Ahn, Seung C.
[VerfasserIn]
;
Gadarowski, Christopher
[Sonstige Person, Familie und Körperschaft];
Perez, Marcos Fabricio
[Sonstige Person, Familie und Körperschaft]
Two-Pass Cross-Sectional Regression of Factor Pricing Models
Anmerkungen:
Nach Informationen von SSRN wurde die ursprüngliche Fassung des Dokuments Octuber 31, 2010 erstellt
Beschreibung:
This paper examines the asymptotic and finite sample properties of the two-pass cross-sectional regressions estimators, when the factors and the asset returns are conditionally heteroskedastic and/or autocorrelated. Using a minimum distance approach, we derive heteroskedasticity- and/or autocorrelation-consistent (HAC) standard errors for the two-pass estimators in a systematic and intuitive way. Moreover, we identify the optimal two-pass estimator that is asymptotically more efficient than other two-pass estimators. Also, we derive an HAC model specification test statistic which is an extension of Shanken's GLS residual test statistic (1985). Our Monte Carlo simulations provide evidence of the importance of controlling for autocorrelation in the two-pass estimation. We show that the t-tests computed with HAC standard errors produce more reliable inferences than the tests computed with the standard errors by Fama and MacBeth (1973) or Shanken (1992) when model disturbances are autocorrelated. We also find that the t-tests based on the optimal two-pass estimator require the use very large samples to provide reliable results. Lastly, we find that neither Shanken's (1985) specification test nor the robust specification test derived in this paper have desirable finite-sample properties when disturbances are autocorrelated