Bundling and quality assurance

• Author: Kathryn E. Spier,James D. Dana
• Published date: 01 March 2018
• Date: 01 March 2018
• DOI: http://doi.org/10.1111/1756-2171.12222

RAND Journal of Economics

Vol.49, No. 1, Spring 2018

pp. 128–154

Bundling and quality assurance

James D. Dana Jr.∗

and

Kathryn E. Spier∗∗

With imperfect private monitoring, a firm selling two experience goods can increase both pro-

ducer and consumer surplus by bundling. Bundling constrains consumers to buy two products,

making consumers better informed and ensuring that they use tougher punishment strategies.Both

increased monitoring and increased punishment benefit other consumers, so bundling overcomes

a free-rider problem. The social value of bundling is even larger if consumers cannot attribute a

negative signal to the specific product that generated it, or if one of the two goods is a durable

and the other is a complementary nondurable. Our results are robust to mixed bundling.

1. Introduction

Although tying and bundling have been viewed with suspicion by US courts—and are

per se illegal in some circumstances—it is common for quality assurance to be used as an

explicit efficiency defense in antitrust cases.1In the 1930s, General Motors successfully invoked

quality assurance to defend their business practice of requiring their dealers to use only General

Motors parts in the aftermarket service and repairs of their cars.2In the 1980s, Mercedez-Benz

successfully argued that if their dealers could procure parts directly from independent suppliers

rather than through Mercedes-Benz of North America (MBNA), then dealers would shirk on

quality-control testing, free-riding on the efforts of others, and deliver a substandard product to

consumers. The court found “ample evidence to support a finding that the tying arrangement is a

∗Northeastern University; j.dana@northeastern.edu.

∗∗Har vardUniversity and NBER; kspier@law.harvard.edu.

We would like to thank Heski Bar-Isaac, Claire Chambolle, Louis Kaplow, James Hosek, Howard Marvel, Marco

Ottaviani, James Peck, Martin Peitz, Patrick Rey, Luis Vasconcelos, Michael Whinston, the referees, participants at the

2014 E.A.R.I.E. conference in Milan, Italy,and especially Barry Nalebuff, for helpful comments. We thank Susan Norton

for editorial assistance. Dana acknowledges financial support from Harvard Business School and the Yale School of

Management. Spier acknowledges financial support from NSF grant no. SES-1155761. An earlier version of this article

was circulated with the title “Bundling and Firm Reputation.”

1Tying is permissible “if implemented for a legitimate purpose and if no less restrictivealter native is available.”

Phonetele, Inc. v. American Tel. & Tel. Co. (664 F.2d 716, 738-39 [9th Cir. 1981]). Kaplow (1985, p. 545 at N. 121)

provides discussion of additional cases.

2Pick Mfg. Co. v.General Motors Corp. et al. (80 F. 2d 641 [7th Cir.1935]). “Defective parts, preventing efficient

operations of cars, bring dissatisfaction with automobiles themselves. The material result is blame of the manufacturer

and consequent loss of sales.”

128 C2018, The RAND Corporation.

DANA AND SPIER / 129

legitimate means of maintaining the quality of Mercedes replacement parts supplied by dealers,

and thereby protecting the reputation of the MBNA product.”3

This article argues that bundling may be necessary to assure the quality of experience

goods when monitoring is private and imperfect. We show that a multiproduct firm that bundles

its products together can assure quality more effectively than a multiproduct firm that sells its

products separately and unbundled. Intuitively, consumers who purchase the full product line

from one firm develop a deeper personal experience with that firm’s products and capabilities

than consumers who mix-and-match their purchases across different firms. Purchasing a bundle

of products makes consumers better private monitors and better private enforcers of product

quality. With bundling, consumers detect product quality deviations more readily and retaliate

with greater force by boycotting all of the bundled products.4Because private monitoring and

private enforcement are public goods, product bundling raises social welfare.

We consider a formal model in which a multiproduct branded firm competes with a fringe

of competitive manufacturers who produce unbranded, low-quality products. Consumers do not

directly observe product quality; instead, consumers receive imperfect private signals of quality

after making their purchase decisions. The branded firm is (almost surely) capable of producing

high-quality products but can also produce low-quality products. In each period, the branded firm

decides whether or not to invest in product quality for each of its twoproducts and sets the prices

for those products. Consumers have heterogeneous preferences for the branded firm’s products.

Consumers purchase each period, and those that buy from the branded firm observe imperfect

private signals of the branded firm’s quality,which they then use to update their beliefs about the

branded firm’s product quality over time.

We begin by considering a technology where a negative private signal about one product is

informative about the future quality of both products. In this case, a consumer would naturally

stop purchasing both branded products after observing a negative signal about just one of them.

Bundling is valuable in this setting, because it forces consumers to experience both products

instead of just one, increasing the chance that low quality will be detected. Specifically, we show

that the range of parameter values for which high quality is sustainable in a perfect Bayesian

equilibrium is larger with bundling than without bundling. For parameters in this range, bundling

increases consumer surplus as well as producer surplus.

Wethen extend the model to consider a technology where a negative signal about one product

is informative about the future quality of onlythat product. In this case, a consumer would naturally

stop purchasing a product after observing a negative signal about that particular product, but may

rationally continue to purchase the firm’s other product. Bundling creates value in this setting

in two ways. As before, bundling constrains consumers to experience multiple products and to

be better monitors. However, in this case, bundling can also make the punishments more severe.

Many consumers would balk at paying a high bundled-product price for just one high-quality

product, and therefore many consumers would stop buying the bundle after just one negative

signal. Again, more intense monitoring and more severe punishments benefit consumers as well

as the firm.

Taken together, the analysis shows that bundling can solve two free-rider problems. First,

without bundling, some consumers shirk on monitoring by purchasing only one of the firm’s

products. Bundling increases the proportion of consumers purchasing two products, and this

improves the quality of their privatemonitoring. Second, without bundling, a consumer may shirk

on punishing the firm by continuing to purchase one of the branded firm’s products even after

observing a negative signal about the other product. Bundling makes more severe punishments

3Mozart Co. v. Mercedes-Benz of North America Inc. (833 F. 2d 1342 [9th Cir. 1987]). Although this example

is significantly more complex than our model because MBNA is a reseller, not a manufacturer, of automobile parts, it

nevertheless clearly illustrates that quality assurance is facilitated by constraints on buyerswho would otherwise free-ride

on other buyers’ behavior.

4Bundled pricing makes such enforcement rational. Following a bad experiencewith just one product, consumers

are not typically willing to pay the bundled price for the remaining products.

C

The RAND Corporation 2018.