Research Summary
Research Summary
Portfolio Betas Do Not Make for Better Asset Pricing Tests
Description
Many papers claim that because using portfolios instead of individual stocks as test assets minimizes idiosyncratic volatility, their use also yields more precise estimates of risk premia. I show that while portfolio formation does lead to more efficient beta estimates, the standard errors of factor risk premia estimates depend only on the cross-sectional distribution of the factor loadings, not on the precision of the estimated betas. In contradiction to the standard claim, portfolio formation actually increases these asymptotic standard errors and handicaps tests of asset pricing models.