Voter Behavior and Government Performance : Theory and Empirical Application in Sub-Saharan Africa
It is a commonly accepted view that democracy when compared to other non-democratic systems is a superior political system promoting economic growth and well-being. As a result in the past years many former autocratic regimes became democracies, especially in Africa during the third wave of democratisation. However, controlling for demographic and economic structures a large variance regarding policies, implied growth and well-being can be observed across democracies. The question arises how these differences across economically similar democratic countries can be explained. We assume that there exists an incentive problem that especially governments in developing countries are facing. Depending on the political system, there are different incentive mechanisms. In democratic systems, electoral competition is understood as a fundamental mechanism to guarantee that governmental policies reflect society’s interests. But, although multiparty elections take place in most African countries by now, electoral competition is often restricted and leads to biased policy outcomes as well as poor government performance. Basically, policy biases result from two major mechanisms: government capture and a lack of governmental accountability. Consequently, there exists a relationship between individual voting behavior and government performance. In this thesis, we analyze the relationship between voting behavior and government performance theoretically and empirically. In the first part we extend the already existing voter model of Baron, Grossman and Helpman by generalizing the assumption of uniform distributed voter preferences. Instead we take voter preferences that are extreme value distributed assuming that a local Nash equilibrium exists. Further we derived indices that measure government accountability and government capture. The second part of the thesis concentrates on the empirical application. Voter surveys were conducted in Ghana, Senegal and Uganda. In each country voting behavior was estimated with a multinomial logit model and a latent class model. The latter explicitly includes heterogeneity in the probabilistic voter model, which is necessary for measuring capture. The results show that voting behavior in the three case study countries strongly depends on non-policy issues, but also on retrospective issues and policy factors. Further we derived a relationship between voting motives and our indices measuring government performance. The higher the degree of democracy within a country the more dominant is policy oriented voting, which leads to better incentives and consequently to enhanced government performance.