Standard procedures in empirical asset pricing suffer from various issues that are common to all regression-based methods. This work reviews recently introduced approaches that aim to mitigate problems associated with omitted factors and errors-in-variables. New methods addressing the omitted-variable bias suggest procedures for selecting appropriate control variables, aggregating the information from a large set of factors, or making existing methods robust against omitted factors. While the omitted-variable problem is present in almost all standard empirical asset pricing methods, the errors-in-variables problem is largely limited to the estimation of factor premia via two-pass regressions. New methods addressing the errors-in-variable bias implement an instrumental variable approach, suggest a generalized version of the widely used portfolio sorts procedure, or correct estimates based on an analytic expression for the bias. Ultimately, all of these new methods represent highly relevant advances for the area of empirical asset pricing, and the possibility to synthesize the most promising approaches might be worthwhile to investigate in the future.

中文翻译：

---

实证资产定价新方法的文献综述：遗漏变量和变量误差偏差

经验资产定价的标准程序存在各种问题，这些问题是所有基于回归的方法所共有的。这项工作回顾了最近引入的方法，旨在减轻与遗漏因素和变量误差相关的问题。解决遗漏变量偏差的新方法提出了选择适当控制变量、从大量因素中汇总信息或使现有方法对遗漏因素具有鲁棒性的程序。虽然几乎所有标准的经验资产定价方法都存在遗漏变量问题，但变量误差问题主要限于通过二次回归估计因子溢价。解决变量误差偏差的新方法实施了工具变量方法，建议广泛使用的投资组合排序程序的广义版本，或基于偏差的分析表达式的正确估计。最终，所有这些新方法都代表了经验资产定价领域高度相关的进步，未来可能值得研究综合最有前途的方法的可能性。