PhD Oral Exam - Yushen Li, Economics

When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.

Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.

Abstract

This thesis consists of three chapters. Chapter 1 theoretically studies a discrete model of the exchange of indivisible favors between two infinitely lived agents. With private information, the ex-post efficient exchange of favors is not incentive compatible. Focusing on Perfect Public Strategies, we consider two types of pure strategy within the class of Markov strategies, which we call Bounded Favors Bank strategy (BFB hereafter). In the BFB strategies the net number of favors received by the players serves as a state variable, which is bounded in equilibrium. In interior states, the BFB strategies prescribe the static efficient exchange of favors to be played. In boundary states, we consider two types of the BFB strategy which differ in what is played. On one hand, a type of pure strategy with reward prescribes an agent who provided the most favors (in the negative boundary state) is rewarded by exonerating from favor provision until she receives a favor back; on the other hand, a type of pure strategy with punishment prescribes an agent who received the most favors (in the positive boundary state) is punished by being forced to provide a favor even if the behavior is highly costly. We prove that the two types of the BFB strategy constitute equilibria under private information given a proper set of the payoff parameters and the discount factor. We also show that the payoffs of these two types of BFB strategy can approximate the efficient outcome under private information conditional on agents are sufficiently patient. We then compare the long-run payoffs of the BFB strategies with other important strategies in the literature and find that the BFB strategy with punishment is possible to achieve the highest long-run payoff.
Chapter 2 uses a controlled lab experiment and econometric methods to test the theoretical model in Chapter 1. In the experiment, we examine the behavior of subjects and infer the strategies subjects employ when the information about the costs of favor exchange is complete and incomplete, respectively. From the experiment, we find that, in the favor exchange game with complete information, subjects cooperate to exchange favors substantially more than in the game with incomplete information. When analyzing the strategies, we find that in the treatment with complete information the strategy that is most likely to be played is the efficient one. We also find that subjects only rely on simple strategies under complete information. However, strategies are more complex and the BFB strategies are played with a statistically significant probability under incomplete information. Furthermore, the reward strategy is estimated to be played more often than the punishment strategy when information is incomplete. Our findings suggest that when dealing with a long-term bilateral relationship with private information, using a form of reward may have more compliance than using a form of punishment to achieve higher efficiency.

Chapter 3 studies the behavior and strategies of online users in making contributions to activity of answering questions on an online Question-and-Answer platform. We provide both theoretical and empirical findings on the incentive effect of peer recognition on online content provision. Our theoretical model illustrates how a Bayesian influencer strategically determines her online contribution. Using unique data from the largest Chinese Question-and-Answer platform, we analyze the content provisions of all the influencers with more than 10,000 followers on the platform over two years. By using an instrumental variable approach, we find that a simple OLS method is likely to under-estimate the incentive effect of peer recognition for content provision since the reputational and privacy concerns might have an adverse effect. Our findings suggest that while badges in the Q&A platform make it easier for users to identify the quality of an influencer, those badges may also limit the content contribution since influencers have concerns about their reputation and privacy. It indicates that platforms’ attitudes towards real-name policies may depend on the trade-off between incentives for content creation and content regulation. Policies based on reputation and privacy may have a backlash against policies that encourage traffic.