In many businesses it is customary to advertise a base price
for a product and to try to sell additional “add-ons” at high prices
at the point of sale. The quoted price for a hotel room typically
does not include phone calls, in-room movies, minibar items, dry
cleaning, or meals in the hotel restaurant. Personal computers
advertised in weekly sales circulars typically have little memory,
a low-capacity hard disk, and no separate video card. Appliance
stores push extended warranties. Car rental agencies push insurance
and prepaid gasoline. Manufacturers of new homes offer a
plethora of upgrades and options that can add tens or hundreds of
thousands of dollars to a home’s price. In some cases, add-ons can
be thought of as a classic price discrimination strategy: the base
good and the base good plus the add-on are two different quality
levels. In many of the above applications, however, there is a
noteworthy feature absent from the classic price discrimination
model: add-on prices are not advertised and would be costly or
difficult to learn before one arrives at the point of sale.
In this paper I address two questions: why do firms offer
high-priced add-ons; and what difference does it make? More
attention is given to the latter question. I focus on add-ons with
unadvertised/unobservable prices, but also discuss the more
traditional model with all prices observable. Add-ons are
clearly a major source of revenue for many firms, and some
consumer groups complain bitterly about them.1 Whether we
should care much about add-ons is not obvious, however. The
above examples all involve fairly competitive industries. The
classic Chicago-school argument would be that profits earned
on add-ons will be competed away in the form of lower prices
for the base good.2
The analysis focuses on a simple competitive price discrimination
model. Two firms are located at the opposite ends of a
Hotelling line. Each firm has two products for sale: a base good
and an add-on. The add-on provides additional utility if consumed
with the base good. The consumer population has both
vertical and horizontal taste heterogeneity. There are two continuums
of consumers: “high types” with a low marginal utility
of income; and “low types” or “cheapskates” with a high marginal
utility of income. Within each subpopulation, consumers
have unit demands for the base good with the standard uniformly
distributed idiosyncratic preference for buying from
firm 1 or firm 2.
A crucial assumption inherent in this specification is that
“high type” consumers are both more likely to buy high-priced
add-ons and less likely to switch between firms to take advantage
of a small price difference. This is intended to fit two types of
applications. The traditional application would be discrimination
between wealthy versus poor consumers (or businessmen versus
tourists). Both assumptions about behavior would be natural
consequences of wealthy consumers’ having a lower marginal
utility of income. A second “behavioral” application would be to
sophisticated versus unsophisticated consumers, with unsophis-