Abstract

The uniqueness of art objects need to be taken into account in the construction of any art market price index. Yet the most widely used methods typically rely on biased samples, discarding a very large proportion of the information available (the repeated sales approach) and/or require strong assumptions regarding the structure and time stability of the market (the hedonic regression approach). In this paper a refined hedonic index is developed that explicitly addresses these problems. An empirical illustration comparing these methods is presented using a dataset of symbolist paintings appearing at auction over the period 1990-2001.