1 Introduction
A firm’s reputation is based on the information its customers hold about its past performance.
Such information may simply result from own past experience but more often than not it will
actually stem from information exchange—from what consumers have heard from others. In
fact, most brand reputation appears to be based on such word of mouth. In many cases, such
as luxury cars or high-end stereos, consumers who know about a product’s high quality far
outnumber those who have actually tested it. Consequently, the way consumers exchange
information matters for firms’ reputation and, ultimately, for market outcomes. Consumer
networks are the carrier of firms’ reputations.
In this paper we use a simple experiment to shed some light on how the structure of consumer
networks affects the performance of markets that suffer from moral hazard. Firms sell
experience goods of unknown quality. Quality may be either good or bad. Buyers are matched
to sellers and decide whether or not they want to buy a single unit of the good. When making
this decision, buyers are provided with some information about the seller’s past track record.
The amount of information they are provided with depends on the underlying network
structure of the buyer population. We consider four treatments: (NO) an anonymous one-shot
benchmark where firms have no labels and reputation building is excluded by design;
(DEGENERATE) a degenerate network without information exchange such that each buyer
only knows his own experience; (PARTIAL) a partial network where buyers are located on a
circle and everybody knows their own experience and that of their right-hand neighbor;
(FULL) a full network where everybody has access to the entire history of all buyers and, due
to one-to-one matching, the entire history of all sellers.
We find that markets fail completely when interaction is anonymous and one-shot. Without
identification of firms there is no reputation building and market failure is almost complete.
Only 5% of all matches result in mutually beneficial trade. Once there is identification that
does allow reputation building market performance is vastly enhanced. In treatment
DEGENERATE where all buyers just remember their own experiences the number of
mutually beneficial trades almost quadruples compared to the anonymous benchmark. Buyers
trust more often and sellers offer vastly better quality (see also Bolton et al. 2004, Bohnet and
Huck 2004, Bohnet et al. 2005, or Bracht and Feltovich 2007).