Selected Publications

Working Papers

We study the ability of multi-group teams to undertake binary projects in a decen- tralized environment. The equilibrium outcomes of our model display familiar features in collaborative settings, including inefficient gradualism, inaction, and contribution cycles, wherein groups alternate taking responsibility for moving the project forward. Expected delay grows more than proportionally with project size, and some welfare- enhancing projects are not completed, even as agents become arbitrarily patient. A team composed of two equally large groups can complete larger projects than a fully homogenous team, even as the difference in preferences for completion among the two groups is arbitrarily small. Moreover, if the project is sufficiently large, the two-group team always completes the project strictly faster.

Democratic representation is constrained by the alternatives available to voters. In this paper, we develop a methodology to gauge the extent to which the “supply side” of politics hinders voter welfare. Using rich data on thousands of candidates in three Brazilian legislative elections, we quantify the relative value voters place on candidates’ policy positions and non-ideological attributes, and we evaluate voters’ welfare given the set of candidates they face. Our estimates uncover substantial welfare losses to voters relative to three alternative benchmarks of ideal representation. On average, the typical voter suffers only a moderate loss due to policy incongruence but a large loss due to shortages in candidates’ non-ideological characteristics. To evaluate the welfare consequences of potential institutional reforms, we develop and estimate a model of equilibrium policy determination. Through counterfactual experiments, we show that institutional reforms aimed at improving the quality of representation may have sizable unintended consequences due to equilibrium policy adjustments.

A fundamental result in the literature on congressional control of the bureau- cracy is that optimal institutions take the form of limited delegation of decision-making authority to the agency. We revisit the question of optimal institutional design for cases in which the extent to which the policy is funded determines its effectiveness. In the optimal separating mechanism, the legislator imposes larger reductions of the budget for policies that go further in the direction of the agent’s preferences (sufficiently high states); even more, in some states, than what the legislator herself would want to implement given full information. Policy over-reacts to the agent’s information in low states, and under-reacts in high states. For sufficiently high states, both policy and budget are below the first best for both the agent and the legislator. Our results imply that the optimal mechanism can not be implemented through delegation of authority to the agent.



We introduce a dynamic model of position-taking and advertising, in which politicians trade off policy and reelection. We estimate career concerns for US senators, and quantify how preferences for office and policy affect electoral accountability. Senators are typically willing to make signicant policy concessions for electoral gains: the median senator is willing to give up 5% of the average distance between voters and politicians for a 1% increase in the probability of reelection. There is, however, substantial heterogeneity, which covaries with observable characteristics. Given estimated voter responsiveness, accountability is low on average, but high for office-motivated senators in competitive elections.



We consider a dynamic process of coalition formation in which a principal bargains sequentially with a group of  agents. This problem is at the core of a variety of applications in economics and politics, including a lobbyist seeking to pass a bill, an entrepreneur setting up a start-up, or a firm seeking the approval of corrupt bureaucrats. We show that when the principal's willingness to pay is high, reallocating bargaining power from the principal to the agents generates delay and reduces agents' welfare. This occurs in spite of the lack of  informational asymmetries or discriminatory offers. When this collective action problem is severe enough, agents prefer to give up considerable bargaining power in favor of the principal.


 We quantify the effect of pre-vote deliberation on the decisions of US appellate courts. We estimate a model of strategic voting with incomplete information in which judges communicate before casting their votes, and then compare the probability of mistakes in the court with deliberation with a counterfactual of no pre-vote communication. The model has multiple equilibria, and judges' preferences and information parameters are only partially identifed. We find that deliberation can be useful when judges tend to disagree ex ante and their private information is relatively imprecise; otherwise, it tends to reduce the effectiveness of the court.


We consider a class of dynamic collective action problems in which either a single principal or two competing principals vie for the support of members of a group. We focus on the dynamic problem that emerges when agents' decisions to support an alternative are irreversible, and agents negotiate and commit their support to the principal sequentially. A danger for the agents in this context is that a principal may be able to poach agents to her side by exploiting competition among members of the group. Would agents benet from introducing competition between opposing principals? We show that when the principals' policies provide value to the agents, competition generally reduces agents' welfare. We study applications to endorsements in elections and corporate takeovers.


We argue that a model of judicial behavior that accounts for dierences in justices' ability and ideology provides a fruitful alternative for the empirical analysis of judicial decision-making around the world, and illustrate this by focusing on the case of the UK.We show that the model explains the decisions of the Lords of Appeal remarkably well, and improves the t of a purely ideological model. We use our estimates to tackle previously unaddressed questions about the relative role of justices' preferences and ability in the Appellate Committee.

We study a model of decentralized legislative bargaining with policy externalities. We establish the emergence of middlemen in legislative bargaining as a robust equilibrium phenomenon. We show that legislative intermediation can impact policy outcomes, and can be inefficient. To fulfill this role, the middleman's policy preferences and bargaining position must be such to make his role of intermediary credible. But the middleman must also directly benefit from policy change. Thus, the political broker is both an intermediary and a client. This result highlights a fundamental difference between intermediaries in politics and exchange economies.

In this paper, we study the conditions under which members of Congress incorporate policy-specic information in their voting decisions. To do this, we estimate an empirical model that accounts for uncertainty and private information about the quality of the proposal. We show that seniority and uncompetitive elections lead to higher ideological rigidity, and curtail the role of information in policy-making. 


 We link the intensity of campaign competition in different electoral systems with the number of candidates running for public office and their ideological differentiation. We show that proportional elections have more candidates, competing less aggressively in campaign spending, than those in majoritarian elections. Candidates’ ideological positions, however, can in general be more differentiated in majoritarian or proportional elections. We also study the equilibrium effects of plurality premiums and the consequences of heterogeneity among candidates in non-ideological characteristics.

In this paper, we address empirically the trade-offs involved in choosing between bureaucrats and politicians. In order to do this, we map institutions of selection and retention of public officials to the type of public officials they induce. We do this by specifying a collective decision-making model, and exploiting its equilibrium information to obtain estimates of the unobservable types. We focus on criminal decisions across US states' Supreme Courts. We find that justices that are shielded from voters' influence (bureaucrats) on average (i) have better information, (ii) are more likely to change their preconceived opinions about a case, and (iii) are more effective (make less mistakes) than their elected counterparts (politicians). We evaluate how performance would change if the courts replaced majority rule with unanimity rule.

We estimate a model of voting in Congress that allows for dispersed information about the quality of proposals in an equilibrium context. In equilibrium, the Senate only approves House bills that receive the support of a supermajority of members of the lower chamber. We estimate this endogenous supermajority rule to be about four-fifths on average across policy areas. Our results indicate that the value of information dispersed among legislators is significant, and that in equilibrium a large fraction of House members’ (40–50%) votes following their private information. Finally, we show that the probability of a type I error in Congress (not passing a good bill) is on average about twice as high as the probability of a type II error (passing a low-quality bill).


We study a model of elections in non-majoritarian systems that captures the link between competition in policies and competition in campaign spending. We argue that the overall competitiveness of the political arena depends on both the endogenous number of parties contesting the election and the endogenous level of campaign spending. These two dimensions are linked together through their combined effect on the total equilibrium level of political rents.We illustrate the key insights of the model with an analysis of the competitive effects of campaign spending limits. We show that under some conditions spending caps can be pro-competitive, leading to an increase in the number of parties contesting the elections.

We estimate an equilibrium model of decision making in the US Supreme Court that takes into account both private information and ideological differences between justices. We measure the value of information in the court by the probability that a justice votes differently from how she would have voted without case-specific information. Our results suggest a sizable value of information: in 44 percent of cases, justices’ initial leanings are changed by their personal assessments of the case. Our results also confirm the increased politicization of the Supreme Court in the last quarter century. Counterfactual simulations provide implications for institutional design.


This paper examines the institutional determinants of discipline in legislative parties. The model formalizes the tradeoff between resources at the leader’s discretion, and the leader’s need to maintain a minimum level of support to continue leading. The value of the leader’s promises of future benefits is here endogenously determined by the backbenchers’ beliefs about the extent of support to the leader among other party legislators. Rewards that can be distributed publicly and on the spot are effective tools to coordinate beliefs about the stability of the leader, and thus also increase the value of the leader’s promises of future benefits. These spot resources are in fact necessary for the leader to be powerful: without them, the leader can use promises of future benefits to sway members’ behavior only if a majority of the party agrees (ex ante) with the leader’s preferred position in the first place.



This paper links the theory of interest groups influence over the legislature with that of congressional control over the judiciary. The resulting framework reconciles the theoretical literature of lobbying with the negative available evidence on the impact of lobbying over legislative outcomes, and sheds light to the determinants of lobbying in separation-of-powers systems. We provide conditions for judicial decisions to be sensitive to legislative lobbying, and find that lobbying falls the more divided the legislature is on the relevant issues. We apply this framework to analyze supreme court labor decisions in Argentina, and find results consistent with the predictions of the theory.


Argentina's constitution and electoral rules promote a fragmented polity. It is in those environments that independent judiciaries develop. Instead, most analysts do not consider the Argentina judiciary as independent. In this article we attempt to explain this contradiction by showing that this perception is inappropriate. We develop a test of the hypothesis that the judiciary is independent by empirically examining the political incentives faced by individual justices in their decision making. Our results show an often-defiant Court subject to constraints. Our measure of defiance is the probability of a nonaligned justice voting against the government. We find that judicial decision making was strategic. The probability of voting against the government falls the stronger the control of the president over the legislature, but increases the less aligned the justice is with the president. Thus, politics and process matter in understanding Argentina's Supreme Court decisions. Institutions matter in Argentina as well.