The Political Violence Cycle∗

Do elections cause violence? Some scholars point to higher levels of unrest around elections, especially in developing countries, to assert that elections cause political violence. We develop a formal model to demonstrate that this conclusion is unfounded. In our baseline model, violence is more effective in electoral periods, leading to a spike in fighting surrounding elections. However, the presence of elections also decreases violence in other periods, as the prospect of future electionsmakes fighting less valuable today. So, while we observe a “political violence cycle” that peaks at the election, the overall effect of introducing elections on aggregate levels of violence may be positive or negative. When elections also provide a sufficiently effective non-violent means to contest for power, they unambiguously reduce violence compared to the non-electoral benchmark while still producing a qualitatively similar cycle. Thus observing a spike in violence does not imply a country would be more peaceful in the absence of elections; in fact, it is a symptom of the fact that elections are consequential.