Beyond Sorting: A More Powerful Test for Cross-Sectional Anomalies

Abstract

Standard-Nutzungsbedingungen:Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden.Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen.Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may AbstractMany researchers seek factors that predict the cross-section of stock returns. The standard methodology sorts stocks according to their factor scores into quantiles and forms a corresponding long-short portfolio. Such a course of action ignores any information on the covariance matrix of stock returns. Historically, it has been difficult to estimate the covariance matrix for a large universe of stocks. We demonstrate that using the recent DCC-NL estimator of Engle et al. (2016) substantially enhances the power of tests for cross-sectional anomalies: On average, 'Student' t-statistics more than double.

The Power of (Non-)Linear Shrinking: A Review and Guide to Covariance Matrix Estimation

“…But we did compare DCC-NL to DCC-S, which uses the sample covariance matrix of the series {ŝ t } to estimate C, and to DCC-Lin, which uses linear shrinkage applied to the series {ŝ t }. We found that for N = 100, all three methods performed about equally well but that for N = 500 and N = 1, 000, DCC-Lin and DCC-NL outperformed DCC-S by a considerable margin, with DCC-NL being the clear winner; more specifically, the improvement of DCC-NL over DCC-Lin was of the same magnitude as the improvement of DCC-Lin over As a further application, in Ledoit et al (2019), we showed how to use the DCC-NL estimator to construct more powerful tests for cross-sectional anomalies, that is, more powerful tests to establish the validity of a so-called return anomaly (also called factor or return-predictive signal) whose goal it is to explain the cross-section of expected stock returns. Traditional tests construct dollar-neutral long-short portfolios that load on the return anomaly under study by sorting the stocks into quantiles according to their anomaly scores; if such a zero-cost portfolio can be shown to deliver a positive expected return with statistical significance, the anomaly under study is established as 'successful'…”