B.J.Pol.S., Page 1 of 31 Copyright r Cambridge University Press, 2010 doi:10.1017/S0007123410000323 Context and Economic Expectations: When Do Voters Get It Right? RAYMOND M. DUCH AND RANDOLPH T. STEVENSON* This article discusses the accuracy and sources of economic assessments in three ways. First, following the rational expectations literature in economics, a large sample of countries over a long time period permits tests of the unbiasedness implication of the rational expectations hypotheses (REH), revealing much variation in the accuracy of expectations and the nature of the biases in expectations. Secondly, a theory of expectation formation encompassing the unbiasedness prediction of the REH and setting out the conditions under which economic expectations should be too optimistic or too pessimistic is elucidated. Zaller’s theory of political attitude formation allows the identification of variables conditioning the accuracy of expectations across contexts, drawing a link between the thinking of political scientists and economists about expectation formation. Finally, the theoretical argument that political context impacts the accuracy of average expectations is tested. A great deal of work in both political science and economics has asked the question of whether subjective economic assessments (i.e., retrospective perceptions and/or expectations) are accurate – at either the individual level or when averaged over a representative sample of individuals. 1 In economics, interest in this question is animated by the desire to test the assumptions of rational expectations models that assume that the individual’s economic assessments are as accurate as they can be. In political science, interest in the question stems from the ubiquitous use of, and apparent empirical power of, economic assessments in individual models of the vote choice. In this article, we hope to advance the general discussion of the accuracy and sources of economic assessments in three ways. First, following the large literature testing rational expectations in economics, we use a new dataset that allows us to test the unbiasedness implication of the rational expectations hypotheses (REH) in a larger sample of countries (over a larger time period) than has previously been possible. This exercise reveals a great deal of country-by-country and over-time variation in both the accuracy of expectations and the nature of the biases in expectations (i.e., whether they are pessimistic or optimistic). In addition, in only three of our ten countries do we fail to formally reject the unbiasedness condition of the REH. Secondly, given this variability in the accuracy of expectations across * Nuffield College, Oxford; and Department of Political Science, Rice University, respectively. Raymond Duch is the corresponding author (email: raymond.duch@nuffield.ox.ac.uk). The authors wish to acknowledge the generous support of the National Science Foundation that funded this project with grant SBR-0215633. Earlier versions of this article have been presented at Universite´ de Montre´al, Indiana University, and the Annual Meeting of the American Political Science Association, San Francisco, 2001, where we received extremely helpful suggestions. The authors are particularly appreciative of the help they have received from James Alt and Jim Granato. 1 We use the term economic ‘perceptions’ exclusively to refer to subjective perceptions of the past or current economy and economic ‘expectations’ to refer to expectations of the future economy. When we mean to refer to both perceptions and expectations, we use the term ‘assessments’ or ‘evaluations’.