Research

“Bayesian persuasion under moral hazard” (Job Market Paper)

In the standard Bayesian persuasion setting, a receiver wants to learn about the state of the world to make a final decision. The sender, who prefers one decision to another, designs an experiment with the intent of manipulating the receiver to choose their own preferred choice with the highest probability. It is usually assumed that the receiver accepts the experiment as designed by the sender.

In this paper, I modify the standard framework to capture the following fact of many real-life scenarios. In my model, the receiver, aware of the intent of manipulation, can make unobservable actions that alter the probability distribution of the experiment and change the inference that each message carries. This allows her to learn about the state differently than how the sender had originally specified. In equilibrium, the sender will only design experiments that the receiver will accept. This fundamentally changes the result of the standard model.

In the model, the unobservable action plays no role in the actual distribution of the state of the world and simply captures the economic intuition that the receiver is not a passive agent in the interaction and will react to the experiment to try to undo the sender's manipulation. I show how many of the properties of the standard solution are lost. I create an example in which the sender has no better alternative than to release all information, even if he would have benefited from persuasion in the standard setting. Finally, I gave some conditions under which the standard Bayesian persuasion solution can be applied even in the presence of moral hazard.

“Signaling with evidence” working paper with Nour Chalhoub.

We study the interaction between a developer and an investor. The developer needs funding to produce a product, which is provided by the investor. Both developer and investor want the product to be completed, but the investor pays the cost of continuing. How much the project advances each period is private information of the developer and the distribution of advancements can change over time.

We made a key assumption to model a feature of certain products that can be partially tested before they are fully developed: if asked for evidence on how advanced the project is, the developer can always show less of what is done, but never more. We show that this assumption allows for a very simple class of equilibria in which the developer only wants to show the minimum advancement required to get the project to be continued each period, saving current work for the future as insurance. Nevertheless, this one-sided restriction is enough for projects to be started and completed with possible probability, even if the project would not get started if the developer could not monitor the partial development before the project is completed or if the investors could lie in any direction.

“Market Structure and Innovation with Private Upgrades”

This paper studies an industrial organization model of vertical discrimination. Consumers value higher technology, which is a characteristic of the market goods that the firm can improve. Available upgrades are private information of each firm. Decisions to invest are made before the information is revealed to their rivals. I find that as long as monopoly profits are higher than oligopoly profits, a potential entrant will enter a monopoly market structure with a higher probability than an oligopoly market structure. This result does not extend to incumbent firms. I build a simple example in which by making small changes to the current level of technology or the distribution of upgrades it is possible to get any order in the probability of upgrading when comparing between monopoly and oligopoly market structure.

“Plan jefes y jefas de Hogar y Programa Familas por la Inclusión Social: impact en la duración del desemplo” with Manuel Benites, Mariela Santoro and Alejo Vidal (Pre Phd work)

This paper uses panel data to measure the effect on subsidies in Argentina for the period of 2003-2010. After the economic crisis in Argentina through the years 2001-2002, poverty levels raised to a historic 53\% for urban regions, and unemployment rates reached as high as 21.5\%. To palliate the effect of the crisis, the unemployed with underaged children or disabled children under the poverty level could apply for a subsidy below the minimum wage.

We use the Cox proportional hazard model to estimate the effect on unemployment spells. While the sign of the effect is as expected, we found that the magnitude is particularly small, changing from 355 days with a plan to 342 without it. A potential explanation is that participants were avoiding taking formal jobs as that would invalidate them to perceive the plan and were taking up jobs in the relatively large informal sector of the Argentinian Economy.