These two types of price discrimination can be difficult to distinguish empirically,
because the variables that would be used to reveal a buyer's cost of switching among sellers
or searching for low prices, e.g., income or employment status, would also be correlated
v^dth his reservation price. In recent years, however, the cost of switching sellers has increased
for buyers of leaded gasoline as a result of a change that is not directly related to their
demographic characteristics. Since 1985, when nearly all stations carried both types of regular
gasoline, the proportion of stations that sell leaded gasoline has declined. The average distance
between sellers of leaded gasoline has increased and, more important, has ceased to be equal
to the average distance between sellers of unleaded gasoline.^ If buyers' costs of switching
retailers are important determinants of the margin on gasoline, then an increase in the
distance between sellers of leaded gasohne, without a similar change in the distance between
sellers of unleaded gas, would be expected to increase the margin on leaded gas relative to
the margin on unleaded gas. In the empirical section, I use information on the share of
gasohne in an area that is leaded, and some direct information on the share of stations
carrying leaded gasoline, to diagnose price discrimination based on heterogeneity in switching
costs