This paper reconsiders the division of the literature on electoral competition into models with forward-looking voters and those with backward-looking voters by combining ideas from both strands of the literature. As long as there is no uncertainty about voters' policy preferences and parties can commit in advance to a policy platform but not to a maximal level of rent extraction, voters can limit rents to the same extent as in a purely backward-looking model. At the same time, the policy preferred by the median voter is implemented as in a standard forward-looking model of political competition on an ideological policy dimension. Voters achieve this outcome by following a simple lexicographic voting strategy. They cast their vote in favor of their preferred policy position, but make their vote dependent on the in-cumbent parties' performance in office whenever they are indifferent. When uncertainty about the bliss point of the median voter is introduced into the model, voters have to accept higher rent payments, but they still retain some control over rent extraction.