Efficient Wine Pricing Using Stochastic Frontier Models

We specify a two-tier stochastic frontier model to estimate a price equation for U.S. Rieslings and determine overpricing and underpricing of specific wines. The data is for 2000-2016. Using tasting score and tasting score squared as sole independent variables, we found evidence of overpricing and underpricing of wines. When additional location variables were entered, empirical estimates for this two-tier model were infeasible, suggesting that both underpricing and overpricing did not occur simultaneously in this market. Separate one-sided stochastic frontier models were then specified and estimated to test for either under or overpricing in this market. Results showed that overpricing exists and not underpricing, but the extent of overpricing is minor, suggesting that wineries do an efficient job pricing their wines, given tasting scores, conditional on regional location premiums or discounts. We also investigated the effect of scale on pricing for a smaller set of wineries.