Research
Publications
(Journal of Mathematical Economics, Vol. 97, Dec 2021)
We study a model of Bayesian persuasion between a designer and a receiver with one substantial deviation—the designer offers once and for all a single statistical experiment from which the receiver can acquire costly i.i.d. signals over time. Taking a 2-state-2-action environment and employing a tractable continuous-time framework, we fully characterize the optimal persuasion policy. When the receiver features high skepticism, the optimal policy is to immediately reveal the truth, which is true for a large set of primitives. We construct the designer's maximum payoff and find a discontinuous drop in it as compared with the standard model. Unlike in many standard persuasion models, the designer is not able to appropriate all the rents of information disclosure while the receiver often achieves the highest possible benefit from being able to repeatedly sample from the strategically offered information structure.
Working Papers
We study the problem of a government that is deciding how to optimally support a financial firm whose solvency is perceived to be in danger, so that the market is on the edge of withdrawing investment to it, thus jeopardizing its growth rate. Tools available to the government are fiscal interventions (a bailout or ongoing financial injections) and the possibility of information disclosure regarding the firm's health (say, via stress tests). We fully characterize optimal interventions. The optimal course of action is to bail out only firms of intermediate size and thereafter inject funds to ensure their solvency. If information disclosure (modeled as a dynamic persuasion game) is ever chosen, it will be tried before any fiscal intervention and subsequent intervention is triggered only after the stress test reveals the firm’s state. We fully characterize the optimal dynamic information structure under some technical conditions and show that the optimal information structure consists of the realization of a signal that reveals the firm’s health to be poor and that the government immediately bails out the firm upon such a signal realization. We explore the relationship between these two policies and show that they are complements. Moreover, we examine when full and immediate information disclosure is optimal and provide a necessary and sufficient condition for the optimality of such a policy.
We study a model of Bayesian persuasion between a designer and a receiver with one substantial deviation—the receiver can acquire costly i.i.d. signals from the offered information structure. Taking a 2-state-2-action environment as the baseline and using continuous approximation, we fully characterize the optimal persuasion policy. When the receiver features high skepticism, the optimal policy is to immediately reveal the truth, which is true for an unexpectedly large set of primitives. We locate the designer’s maximum payoff, find the setup cost of persuasion the designer incurs, and identify a wedge that measures the value of dictating information acquisition for the designer. Some extensions of the baseline model are discussed.
Recent decades have witnessed a dramatic increase in accessibility to information and, as a result, a decrease in its cost. We investigate the implications of this trend on the incidence of coordination failure when information choices can be made sequentially. To do so, we consider sequential information acquisition in a standard global game model with a normal information structure and improper prior. We show that the likelihood of coordination on welfare-inferior equilibrium is invariant to either the information cost or the maximum number of signals agents can acquire. Such invariance is driven by the fact that agents with optimistic (pessimistic, resp.) beliefs acquire too much (too little, resp.) information from the social perspective. We demonstrate that more accessible information may result in a less efficient outcome and characterize how this inefficiency depends on model parameters.
Work in Progress
We study a global game in which the public information is transmitted via signaling. Like each individual in the economy, the government learns a noisy signal about the fundamental and then implements a publicly observable tax policy. Our attempt is to establish a causal relation between the accuracy of the government's private information and equilibrium-multiplicity. When the government's private signal is perfect (resp. totally uninformative), we have multiple equilibria (resp. a unique equilibrium). Building on this observation, we conjecture that there is a threshold of accuracy that divides the world with a unique equilibrium and that with multiple equilibria, and in equilibrium the policy maker will only separate states at the tails and those intermediate. We prove that a unique BNE exists if and only if the policy maker's signal is sufficiently inaccurate, given our conjecture about the equilibrium, which partially confirms the threshold mechanism behind equilibrium-multiplicity determination.
Research versus Review
I consider an experimental tournament with little information, in which the state can be endogenously evolving. A group of experimenters are facing the same uncertainty in terms of the state of a slot machine, either good or bad. In the good state, the revelation of the true state arrives at a Poisson rate rescaled by (costly) research intensity, but only the first discoverer of the good state will be rewarded. Nobody can directly observe others' behaviors or outcomes, but other people's success arrives at a Poisson rate rescaled by (costly) review intensity. Hence, assuming a constant outside option value, each experimenter faces a problem of allocating effort to each of the two arms, and when dropping out. The problem is featured by non-monotone posterior and changing state. I am seeking in this setting the optimal tandem of pulling two arms, how the nature of changing state affects the optimal choice, and its implication on social learning. It is shown that in a version of the model with fixed changing rate of the state, the optimal strategy exhibits a one-switch property, meaning that one can at most switch once from review to research, and the threshold of switch can be characterized.