Resurrecting Earnings-to-Price with Robust Control for Outliers
用理论上合理的稳健截面回归方法控制异常值，证明在单因子和多因子模型中，盈利价格比因子和综合盈利预测、修正及广度因子在CRSP、R3000和R2000样本中统计显著，而最小二乘回归因异常值影响未能发现其显著性。
In this article, the authors use a theoretically justified robust cross-section regression method of controlling for outliers to show that, when used in single and multifactor models, the earnings-to-price factor (EP), and a composite earnings forecasts, revisions, and breadth factor (CTEF), are statistically significant for the CRSP®. R3000, and R2000 universes for the time periods 1980–2007 and 2008–2020. Because of adverse influence of outliers, the least squares (LS) regressions with standard 1% Winsorization fail to indicate that EP and CTEF are significant factors. Moreover, the robust regression method is a powerful diagnostic method for detecting overlooked outliers influence on LS results.