Research papers and work in progress
Job Market Paper:
"Standing in Line: Demand for investment opportunities with exogenous priorities" (.pdf)
Abstract: We look at a model in which agents can invest in a project with a limited number of available slots. Agents have incomplete information about the project’s expected payoffs. Based on that, they must decide whether to invest in the risky project or take a safe outside option. Slots are assigned following an exogenous priority order. Low priority agents may face a winner's curse: if they choose to invest and obtain a slot in the project it
must be that agents with higher priority choose not to do so. In equilibrium, only high priority agents choose to invest when their private information indicates they should. Low priority agents take the outside option independently of their private information. This feature of equilibrium is maintained when we look at variations of the model with priorities assigned by lottery or determined by a Bernoulli process. We perform relevant comparative statics and compare equilibrium outcomes of our simultaneous action model with the ones from a social learning model. Our analysis highlights unexplored links between market design features and the performance of such markets. In particular, agents' knowledge of the priority order affects both demand and efficiency. Furthermore, herding behavior occurs even in the absence of social learning.
Other Papers:
"Seeing is Believing: an experiment on strategic thinking." (.pdf)
Abstract:The order and observability of actions in a game determine the informational inferences players can make. Intuition suggests that such inferences require a higher level of sophistication when they concern actions that are not directly observed, like in simultaneous action games, compared to sequential games where a player can observe others' actions before making decisions. This intuition contrasts with the assumption of full sophistication embodied in the Bayes-Nash equilibrium concepts. Informational cascades the winner's curse may depend on, respectively, the ability or inability to make such inferences. We use a novel experimental design in which subjects play, both simultaneously and sequentially, a game in which either of these phenomena can occur. We find that, in accordance to our intuition, some subjects participate in informational cascades in the sequential game and suffer a winner's curse in the simultaneous game. "Level-k'' thinking and "cursed equilibrium'' are theories that have been proposed to explain why an individual may suffer from the winner's curse in common value auctions and other environments. Nevertheless, according to these theories the same individual could not participate in an informational cascade. Therefore, our results contradict the predictions of both classical and behavioral theories.
"Voting rules and the collective demand for information." (in progress)
Abstract: We set up a model of collective decision making where the group can learn about the state of nature before
making a decision. Group members preferences have both a private and a common value component. They have to vote
upon two policy alternatives. Before the vote they can exert a fixed amount of effort that increases the chances that the
group receives a public signal on the state of nature. We fully characterize agents’ behavior. We show how some agents
the public signal may have a negative value and therefore they exert no effort. The size of this set of agents depends on
the voting rule. In such an environment the optimal voting is a super-majority rule. It efectively reduces phenomenon
and therefore improves the quality of decision making.
"Dynamic multi-sender cheap-talk." (in progress)
Abstract: The paper studies a multi-sender dynamic cheap talk game. Senders may be biased in favor of one of the two
available choices. Reputational effects are created endogenously due to the dynamic nature of the game. Both biased and
unbiased senders may misrepresent their private information in the first round in order to build up their reputation for
the second round. We characterize the equilibria of the game. Preliminary results indicate that an increase in the number
of senders enhances the incentives of biased agents to reveal their information.
