Abstract This study proposes a novel methodological framework that integrates extreme value theory and hedonic regression models to analyse the price formation in fine art auctions. Hammer prices reflect extreme upper-tail realizations of the distribution of bidders’ reservation prices, and standard hedonic approaches centred on average outcomes are not well-suited to capture this crucial feature of the pricing process. In the paper, hammer prices are modelled explicitly as upper-tail observations using a Generalized Extreme Value (GEV) specification embedded within an otherwise standard hedonic framework. Using a sample of Picasso paintings sold at auction between 2000 and 2024, the analysis constructs and compares hedonic price indices based on OLS, median regression, and GEV. The analysis shows that explicitly accounting for tail behaviour results in more stable and informative measures of price dynamics. Bridging the gap between the traditional hedonic approaches and the actual auction pricing mechanisms, this paper aims to provide an integrated framework for constructing art price indices in thin and volatile markets.
Scorcu et al. (Tue,) studied this question.