In this dissertation I study the interaction between institutions and decision-making process. In Chapter 1 Performance reviews are mechanisms by which an agent is held accountable to a pre-specified target, and are used in many organizations to monitor and evaluate agents. How well do these review mechanisms work? In this paper we study two commonly used classes of such mechanisms: those with duration targets, and those with discretionary limits. We describe a constrained dynamic mechanism design problem for a sequence of agents that are periodically replaced, and who observe some private information in each time period. With duration targets there is a performance target to be achieved over a number of periods. We show that a succession of agents implement a pattern of decisions that is a function of the time remaining till the review: in earlier periods more weight is put on the agents' preferred actions, whereas in later periods the weight is shifted towards meeting the goal of the review. Duration targets perform better than full discretion. For high enough costs of review, they also perform better than reviews with limited discretion, where reviews take place every period. We describe a methodology to determine the minimum length of the review phase given the parameters of the model and the cost of review.

In Chapter 2 my co-author, T. Rene Bowen, and I ask "What prevents majorities from extracting surplus from minorities in legislatures?" We study an infinite horizon game where a legislative body votes to determine distributive policy each period. Proposals accepted by a simple majority are implemented, otherwise the status quo allocation prevails. We construct symmetric Markov perfect equilibria that exhibit compromise in the following sense: if the initial status quo allocation is "not too unequal", then the Markov process is absorbed into allocations in which more than a minimum winning majority receives a positive share of the social surplus. The compromise is only sustainable if, starting from the "unequal" allocations, the Markov process is absorbed into allocations in which there is a complete absence of compromise. These compromise equilibria exist when discounting is neither too small nor too large. We find that, contrary to intuition, the range of discount factors for which these equilibria exist increases as the number of legislators increases. In this sense, compromise is easier in larger legislatures.

In Chapter 3 my co-author, Anders Olofsgård, and I look at the impact of broad policy reforms on the levels of corruption. We use a structural break approach to identify country specific time periods in which significant shifts in corruption levels take place. We then correlate these times of change with a set of co-variates with specific focus on the impact of democratization, and trade and equity market liberalization. We find robust support for the hypothesis that episodes of reduction in corruption levels tend to be correlated with democratization and equity market liberalization.