“…In the following analysis we will mainly analyze hedonic regressions that relate the log of the price of paintings to a number of explanatory variables and time dummies. An advantage of our empirical analysis compared to previous investigations on historical markets (Etro and Pagani, , ; Etro et al., ) is that we are able to control not only for fundamental characteristics such as size, support, originality of the work, attribution to a school and a painter and so on, but also for the characteristics of the transaction that vary from auction to auction: the lot number, the length of the catalogue description, the average price of the other paintings sold in the auction, the characteristics of the dealers organizing the sale and buying the lots and the associated network structure. To some extent, our information is even more complete than what is usually available for modern auctions: indeed, the identity of the buyer is typically secret in modern auctions, which makes it impossible to analyze networks of buyers and collusive practices between auction organizers and buyers.…”