This article supports two theoretical changes to models of comparative economic voting. The first is that the distinction between expected and unexpected components of inflation and economic growth is important. We posit that voters are primarily concerned with unexpected inflation and unexpected growth since these changes have real income effects and serve as better indicators of government competence. Empirical analyses of data from nineteen industrialized nations in 1970–94 reveal stronger electoral effects for the unexpected components of inflation and growth than for their overall levels. The second innovation is the relaxation of the assumption of homoscedasticity, which led to the finding that the relationship between economic factors and incumbent vote has become more volatile over time and is less volatile when policy-making responsibility is more obscured.