Taming the Factor Zoo: A Test of New Factors

Abstract

We propose a model-selection method to systematically evaluate the contribution to asset pricing of any new factor, above and beyond what a high-dimensional set of existing factors explains. Our methodology explicitly accounts for potential model-selection mistakes, unlike the standard approaches that assume perfect variable selection, which rarely occurs in practice and produces a bias due to the omitted variables. We apply our procedure to a set of factors recently discovered in the literature. While most of these new factors are found to be redundant relative to the existing factors, a few-such as profitability-have statistically significant explanatory power beyond the hundreds of factors proposed in the past. In addition, we show that our estimates and their significance are stable, whereas the model selected by simple LASSO is not.

Risk Factor Centrality and the Cross-Section of Expected Returns

“…The authors also point out that a non-sparse model (up to 15 explanatory variables) is needed whenever risk factors are applied to explain cross-sectional returns. Feng, Giglio, and Xiu (2020) and Freyberger, Neuhierl, and Weber (2020) use a lasso-type estimator to reduce their respective set of risk factors, also reaching a non-sparse result whenever the issue at 10 Ledoit and Wolf (2004), Engle, Shephard, and Sheppard (2008) and Brito, Medeiros, and Ribeiro (2018). 11 Given a matrix J, the conditional number can be expressed as KLMN%J+ $ OPO Q OPO 5' , whereas %J+ $ R if J is singular.…”

“…It consists of 51 risk factors, the first being the Excess Market Return 43 gathered from the French Library 44 , while the remaining 50 are zero-investment, long-short portfolios constituted by well-known traits described in the asset-pricing literature. In accordance withFeng, Giglio, and Xiu (2020), we split each risk factor into six types of groups: Value…”

“…To learn more, seeHastie, Tibshirani, and Friedman (2009). 5 For definition purposes, we consider high-dimensional environmental models yielding more than 10 predictors.6 Harvey, Liu, and Zhu (2016),Green, Hand, and Zhang (2017),Yan and Zheng (2017), Feng, Giglio, and Xiu (2020), Freyberger, Neuhierl, and Weber (2020 andKozak, Nagel, and Santosh (2020).! 7 A key component of the precision matrix is the most central element, according to some centrality measure, of the adjacent matrix whom is computed by the precision matrix.8 The precision matrix is given by the inverse of the covariance matrix for any given vector of random variables.Electronic copy available at: https://ssrn.com/abstract=3691331…”

“…Factor loading is the correlation coefficient between the principal component and the original random variable.38 Feng, Giglio, and Xiu (2020), Freyberger, Neuhierl, andWeber (2020) andKozak, Nagel, and Santosh (2020).! Electronic copy available at: https://ssrn.com/abstract=3691331…”

Are stock prices driven by expected growth rather than discount rates? Evidence based on the COVID-19 crisis

“… 1 Recent empirical asset pricing models employ historical returns and factors to predict stock prices. We omit a voluminous literature review related to asset pricing and, instead, refer to Harvey et al ( 2016 ), Pätäri and Leivo ( 2017 ) and more recently Feng et al ( 2020 ) for a thorough review of empirical asset pricing models, challenges and skepticism pertaining to them. …”