Overconfidence in the Markets for Lemons

Date01 April 2016
Published date01 April 2016
DOIhttp://doi.org/10.1111/sjoe.12135
AuthorDaniel Müller,Fabian Herweg
Scand. J. of Economics 118(2), 354–371, 2016
DOI: 10.1111/sjoe.12135
Overconfidence in the Markets for
Lemons
Fabian Herweg
University of Bayreuth, DE-95440 Bayreuth, Germany
fabian.herweg@uni-bayreuth.de
Daniel M¨
uller
University of Bonn, DE-53113 Bonn, Germany
daniel.mueller@uni-bonn.de
Abstract
We extend Akerlof’s “Market for Lemons” (1970, Quarterly Journal of Economics 84, 488–
500) by assuming that some buyers are overconfident. Buyers in our model receive a noisy
signal about the quality of the good that is on display for sale. Overconfident buyers do
not update according to Bayes’ rule but take the noisy signal at face value. We show that
the presence of overconfident buyers can stabilize the market outcome by preventing total
adverse selection. However, this stabilization comes at a cost: rational buyers are crowded
out of the market.
Keywords: Adverse selection; behavioral industrial organization; noisy quality signals; over-
estimation; overprecision
JEL classification:D82; L15
I. Introduction
In his influential contribution, Akerlof (1970) has shown that if markets
are plagued by asymmetric information, only inferior goods – or, in the
extreme, no goods at all – are traded in equilibrium. This theoretical predic-
tion of adverse selection is at odds with casual observations about markets
for used goods functioning quite well. For instance, consider the epony-
mous market for used cars. Here, it seems reasonable to presume that the
potential seller of a used car has a good idea of the quality of her car,
whereas the potential buyer obtains only little information whether the car
on display is a “lemon” or a “cream puff”.1According to Akerlof, the trade
We have benefited from comments made by Magdalena Helfrich, Johannes Maier, Takeshi
Murooka, Piers Trepper, Xiaoyu Xia, and two anonymous reviewers. All errors are, of course,
our own.
1The terms “lemon” and “cream puff” are colloquial language. The former term describes
a car that is found to be defective only after it has been bought, whereas the latter term is
used to describe a car in very good condition.
CThe editors of The Scandinavian Journal of Economics 2015.
F. Herweg and D. M¨
uller 355
volume and qualities offered on this market should be low. However, the
trade volume of eBay Motors in the US is approximately 36,000 cars sold
each month (Lewis, 2011), which is reasonably high. Empirical works in-
vestigating markets for used cars also find only weak evidence for adverse
selection (Bond, 1982; Genesove, 1993).
We provide one possible explanation for this puzzling gap between theo-
retical predictions and empirical observations by positing that buyers might
be overconfident in the following sense: an overconfident buyer believes
he can judge the quality of the good by inspecting it when, in fact, he only
receives a noisy signal about the good’s true quality. For instance, the buyer
might consider himself an expert for used cars, believing that after a test
drive he will know whether the car is a lemon or a cream puff. However,
a short test drive does not necessarily reveal whether the car was involved
in an accident and then cosmetically repaired, making a test drive a noisy
signal at best.
We show that if the buyer is overconfident with positive probability,
there always exists an equilibrium in which high-quality goods are traded.
Moreover, if it is sufficiently likely that the buyer is overconfident, all
pure-strategy equilibria are pooling equilibria in which all qualities are
traded. The presence of overconfident buyers can thus stabilize the market
and help to prevent the total adverse selection outcome. However, this
market stabilization comes at a cost: rational buyers no longer participate
in the market. Furthermore, the presence of overconfident buyers can create
incentives for sellers to invest in signal obfuscation (i.e., shrouding the true
quality even more). Although it might increase the probability of trade if
the signal is made noisier, which in turn increases expected welfare, we
show that sellers’ costly investment in signal obfuscation often reduces
welfare, in particular if the problem of adverse selection is severe.
We introduce buyer overconfidence into an otherwise standard and fairly
simple model of a market with asymmetric information. There is a sin-
gle seller and a single buyer. The seller, who owns one unit of the good,
knows whether her item is of high or low quality. The buyer, however,
cannot directly observe the quality of the good but receives a private
noisy signal about the quality of the item on display. We focus on highly
uninformative signals so that in the benchmark case, where the buyer
is a rational Bayesian updater, the unique perfect Bayesian equilibrium
outcome in pure strategies is total adverse selection: only items of low
quality are traded in equilibrium. This benchmark outcome is in contrast
to the findings we obtain when the buyer might be overconfident, be-
cause then high-quality items are always traded in equilibrium with positive
probability.
CThe editors of The Scandinavian Journal of Economics 2015.

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT