Do U.S. paintings follow the CAPM? Findings disaggregated by subject, artist, and value of the work

Abstract

This paper investigates to what extent paintings by U.S. artists born before WWII can be treated like capital assets, and whether the findings are specific to artist, subject matter, and value of the work. The capital asset pricing model (CAPM) in its standard static form is applied to painting returns from 1971 to 1996. Price indices and returns for various groupings of paintings derived from large sample hedonic regressions are used to test alternative forms of the standard CAPM. In the first stage time series estimation, betas for various data groupings are computed to test the degree to which the CAPM explains returns. In general the CAPM signals no factors other than market risk which might explain painting returns. Betas generally are found to be below one with high priced works having betas close to zero and sometimes negative. U. S. paintings appear to have little systematic risk, and thus may provide useful diversification. In a second stage test of the CAPM the computed betas are treated as a long run characteristic accounting for excess returns of the asset. In this cross sectional re-estimation, little support is found for the consistency of the CAPM although high priced paintings show some support. U. S. paintings appear to follow the CAPM to a degree similar to that of traditional capital assets, and thus behave like capital assets regardless of investment desirability. For high value works the CAPM conformity is strongest and diversifications value the highest.