Satisfaction and adaptation in voting behavior: an empirical exploration

Dynamic models of learning and adaptation have provided realistic predictions in terms of voting behavior. This study aims at contributing to their scant empirical verification. We develop a learning algorithm based on bounded rationality estimating the pattern of learning process through a two-stage econometric model. The analysis links voting behavior to past choices and economic satisfaction derived from previous period election and state of the economy. This represents a novelty in the literature on voting that assumes given voter preferences. Results show that persistence is positively affected by the combination of income changes and past behavior and by union membership.