I am an Assistant Professor in the Department of Economics at the University of Wisconsin–Madison. My research interests are in dynamic games and strategic interactions in networked environments.
For dynamic games with imperfectly observed actions, continuous-time models provide a very tractable model to characterize equilibrium payoffs and behavior. The characterization of the payoff boundary establishes a meaningful analytic relationship between the costs of providing different types of incentives, and it determines the equilibrium behavior that supports extremal payoffs.
In applied work, dynamic games are often studied through Markov-perfect equilibria (MPE), which rule out history-dependent behavior such as tacit collusion or reciprocal effort in long-term partnerships. This paper characterizes the perfect public equilibrium (PPE) payoff correspondence in continuous-time stochastic games with discrete states, allowing us to study equilibrium behavior beyond MPE. In a two-stage patent race, efficient PPE sustain tacit non-compete agreements in which competing labs take turns investing at low intensity and use noisy public signals to monitor compliance, achieving substantially higher joint payoffs than in MPE. [Download]
This paper develops the continuous-time approach for two-player games with Poisson information, as well as games with both Poisson and Brownian information. Poisson information can sustain equilibrium behavior that Brownian information alone cannot support, through the efficient use of mutual punishments. The characterization of equilibrium payoffs reveals how the two sources of information are efficiently traded off. [Download]
We establish a folk theorem for games with Brownian information. Equilibria in such games may involve infinitely many action switches over a finite interval, but our construction shows that nearly efficient equilibria need not rely on them them. This makes the prescribed behavior amenable to approximation by strategies with discrete adjustment, as is natural in economic applications. [Download]
My research on networks studies how shocks and policy interventions propagate through networks. Whether their effects are amplified or dampened depends on both the network structure and the strategic responses of the firms and financial institutions involved.
This paper studies optimal design of financial networks with endogenous default resolution. The welfare-maximizing financial system exhibits a two-tier structure. A fully connected low-exposure layer serves to absorb losses without intervention so that the regulator can more credibly threaten not to bail out the banks and a bail-in is possible. A sparsely connected high-exposure layer serves as an insurance in case of default because highly exposed banks are willing to make large contributions to a bail-in.
Earlier work largely studied how interbank networks propagate shocks in the absence of intervention. We endogenize bailout and voluntary bail-in decisions through the strategic interaction between a regulator and banks. Sparse networks can facilitate private rescues by concentrating creditor's exposure, reducing free riding, and strengthening the credibility of the regulator's no-bailout threat for large shocks. These effects can reverse the welfare ranking of networks: a network that would transmit larger losses without intervention can nevertheless deliver higher welfare when intervention is endogenous. [Download]
We characterize the unique equilibrium among capacity-constrained Cournot competitors selling to multiple regional markets. Because firms producing at capacity must divert supply from other regions to expand sales in one, changes in tariffs and production subsidies affect competition throughout the network. These spillovers can have surprising consequences: reducing tariffs can lower aggregate consumer surplus and the profits of all firms. [Download]
Whether you are applying for a job, launching a start-up, running for office, or playing poker, the outcome in many situations depends not only on your own decisions but also on those made by others. Game theory examines how to model such interactions formally and which outcomes to expect when everybody takes into account the objectives and decisions of others.
The second quarter of the microeconomics core sequence introduces the core concepts of game theory needed to model and solve games with complete and incomplete information, both in static and in dynamic settings. We study the main solution concepts and understand the underlying assumptions they impose on the rationality and knowledge of the players. Applications range from job applications, procrastination, and politics to social encounters and board games.
Climate agreements, research collaborations, and market competition are examples of long-lasting interactions between strategic decision makers. In addition to maximizing their short-term objective, people in long-run relationships have to take into account how their actions will affect the future of their relationship. Repeated and stochastic games provide the fundamental tools to study intertemporal incentives that support cooperation in such relationships.
A game theory package that places emphasis on user-friendly input and coloring each player's name, actions, and payoffs in a dedicated color for ease of distinction. Information about the game entered within one environment is stored after it is parsed and can be re-used in another environment. Currently still in development. Bug reports are very welcome. Documentation
Package for creating problem sets, exams, and question banks with emphasis on ease of use. Single compilation generates three pdfs: the problem set, the solutions, and the grading scheme. Points assigned in the grading scheme are totaled up and displayed on the exam automatically. Problems can be filtered based on labels and/or difficulty. Documentation
This is a minimally intrusive package that changes the default overlay behavior of the tikz commands within the beamer document class from the \only to to the `\onslide` behavior. As a consequence, the bounding box of any `tikzpicture` environment is fixed for the entire frame - hence the portmanteau fikz - which prevents images from "jumping around" from slide to slide. Documentation