Credence Goods in the Field - A Forensic-Economic Investigation

Credence goods come in different qualities and have the property that customers do not know which quality fits their need best. They have to seek advice from an expert seller who is also the provider of the good and might have incentives to use his informational advantage to defraud customers. Credence goods are consumed frequently in daily life. Just think of car repair services, where customers hardly know whether the generator has to be replaced, or a new fan belt would do the job; or of medical treatments, where patients cannot judge whether physical therapy cures their backache or a major surgery is needed. Empirical studies typically document the existence of expert misbehavior in markets for credence goods without systematically exploring the conditions leading to it because these studies generally lack a controlled variation of factors potentially important for the extent of fraud. Lab experiments allow for a systematic variation of treatment variables but raise the question of external validity. Natural field experiments are a promising compromise because they allow for a systematic variation of treatment variables, while still investigating the behavior of real people in their natural environment. In the proposed project, we plan to conduct a large scale natural field experiment in the market for computer repair services to provide insight into which kind of fraud is most prevalent in which context and why, and which intervention can contain the amount of fraud. To achieve this goal, we plan to send undercover experimenters with manipulated test computers to different computer experts and ask for a repair. By manipulating the expert`s perception of the information of the consumer about the needed quality, of the social distance to the customer and of other characteristics that are important for an expert`s behavior in a market for credence goods in theory we intend to find out whether there is an relationship between our treatment variables and the behavior of experts and, more specifically, how the extent and the type of fraud depend on those variables and which interventions can contain the amount of fraud. The project intends to add to the existing credence goods literature by systematically exploring the existence of fraudulent expert behavior, the conditions leading to it and possible ways to contain it in a natural field experiment. Ultimately, the results of this study might help policy makers to design institutions for credence goods markets in such a way that fraud is more effectively contained and people are happier about the way they interact on such markets.

In markets for credence goods the providers are better informed about the needs of their customers than the customers themselves. An important consequence of this informational asymmetry is that experts in credence goods markets can defraud their customers. Typical examples of credence goods markets are the health care market, different repair markets and the market for taxi rides in an unknown city. In the context of this project we have analyzed in a series of natural field experiments which behaviors on the customer side could potentially reduce misbehavior on the seller side. One of our main findings is that customers in credence goods markets should only reveal absolute essential information to the expert provider. Additional information like a (correct or incorrect) self-diagnosis, different price- limits or indications that a successful repair would be very important or that an insurance company covers the cost of the repair have in the best case no positive consequences on the repair price. To the contrary, some of these pieces of information even increase the repair price significantly compared to situations where no such information is provided. Furthermore, we show that the probability of defraudation could be reduced by checking internet ratings provided that experts with a lot of 1 and 2 star ratings are avoided. The results of our studies could potentially benefit a lot of people because almost everyone interacts in credence goods markets on a daily basis. To maximize the benefit of this project it is necessary to bring the findings beyond the scientific community to a broader audience. When it comes to reprocess the results of a scientific study for a broader audience there is a potential risk of misunderstandings. Such misunderstandings could be especially problematic in the context of recommendations in credence goods markets. For this reason it must be unambiguously communicated that our findings refer to a specific setting (problems in repair markets which can be diagnosed by an expert without uncertainty) and should not be interpreted as universal results applicable in all credence goods transactions. In health care markets, for example, it could indeed be beneficial if customers bring themselves in when it comes to diagnose the problem.