Reduced Form Evidence on Belief Updating under Asymmetric Information - The Case of Wine Expert Opinions

We estimate the effect of revealing expert opinion labels on wine product purchases through a field experiment where a random subset of wine products within the con- sumers' retail shelf choice set are labeled in the treatment store. We use a detailed weekly product level panel scanner data set for labeled and unlabeled wines in the treatment and comparable control stores before and after the implementation of a product-level labeling field experiment. We combine the scanner data with additional information on the characteristics of each product, such as brand, varietal, region of production, and price to estimate the average and heterogeneous effects of the field experiment on wine consumption. Consistent with earlier work, we find there to be a positive and significant overall average effect and that demand increases more for higher score wines than for lower score wines. We advance the literature with the following: one, higher scores matter more for prices in the lower quartile of the overall wine price distribution whereas demand does not move for the higher priced wine quartile, once quality is revealed; the result is consistent with pre treatment consumer behavior where consumers infer high quality for high prices. Two, we find positive spillover effects of this experimental treatment within brand for untreated wines as the displayed average score of the wine brand increases. However, we also obtain negative spillover effects for untreated wines that belong to intensively treated brands.