Book publishers price discriminate among
their customers by producing a book in hardcover and paperback formats
which they introduce into the market at different points in time. I argue
that this behavior is hard to justify using conventional models where the
distribution of consumer preferences is unimodal. Thus demand estimation
that only allows for single-peaked preferences can lead to erroneous conclusions.
A generalization to bimodal preferences eliminates the problem and affords
a
natural interpretation of the two
peaks as representing libraries and individuals. These two "types" have
very different valuations for a book's content, as well as for time and
quality. Estimation of a structural econometric model that incorporates
the firm's product selection problem confirms the bimodality of consumer
tastes and produces estimates that have very realistic and intuitive implications. |
