Date: Wednesday, 12 February 2025, at 12.30pm (CET)

Dual format: ESSEC Paris La Défense (CNIT), Room TBA
and via Zoom, please click here

Testing for the existence of alpha — the portion of expected returns that cannot be explained by risk exposures — is central in empirical asset pricing, and yet poses a number of issues. Available testing procedure generally exhibit low power, require strong assumptions on the data generating process, and often involve the estimation or inversion of (often large) covariance matrices. We introduce a randomized testing methodology that addresses all these problems as it does not require estimation of any covariance matrix, while allowing both N and T to grow large, with the former possibly faster than the latter. In contrast with extant approaches, the new procedure can accommodate conditional heteroskedasticity, non-Gaussianity, and strong cross-sectional dependence in asset returns. We also propose a de-randomized decision rule to choose in favor or against correct specification of a Linear Factor Pricing model. In simulation, this testing procedure displays satisfactory size and power properties, and compares favorably to several existing tests. Results on the decision rule are equally encouraging. An application to constituents of the S&P 500 shows that linear factor pricing models correctly price large caps U.S. stocks during calm and expansionary market regimes.

Kind regards,
Jeremy Heng, Olga Klopp, Roberto Reno, Marie Kratz and Riada Djebbar (Singapore Actuarial Society - ERM)