On the Estimation of Beta-Pricing Models

Abstract

An integrated econometric view of maximum likelihood methods and more traditional two-pass approaches to estimating beta-pricing models is presented. Several aspects of the well-known “errors-in-variables problem” are considered, and an earlier conjecture concerning the merits of simultaneous estimation of beta and price of risk parameters is evaluated. The traditional inference procedure is found, under standard assumptions, to overstate the precision of price of risk estimates and an asymptotically valid correction is derived. Modifications to accommodate serial correlation in market-wide factors are also discussed.