A Multiprincipal, Multitask Model of Interest Group Competition: An Application to Land Reform Politics in Brazil

Abstract. This paper adapts the model developed by Dixit (1996) to analyze the case of interest group competition over land reform politics in Brazil. The model presents some important advances over earlier seminal economic models of interest group competition in the literature, such as Peltzman (1976), Becker (1983) and Denzau and Munger (1986) in that it considers that government agencies respond to multiple principals each one of which has an interest over a different task performed by the agency, and all of these relationships suffer from information asymmetries. The upshot of this situation is that not only does the asymmetric information push the final outcome away from the first-best and towards a second-best, the standard result in single principal, single task models, but in addition the multprincipal, multitask nature of the relationships lead to a third-best situation. Such third-best outcomes imply that we should expect to see low-powered incentives, a fact commonly observed in the actual organization of government. In the model the government and its land reform agency are the agent, while the principals are the landless peasants (MST) and the general group of voters. Each principal is interested in a specific task performed by the government and effort put towards one of the tasks implies less effort available for the other tasks. The principals do not observe effort but they do observe the outcome in each of the two task dimensions. The model is first solved under complete information, to establish the first-best solution, and subsequently it is solved assuming cooperation amongst principals, to establish the second-best solution. It is then solved with non-united principals and the third-best solution is characterized. This solution is used to portray the case of land reform politics in Brazil by allowing the landless peasants to possess a higher level of information than the voters. By casting the land reform debate in terms of the number of settlement projects established by the government, rather than a measure of actual emancipated and self-sustaining families achieved, the betterorganized groups are able to influence policy towards their own preferences.