The hypothesis that price discrimination based on willingness to pay for quality can occur in multifirm markets is confirmed using microdata on gasoline retailing. A test that discriminates between price structures associated with discrimination and with cost-driven, competitive differentials is developed and implemented with controls for variation in outlet and market characteristics. A second test based on profitability variation rejects a competitive, peak-load pricing explanation for the observed price dispersion. The data suggest that price discrimination at the retail level adds at least 9¢ a gallon to the average price of full-service gasoline.

The hypothesis that price discrimination based on willingness-to-pay for quality can occur in multifirm markets is confirmed using microdata on gasoline retailing. A test that discriminates between price structures associated with discrimination and with cost-driven, competitive differentials is developed and implemented with controls for variation in outlet and market characteristics. A second test based on profitability variation rejects a competitive, peak-load pricing explanation for the observed price dispersion. The data suggest that price discrimination at the retail level adds at least nine cents a gallon to the average price of full-service gasoline. Copyright 1991 by University of Chicago Press.

Matching models usually assume an exogenously given distribution of match productivity, and the act of changing jobs then has the worker taking a new, independent sample from this distribution. Using a "characteristics" approach to matching two heterogeneous populations, this article shows that assumptions concerning the normality and serial independence of match productivity (across successive matches) follow from some simple axioms. Moreover, the normality assumption receives support from an empirical test that uses data on the output of a large group of workers.