Abstract
Using a unique hand-collected dataset, comprising all artwork sales in Italy between 2006 and 2010, we test Rosen’s and Adler’s hypotheses in the modern and contemporary visual art market. We extract our measures of artist talent and fame from a set of observable artist-specific variables by means of a factor analysis and estimate the elasticities of income with respect to talent and fame. Consistent with Rosen’s and Adler’s hypotheses, our results suggest a convex relationship between income and talent and a linear relationship between income fame. Using SUR models to evaluate the effects of artist talent and fame on the average trade prices and number of sales, we find that the number of artwork sales is the main ‘channel’ through which talent and fame influence income. Copula models provide additional insights on the nature of the conditional dependence relationship between average prices and number of sales. Poolability tests suggest a single model of artist income applies to all artists in our dataset whether their works are generally traded in auction houses or galleries, so it is not necessary to specify different models. Finally, quantile regressions reveal that artists in low-income quantiles are not superstars, Rosen’s hypothesis holds only above the median income, and fame plays no role for low-income quantiles.