The Politics of Attention (with Lin Hu)

We develop an equilibrium theory of attention and politics. Following the seminal work of Downs (1957), we assume that voters are rationally inattentive, meaning that they can process the most relevant information for decision making at a cost measured by entropy reduction. In a generalized Downsian model of electoral competition, we fully characterize equilibrium attention, policies and votes and find two salient patterns emerging as we increase voters' attention costs or garble the media technology: first, arousing and attracting voters' attention becomes harder; second, doing so requires the use of extreme and exaggerated policy and issue positions on the candidates' part. We supplement our analysis with historical evidence, and discuss its relevance in the new era featured with greater media choices and distractions, as well as the rise of partisan media and fake news. 

Social Media, Fake News and Political Accountability (with Davin Raiha)

Social media has proliferated throughout the developed world. Blogs, tweets, and posts have become important sources of political news and information for many. But what is the impact of the emergence of social media on politics and political behavior? Does it improve or compromise political accountability? This paper explores how social media affects political accountability in ways that are different from conventional media. Compared to conventional media, social media enables dissenting viewpoints, harsh criticisms, and even fake news to emerge, creating doubts about politicians that can lead to detrimental behaviors on the parts of citizens and politicians. We show how doubts can reduce politicians' incentives exert policy efforts, cause distraction and encourage cosmetic behaviors. We show how it can reduce the ability of citizens to hold politicians accountable, and adversely affect the competence of the candidate pool. 

Optimal Incentive Contract with Endogenous Monitoring Technology (with Ming Yang)

Recent technology advances give firms more flexibility to utilize employee performance data at a reduced and yet significant cost. This paper develops a theory of optimal incentive contracting where the monitoring technology that governs the above described procedure is a subject of the contract designer's strategic choice. In otherwise standard agency models with moral hazard, we allow the principal to partition the agent's raw performance data into any finite categories and to pay for the quantity of information that the output signal carries. Through analysis of the trade-off between giving incentives to the agent and saving the cost of data processing and analysis, we obtain characterizations of the optimal incentive contract such as information aggregation, strict MLRP, likelihood ratio-convex performance classification, group evaluation as a best response to high monitoring cost, and dividing resources across the assessments of various tasks according to the agent's endogenous tendencies to shirk. We examine the implications of these results for workforce management and firms' internal organizations. 

Intermediated Implementation (with Yiqing Xing)

Many real-world problems such as sales, taxation and health care regulation involve a principal, one or more intermediaries, and agents with hidden characteristics. In these problems, intermediaries can specify the full menu of the multi-faceted bundles that they offer to agents, whereas the principal can only regulate some but not all aspects of the sold bundles, due to legal, information and administrative barriers. We examine how the principal can implement in these situations any social choice rule that is incentive compatible, individually rational and feasible among agents. We show that when intermediaries have private values and are perfectly competitive, the principal's goal can be achieved by a per-unit fee schedule that allows intermediaries to break even under the target social choice rule. When intermediaries have interdependent values or market power, per-unit fee schedules cannot generally be used to achieve the principal's goal, whereas regulating the distribution of limited aspects of sold bundles can. We apply these results to the regulation of real-world intermediaries. 

Efficiency in Dynamic Agency Models

We examine a dynamic agency model where the agent's hidden action can affect current and future signals. We show that when players interact for a large number of instances, asymptotic efficiency can be attained if the monitoring technologies satisfies two general conditions called measure concentration and informativeness, and if the agent can be penalized by reductions in the instantaneous consumption or the future payoff. We use this result to establish a Folk Theorem for discrete-time agency models with high discount factors, and to identify signal processes that yield asymptotic efficiency in discretized continuous-time models. 

Robust Incentive Contract with Disagreement

We examine a dynamic agency model where the agent can disagree over the performance evaluations assigned by the principal and create frictions if the actual incentive pay falls short of what he thinks he deserves. The main result is a characterization of the max-min incentive contract, which gives the principal the best profit guarantee against all disagreement processes. When the horizon is long, the max-min contract is near-efficient and resembles the efficiency wage contract. A key step in the proof is the construction of a simple test contract, which satisfies the above described properties even if disagreements can depend arbitrarily on past efforts and performance measures or exhibit persistence over time.

A Folk Theorem with Virtually Enforceable Actions

This paper proves a Folk Theorem for infinitely repeated private monitoring games with virtually enforceable actions. In these monitoring situations with scarce signals, players depart from the efficient outcome occasionally to acquire the information that detects the profitable deviations of their opponents. In a finite horizon setting with monetary transfers and public communication, I devise a novel Budget Mechanism with Cross-Checking (BMCC) which---through linking the players' action choices over time---virtually implements the efficient outcome at a vanishing incentive cost as the horizon grows and the players become patient. As the building block of my equilibrium construction for the infinitely repeated game, BMCC outperforms public-strategy mechanisms in scarce signal environments and carries important policy applications for labor contract design with costly subjective performance evaluation.

Work In Progress

Differentiated Experiments

Robust Voting Mechanism (with Songzi Du)