Research

Abstract: The exorbitant prices of prescription drugs have been a prominent challenge for policymakers everywhere, but more so in the United States. In a bid to combat this, a new law now allows the US government to directly negotiate with pharmaceutical firms to determine prices for at least the more expensive drugs. We take a theoretical approach to study the three-stage bargaining mechanism delineated in the government guidelines. A novel feature of the current mechanism is that the coinsurance for these drugs will be determined after a price has been accepted. Assuming an asymmetry of information about the government's budget allocation that determines the coinsurance for different drugs, we analyze all the PBEs of the bargaining game. In equilibrium, the firm either never learns the coinsurance from the price offer or only learns if the price is high enough. Even with an asymmetry of information about the budget, the 3-stage bargaining game is equivalent to a take-it-or-leave-it offer. The main insight is that the high-budget type is often worse off in equilibrium; in contrast, the low-budget type is better off when compared with a complete information environment. Only when the government's belief that the budget for the drug is high is above a threshold, is it better to commit to a coinsurance along with the price offer. 

Abstract: How does the presence of fake news affect incentives to acquire legitimate information? I study a model of costly information acquisition where either an honest or a fake sender communicates with a receiver through a platform. The honest sender sends a true but noisy signal, whereas the fake sender sends a false and uninformative signal. The platform can verify the signal's authenticity; however, it faces a tradeoff. Fake news, although harmful for the receiver, makes her more skeptical and increases the honest sender's incentives for acquiring information. The platform commits to a policy that indicates the screening probability and a disclosure rule. The optimal screening policy that maximizes the receiver's welfare often requires tolerating fake news, even when such screening is costless. Moreover, not informing the receiver of the screening outcome is sometimes better than full transparency. These findings suggest that complete moderation and fact-checking of content may inadvertently leave the receiver worse off

Abstract: This paper develops a model of disclosure where an unbiased sender communicates with a receiver through a biased intermediary. Evidence is modeled as verifiable information. The sender privately observes the realized evidence and then decides how much information to disclose to the intermediary. An intermediary is a strategic player who can further decide how much evidence to disclose to the receiver. In the baseline model, information can be hidden but not fabricated. When null messages are allowed and if the prior does not favor the intermediary's preferred action, there exist equilibria where the sender can induce the receiver to take the correct action. However, when the prior favors the intermediary's preferred action, such an equilibrium can never be sustained. When null messages are  allowed, the receiver may choose the sender's preferred action for any prior. Perhaps surprisingly, in both cases the receiver may choose the opposite action, even when the sender and intermediary align on their preferred action.

Abstract: This paper studies a novel contract for experimentation in a principal-agent setting called the risk-sharing contract. A principal contracts with two agents for different tasks, each of which needs to be successful for the project to come together. The contract stipulates that no single agent will be paid unless both tasks achieve the desired breakthrough. I solve for the optimal stopping time and optimal wage payments of the agents. While closed-form solutions cannot be attained, some comparative static results are derived.