Previous research has established that, on average, government parties lose support. While this has always been the case, the ‘costs of ruling’ seem to have increased over the past decades. Moreover, the ‘costs of ruling’ depend on economic conditions. When the economy is doing well, government parties tend to lose less, or could even gain, whereas government loss is higher when the economy is doing poorly. However, even though the ‘cost of ruling’ thesis is well established in the literature, there is much less certainty about the role of the economy. Scholars agree that the state of the economy plays an important role in American and British elections. Yet, in multiparty systems (as they exist in the large majority of democracies), the relationship between election outcomes and the state of the economy is much less clear. Moreover, recent studies based on individual level survey data have shown that there is substantial variation between different parties in the extent to which they are held accountable for economic conditions (e.g., Duch and Stevensen 2007; Van der Brug, Van der Eijk and Franklin 2007).
Much of the research on economic voting is conducted at the aggregate level, measuring the loss of governing parties. The largest data set, investigating government loss and economic voting at the aggregate level contains the results of 282 elections in 19 democracies (Paldam and Nannestad 2002). In this paper I will present results at the level of 2292 different political parties, participating in 1126 elections in 77 democratic countries (according to Polity IV). This newly collected dataset contains election results from democracies in all continents, over two centuries (1810 until 2012). These data allow me to focus in more detail on the institutional and party level variables that determine whether governments are blamed or awarded for the state of the economy. Holding government parties accountable is an important aspect of electoral competition, so that the conclusions of my study speak readily to the topic of the workshop.