|
|
|
| |
Seminar by Prof. Guanhao Feng, City university of Hong Kong
|
Date | Thursday, 17 July 2025 |
Time | 10:30 a.m. – 11:30 a.m. |
Venue | RR301, Run Run Shaw Building |
|
Title | Testing and comparing factor models: A portfolio-based framework |
Abstract |
We present two papers introducing novel statistical approaches with economic interpretations for testing high-dimensional asset pricing factor models.
First, we introduce the Sample Splitting Alpha (SSA) test, a novel framework for evaluating models with a large number of test assets. Our SSA test integrates variational representations of existing statistics with sample splitting and ridge-regularization techniques, enhancing the robustness and reliability of the test statistic. Notably, the SSA test directly links to the out-of-sample information ratio, providing a practical metric for evaluating the benefits of adding test assets to a factor model and facilitating direct comparison and ranking of competing models. Monte Carlo simulations demonstrate the feasibility of the SSA test, supported by empirical evidence that the U.S. equity cross-sectional average returns are well-explained by factor models in terms of out-of-sample evaluation.
Second, we propose a Maximum Random Sparse Portfolios (MRSP) test for evaluating asset pricing linear factor models when the number of test assets exceeds the time periods. Our method constructs L random sparse portfolios, each comprising K equally weighted assets, and identifies the most challenging portfolio to test factor models. Under mild regularity conditions, we show that the MRSP test statistic converges to a type I extreme value distribution. Monte Carlo simulations highlight the MRSP test’s robustness, supported by empirical evidence that the U.S. equity cross-sectional average returns are well-explained by factor models in recent years. |
About the speaker |
Guanhao (Gavin) Feng is an associate professor of finance and statistics at the City University of Hong Kong. He earned his Ph.D. and MBA from the University of Chicago in 2017. Gavin focuses on developing methodological solutions, including machine learning, Bayesian statistics, and financial econometrics, to address big data challenges in empirical asset pricing. His work has been published in leading journals such as the Journal of Finance, Journal of Financial Economics, Journal of Financial and Quantitative Analysis, Journal of Econometrics, and International Economic Review. He is the principal investigator for various external research grants, such as the HKRGC ECS and GRF grants, and the NSFC youth science fund. Gavin’s research has been acknowledged by practitioners, receiving research awards from INQUIRE Europe, Hong Kong Institute for Monetary and Financial Research, and the AQR Insight Award.
|
|
|
| |
|
|
|
|
|