we find the unconditional fit is misleading. The models fit well most of the time but not when the profits are produced. Noticeably, q-theory implies time-varying loadings that are not consistent with the data. On the other hand, consistent with an underreaction channel, earnings announcement returns and analyst forecast errors both decrease steeply with lagged volatility. My second essay is on portfolio optimization. We comprehensively examine whether advances in the asset-pricing and covariance matrix literatures can jointly improve the out-of-sample (OOS) performance of mean-variance efficient (MVE) portfolios. Focusing on the 500 largest stocks, we find that, after accounting for transaction costs, MVE portfolios formed using improved inputs do not outperform the passive strategy. However, their after-cost performance can be substantially improved by combining several ideas available in the literature. Portfolios that simultaneously target risk, manage transaction costs, correct the covariance matrix for OOS errors, and use simple linear Fama-MacBeth return forecasts attain net Sharpe ratios greater than one, significantly outperforming the passive portfolio. My third essay is on factor momentum. Factor momentum recently joined the ongoing debate over the causes of stock momentum. We find that neither momentum in “off-the-shelf” factors nor momentum in high-eigenvalue principal component factors can explain any previously proposed momentum driver. Also, compared to previous drivers, factor momentum does not exhibit superior performance in capturing momentum-like anomalies. Like the competing models, it cannot explain stock momentum conditionally. Moreover, it cannot explain stock momentum after accounting for transaction costs while these can explain the persistence of factor momentum, especially in less systematic factors.
Persistent link to this record
Link to Publisher Version
Link to Open Access Version
Additional Link
Conference Proceedings Editor(s)
Other Contributor(s)
Corporate/Industry Contributor(s)
Publication Year
Resource Type
Degree Type
PhD Doctorate
UNSW Faculty