Explaining Costs of Governing

Abstract

What accounts for losses in government support over its time in office? It is well-known that the early part of a period in government is characterised by a peak in support, and that this support gradually erodes over time. This chapter offers a direct test of a range of mechanisms argued to underpin trends in governing costs. Using a newly available measure of subjective performance evaluations in five countries, we argue that governments lose support over time because electorates judge incumbent parties' performance in predictable ways. Crucially, these predictable ways are unrelated to exogenous performance indicators of incumbents, such that we point to the importance of Stimson's (1976: 1) observation, that, "the extraordinary fit of parabolic curves to actual presidential approval leads to the suspicion that presidential approval may be almost wholly independent of the President's behavior in office". We reveal that a) honeymoon periods arise because electorates base their evaluations of the new incumbent on the basis of their evaluations of the old one. This mechanism weakens as voters acquire new information about the current incumbent, and discount information about the previous one. This mechanism differs from the 'weak partisans' explanation, or the 'expectations and disillusionment' explanation but provides an empirical explanation for the peak in government support at the beginning of a period of government (which can also help account for the rise in expectations at the beginning of a cycle, and the dwindling of weaker partisan support). We further show that b) performance information accumulates; such that a performance shock at the beginning of a cycle, after the initial honeymoon period, matters more than a performance shock at the end of the cycle. Once voters have made up their mind about an incumbent, they are increasingly unlikely to revise their opinion about the governing party. This means that positive information, if it exists, has little effect on incumbent popularity once governing costs have set in. Finally, we show that c) these trends in perceived incumbent performance explain losses in governing electoral support, and this is more important than the effects of measures designed to capture 'coalitions of minorities'.