On the uncertainty of art market returns

•We estimate art market returns using public auction data.•We employ a Hedonic Pricing Model and then use a wild bootstrap method.•We find that the resulting confidence intervals are very wide.•Point estimates of returns are insufficient to make art investment decisions.•Our findings question the validity of previous art-related returns research.

We examine the returns based on auction data for two individual artists and two groups of artists. We employ a Hedonic Pricing Model correcting for the log-transformation bias followed by a wild bootstrap method. This approach allows us to specify confidence intervals for the return estimates. We find that the resulting confidence intervals are wide; therefore, relying solely on point estimates of returns to derive conclusions can be misleading. This situation calls into question the validity of previous research and also shows that the very concept of return in the context of art markets is quite elusive.