Intermediaries and product quality in used car markets

• Author: Gary Biglaiser,Yiyi Zhou,Fei Li,Charles Murry
• Published date: 01 September 2020
• DOI: http://doi.org/10.1111/1756-2171.12344
• Date: 01 September 2020

RAND Journal of Economics

Vol.51, No. 3, Fall 2020

pp. 905–933

Intermediaries and product quality in used

car markets

Gary Biglaiser∗

Fei Li ∗

Charles Murry∗∗

and

Yiyi Zhou∗∗∗

We present empirical evidence supporting that used cars sold by dealers have higher quality:

(i) dealer transaction prices are higher than unmediated market prices, and this dealer premium

increases in the age of the car as a ratio and is hump-shaped in dollar value, and (ii) used cars

purchased from dealers are less likely to be resold. In a model, we show that these empirical

facts can be rationalized either when dealers alleviate information asymmetry, or when dealers

facilitate assortative matching. The model predictions allow us to distinguish these two theories

in the data, and we find evidence for both.

1. Introduction

Most transactions are made through a variety of intermediaries such as retailers, dealers,

and brokers. Since there is no place for intermediaries in Arrow–Debreu’s highly stylized world,

to understand the ubiquitousness of intermediaries, one must count on market frictions. One

obvious rationale is offered by Rubinstein and Wolinsky (1987): intermediaries can facilitate

searching and matching between parties in decentralized markets. Moreover, when goods are

heterogeneous, and tastes of agents are idiosyncratic, intermediaries could also improve the allo-

cation or match efficiency. Another popular justification of intermediaries relies on frictions due

to an informational asymmetry between agents. As Biglaiser (1993) and Lizzeri (1999) argue,

intermediaries can serve as information intermediaries, or certifiers, in markets where there are

∗University of North Carolina at Chapel Hill; gbiglais@email.unc.edu,lifei@email.unc.edu.

∗∗Boston College; charles.mur ry@bc.edu.

∗∗∗Stony Brook University; yiyi.zhou@stonybrook.edu.

This article supersedes the previous work titled “Middlemen as Information Intermediaries: Evidence from Used Car

Markets.” The authors thank Eric Bond, Liran Einav, Igal Hendel, Brad Larsen, Qihong Liu, Alessandro Lizzeri, Brian

McManus, Peter Newberry, John Rust, Tobias Salz, Henry S. Schneider, Karl Schurter, Andrei Shleifer, Shouyong Shi,

Senay Sokullu, Randy Wright, Andy Yates, Jidong Zhou, editor Katja Seim, anonymous referees, and participants of

many conferences and seminars for helpful comments.

© 2020, The RAND Corporation. 905

906 / THE RAND JOURNAL OF ECONOMICS

motives for adverse selection or consumer sorting. The idea is that intermediaries have more ad-

vanced technology and experience to distinguish product quality, so goods traded through them

are of higher quality than those traded directly between sellers and buyers. Although the the-

oretical literature has proposed many distinct rationales for intermediaries, empirical research

is limited and focuses almost exclusively on how intermediaries alleviate search frictions. The

goal of this article is to examine the role of used-car dealers more comprehensively. We provide

evidence supporting the hypothesis that car dealers provide high-quality cars for consumers, mo-

tivated either by information asymmetries or by match efficiency motives.

Using administrative registration records of used car transactions from two large states, we

examine the prices and resale patterns of cars sold through dealers and cars sold directly by

sellers. First, we document a dealer price premium: for the same type of car, transaction prices of

dealer sales are higher than transaction prices in the unmediated market.1Second, we show that

the dealer premium, in dollars, is hump-shaped in car age and as a ratio is increasing in the car

age. Third, we document that used cars purchased from dealers are less likely than unmediated

transactions to be resold within a short time after the initial transactions. Although there may be

many sources of the dealer premium, we argue that these observations are consistent with the

hypothesis that part of the dealer premium is due to dealers offering superior cars.

We formalize our dealer quality premium argument with a parsimonious theoretical model

to understand a dealer’s role in a market with a depreciating good that may experience a failure,

or in Akerlof’s parlance, become a lemon. When selling a car, a seller can visit a dealer, who

decides how much, if anything, to offer for the car. The seller can either trade with the dealer or

go to the unmediated market and sell the car directly to buyers.

Based on these ingredients, we show that two prevailing theories about intermediaries can

explain empirical observations. First, we assume that the seller privately observes the quality of

the car, but dealers are experts who can run a test to ascertain quality. The unmediated market

understands that the dealer is an expert and has reputation concerns; therefore, dealers trade

higher-quality cars on average and enjoy a price premium over the unmediated market. Since the

age, or vintage, of a car represents the hazard of the car becoming a lemon, it has an important

effect on the dealer’s price premium. On the one hand, the dealer’s information role increases

as the car ages, but on the other hand, even a high-quality car depreciates as it ages. We show

that this leads to a dealer premium pattern described above. Besides, the dealer’s expertise in

screening car quality generates a selection mechanism: cars purchased through dealers are more

likely to be of high quality than those purchased from sellers directly. By the classic adverse

selection logic, lemons will be resold sooner than high-quality cars.

Second, weconsider a model with complete infor mation but consumer heterogeneity:buyers

have either a high or lowvaluation. In this case, dealers serve as a platform to facilitate assor tative

matching between buyers and sellers in the presence of search frictions. In equilibrium, dealers

only sell high-quality cars and attract high-valuation buyers. In the unmediated market, cars with

both high and low quality are sold, and low-valuation buyers purchase these cars. The dealers’

price premium is justified by the matching efficiency they create. We show that the age profile of

a dealer’s price premium is also consistent with the data. Also, the initial allocation is inefficient

in the unmediated market: buyers with low valuations may purchase high-quality cars, giving

them an incentive to resell their cars. Therefore, the model also predicts that the resale rate is

higher when the car is purchased from the unmediated market than when it is purchased from

a dealer.

After presenting the theory, we turn back to the used car data to distinguish between the

asymmetric information and sorting theories. Distinguishing these two theories is essential for

two reasons. First, it allowsus to identify the source of the inefficiency in the used-car market, by

better understanding the role of dealers in this market. For example, imagine that a dealer sells a

1Throughout the article, we will refer to cars sold by dealers as mediated transactions and cars sold by sellers to

buyers without going through a dealer as direct or unmediated transactions.

C

The RAND Corporation 2020.