high-income store. These differences in size suggest a more limited number of produce items for
lower-income stores. Quantity differences, as shown in Table 1, show similar size effects, but
store 2 is now more comparable to store 5, than it is to store 4. This shift suggests entirely
different purchasing patterns for higher- and lower-income shoppers. That is, if purchase
bundles are similar, then comparable dollar sales should lead to comparable quantity sales. This
expectation stems mainly from the fact that prices are identical across all stores.
Even though prices are identical across all stores, a quick glance at Table 1 shows that
lower-income shoppers pay lower prices for all fruit and vegetables, except bananas and yellow
vegetables. For these two sub-categories, lower-income shoppers pay either slightly higher or
statistically identical prices. These price data suggest that lower-income shoppers make a special
effort to purchase the lowest-priced commodities within a given sub-category. For example, a
lower price can be realized for potatoes by purchasing pre-sorted bags of potatoes, instead of
self-selecting potatoes from bulk bins. Similar tradeoffs can be made for commodities like
apples and citrus. And while exact tradeoffs cannot be observed because the data are at a store
level as opposed to an individual shopper level, observed differences in prices paid clearly
suggest that lower-income consumers are careful shoppers. What is not clear is the extent to
which income constraints dictate these shopping preferences.
Model Development
One of the key factors motivating this research is whether higher- and lower-income
consumers have different demand responses to changes in prices for various sub-categories of
fruit and vegetables. To address this issue, an error component model is specified and estimated.
Since the data set is a cross-section of higher- and lower-income stores over a time period of