In this paper, we propose a method that enables researchers to estimate the treatment effect averaged over an entire population using an instrumental variable (IV) in a setting where a running variable fully determines the treatment. We apply our method to the estimation of the incumbency effect...