Voters are not always able to hold governments responsible for their performance in office, and as a result political accountability and representation are not served in all elections. Existing research looks to variations in electoral institutions to explain citizens' ability to vote retrospectively: when the power of policy making is concentrated into the hands of few political actors (under low-threshold PR, coalition cabinets and systems of multiple levels of governance), it becomes easier for citizens to ascertain who is responsible for policy outcomes and to sanction incumbents (Aguilar and Sanchez-Cuenca 2008, Sanchez-Cuenca 2008). But due to their static nature, these institutional predictors are ill-equipped to explain existing inter-election fluctuations in retrospective voting (reported in Lewis-Beck 1988, Paldam 1991, Rattinger and Kramer 1998).

This paper moves beyond institutional explanations of accountability and focuses instead on a dynamic aspect of electoral systems, namely, the inter-election stability of party organizations. I argue that when the attributes of parties change (name, leadership, structure or electoral participation), citizens are less capable of attributing competences to competing parties. As a result, citizens are less likely to elect parties based on their past performance in office, and political accountability is undermined. I test these expectations with survey data from forty European elections (the Comparative Study of Electoral Systems) and an original data set on party instability.