CHAPTER 1. PACKAGING AND COSTLY CONSUMER SEARCH 2
effect on a firm's profit as setting the service price to zero. Consumers with low
valuations for the service are inefficiently forced to consume the service, and may
be driven away by the higher package price. This is also the conclusion reached by
the bundling literature: the literature suggests that adding the package price may
increase profits. The literature also points out that setting the package price only is
weakly dominated by setting all the prices, and is strictly dominated as long as the
firm does not wish to sell the service to all the consumers (Adams and Yellen 1976,
McAfee et al 1989).2 Metering costs may explain why some services are not priced
separately, but cannot easily explain many examples of packaging.3
The main idea of the paper is related to the search cost literature, which suggests
that consumers incur a cost to find the prices charged by different firms. This paper
extends the search cost rationale to a single firm that sells multiple goods. In order to
acquire information about prices, a consumer must incur an incremental search cost
for discovering each price. Consider the hotel example: in the search cost literature a
potential guest must incur a cost to find the room price at each hotel. Extending the
idea to multiple prices charged by a single hotel, a consumer must incur a (possibly
small) extra cost to find the prices of breakfast, wireless internet, or minibar drinks
prior to booking the hotel.4 A similar situation arises when consumers open a bank
account or choose a calling plan, but cannot easily compare the fee schedules of
different banks or cellular phone operators.
Suppose that a hotel offers rooms and a single service, and guests are heterogeneous
in their valuations for the room and for the service. Further suppose that
guests must incur a cost to find the service price prior to the booking decision (the
2For example, McAfee et al (1989) '...immediately rule out pure bundling as a (uniquely) optimal
strategy, because mixed bundling is always (weakly) better'. Chen and Nalebuff (2007) find that a
firm sets a service price to zero when (among other conditions) the cost of the service is zero.
3In many cases, metering costs seem low: metering internet surfing is easy and done by many
coffeeshops, charging a one-time debit card fee is simple, and cellular-phone airtime must be metered
even when some minutes are included. In the hotel example, metering costs are unable to explain
the variation across services (high-metering-cost minibar is excluded, low-metering-cost breakfast is
included). Moreover, metering costs cannot easily explain the variation in packaging (e.g. internet
or breakfast) across hotels. The results in this paper are qualitatively different from the results
suggested by metering or other menu cost theories (see the discussion in section 1.4.1).
4In a casual experiment, two phone inquiries about the breakfast price at a single hotel lasted 25
minutes.