Do Factor Models Explain Breaks in the Distribution of Equity Returns?
研究了六因子模型（Fama-French五因子加Carhart动量）能否解释美国股票日收益分布中频繁出现的结构性断点，发现1963年至1990年代中期模型能解释20%-60%的断点，但此后解释力崩溃。
Factor models have become ubiquitous in finance to predict asset returns, estimate risk, and construct investment strategies. The authors investigate whether these models explain the frequent breaks observed in the distribution of equity returns. Using a six-factor model encompassing Fama and French’s five factors and Carhart’s momentum, we analyze 60 years of daily US stock returns. They apply an ensemble of unsupervised changepoint detection methods on the stock returns and factor model residuals to identify breaks in the distribution. Their findings reveal that the factor model reliably explained between 20% and 60% of breaks from 1963 to the mid-1990s. However, the factor model’s ability to account for these breaks has since collapsed. These findings have profound implications for the applications of factor models in estimating risk and returns. The authors advocate including changepoint detection methods in the investment practitioner’s analytical arsenal.