Optimal Electoral Timing: Exercise Wisely and You May Live Longer

In many democratic countries, the timing of elections is flexible. We explore this potentially valuable option using insights from option pricing in finance. The paper offers three main contributions on this problem. First, we derive a rationally-based mean-reverting political support process for the parties, assuming that politically heterogeneous voters continuously learn over time about evolving party fortunes. We solve for the long-run density for this process and derive the polling process from it by adding polling noise. Second, we explore optimal timing using the political support process. The incumbent sees its poll support, and must call an election within five years of the last election to maximize its expected total time in office. This resembles the optimal exercise rule for an American financial option. This option is recursive, and the waiting and stopping values subtly interact. We prove the existence of the optimal exercise rule in this setting, and show that the expected longevity is a convex-thenconcave function of the political support. Our model is tractable enough that we can analytically derive how the exercise rule responds to parametric shifts. We calibrate our model to the Labour-Tory rivalry in the U.K., with polling data from 1943-2005 and the 16 elections after 1945. Excluding three elections essentially forced by weak governments, our maximizing story quite well explains when the elections were called, and beats simple linear regressions. We also measure the value of election options, finding that over the long run they should more than double the expected time in power of a fixed term electoral cycle.