Abstract
This thesis includes three essays in empirical asset pricing across financial markets. The first essay examines factor momentum in commodity futures markets. Based on the US and UK data from 1985 to 2022, we first show that a commodity factor’s past returns positively predict its future returns. This predictability is at its strongest over the one-month horizon, and could be explained by mispricing. The factor momentum sug gests mean-variance inefficient commodity factors and negatively impacts the efficiency of pricing models. We then construct the time-series efficient factors, which exhibit higher Sharpe ratios and improve the performance of pricing models. These findings are ro bust across international commodity futures markets, but the transaction costs erode the economic gains of factor momentum and efficient factor strategies due to high portfo lio turnover. Overall, our results point to the potential to time commodity factors and highlight the importance of conditional asset pricing in commodity futures markets.The second essay investigates factor momentum in mutual fund markets. Employing a comprehensive dataset of US equity mutual funds from 1984 to 2024, we identify a significant positive association between past 12-month factor performance and subsequent performance. This persistence holds across multiple performance metrics and is evident at both time-series and cross-sectional levels. The profitability of these strategies is consistent with a mispricing channel. These findings contribute to the long-standing debate on the predictability of fund alphas and offer practical insights for investors seeking to exploit persistent signals in mutual fund characteristics.
The third essay examines the role of memory-based beliefs in shaping asset prices in the Chinese equity market. Leveraging a well-established psychological model of memory and data on A-share stocks from 2000 to 2023, we show that memory cues systematically influence stock prices. The magnitude of these effects is conditioned on key memory related properties: recency, interference, and similarity. Furthermore, after controlling for economic linkages and risk-based factors, we show that memory-induced mispricing remains significant but short-lived, suggesting the effects are not information-driven. This essay advances our understanding of investor trading behavior and its implication for asset pricing in the context of associative memory.
| Date of Award | 15 Oct 2025 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Xiaoquan Liu (Supervisor) & Ying Jiang (Supervisor) |