Research

It is well known in the credence-good literature that in an expert-client relationship, under the Liability assumptions, clients have to reject the expert’s serious-treatment recommendations with a positive probability to ensure that the expert honestly recommends treatments. Inefficiency arises because some socially efficient treatments are not provided. We show that the expert can enhance clients’ trust, or acceptance rate of the serious treatment, by providing intrinsically socially inefficient customer service upon recommending the serious treatment. Enhanced clients’ trust leads to higher efficiency and higher profit for the expert. However, trust cannot be enhanced by providing customer service with different timing.

We analyze the optimal delegation problem between a principal and an agent, assuming that the latter has state-independent preferences. Among all incentive-compatible direct mechanisms, the veto mechanisms -- in which the principal commits to mixing between the status quo option and another state-dependent option -- yield the highest expected payoffs for the principal. In the optimal veto mechanism, the principal uses veto (i.e., choosing the status quo option) only when the state is above some threshold, and both the veto probability and the state-dependent option increase as the state gets more extreme. Our model captures the aspect of many real-world scenarios that the agent only cares about the principal's final decision, and the result provides grounds for the veto delegation pervasive in various organizations.

We study a general credence goods model with N problem types and N treatments. The expert seller and the client communicate through cheap talk. We find that the expert’s equilibrium payoffs can be characterized geometrically as the quasiconcave envelope of his belief-based profits function under discriminatory pricing. We establish the existence of client-worst equilibria, apply the geometric characterization to previous research on credence goods, and provide a necessary and sufficient condition for when communication benefits the expert. For the binary case, we solve for all equilibria and characterize the client’s possible welfare among all equilibria.

An expert seller chooses an experiment to influence a client's purchasing decision, but may manipulate the experiment result for personal gain. When credibility surpasses a critical threshold, the expert chooses a fully-revealing experiment and, if possible, manipulates the unfavorable result. In this case, a higher credibility strictly benefits the expert, whereas the client never benefits from the expert's services. We also discuss policies regarding monitoring expert's disclosure and price regulation. When prices are imposed exogenously, monitoring disclosure does not affect the client's highest equilibrium value. A lower price may harm the client when it discourages the expert from disclosing information.

We consider a large society where players can search for a match at a cost. The matched pairs play the repeated prisoners’ dilemma game subject to voluntary separation. Moreover, players can search for an alternative partner while in a relationship. We find that when the search cost is moderate, the option to perform in-relationship search, while not being exercised on the equilibrium path, promotes cooperation and makes the relationship long-lasting.

A principal hires an expert to collect information and then make a decision, utilizing both the expert’s private information and informative public opinion. The optimal contract induces the expert to sometimes defy public opinion even when public opinion is more informative than his private information. Our finding is robust to allowing for switching the arrival times of different signals, expert reporting his private information, expert's reputational concern and repeated interactions.

We study the marriage and remarriage market in a repeated-game framework with potentially alternating partners. We show that compared with a label-less society, introducing marital labels by a credible institution can i) identify the innocent bachelor(ette)s, and ii) effectively replace inefficient trust-building phase by efficient transfer punishment, therefore achieving welfare improvement. With the presence of marital labels, the socially optimal matching protocol is to arrange as many bachelor-divorced marriages as possible. We also characterize equilibrium behavior for a decentralized society where marital status is either observable or unobservable at dating stage and derive the conditions under which different institution outperforms others.

A key attribute of many digital products is that they may be designed as flexible, allowing buyers to learn and modify their designs post-purchase. In this paper, we examine a digital product market where sellers exert cognitive efforts to find an innovative design, and if unsuccessful, tradeoff between offering basic and flexible designs. We fully characterize pure-strategy equilibria for various market structures and demonstrate that moderate competition promotes seller cognition, enhances buyer surplus, and enhances social welfare, while excessive competition can harm buyers. Among other extensions, we find that government subsidies should be directed toward sellers with more effective cognitive technologies.

We study a bilateral trade environment in which a seller exerts cognitive efforts to identify the optimal product design for a buyer. Should the seller fail to discover the innovative design, he has the option to induce additional cognition from the buyer by making the product flexible at an additional cost. We show that sequential cognition is more likely to occur in the decentralized equilibrium than in the first-best case. Furthermore, the seller always under-provides cognition, while the buyer may incur excessive cognitive costs in equilibrium. In an extended model, we find that the seller may further decrease cognitive efforts when facing competition.

Complementary to the existing literature that extensively studied credence goods markets in static settings, we develop a dynamic model in which a durable good breaks down stochastically after treatments, and the customer meets the expert recurrently. We assume that the minor treatment alleviates the symptom of the major problem but fails to cure it, increasing the future failure rate. In contrast to the literature, we show that the truth-telling equilibrium never exists under the verifiability assumption, because the standard equal-margin condition fails.

In our dynamic setting, the expert has a stronger incentive to undertreat since undertreatment induces more future business. But on the other hand, the customer becomes less willing to pay for the minor treatment for fear of increased future payments. Therefore, depending on the relative magnitude of these two opposing forces, either Undertreatment or Overtreatment can emerge in equilibrium. Surprisingly, the expert’s incentive to undertreat weakens as the increment of failure rates rises.