Diagnosing Asset Pricing Models Using the Distribution of Asset Returns



    Search for more papers by this author
    • School of Business, University of North Carolina at Chapel Hill. This paper is based on my dissertation at The University of Chicago. I would like to thank George Constantinides, Wayne Ferson, Peter Knez, and especially Lars Hansen and John Cochrane for their helpful comments and suggestions.


This paper develops a set of diagnostic tests which can shed light on why a particular model is failing and indicate what steps might be taken to make the model consistent with asset returns. Theoretical bounds on the moments of a stochastic discount factor are derived as a function of the moments of observed asset returns. Particular attention is paid to restrictions on moments other than the variance. These bounds can also be used to measure the information about the distribution of the discount factor contained in the moments of various asset returns. As an application of this methodology, bounds on the discount factor are estimated using size-based portfolios, and the results are used to analyze the small firm effect. Empirical results indicate, for the period 1926–1975, that moments of the returns of small firms contain information about the discount factor that is not contained in the moments of the returns of large firms and/or a proxy of the aggregate wealth portfolio. However, this difference disappears when more recent data is included.