When Does Ranked-Choice Voting Reduce Polarization?

Under what conditions does ranked-choice voting (RCV) reduce polarization? Previous research has provided two competing theories, leading to little consensus on the effect of RCV on polarization. We introduce a theoretical model of candidate competition in group-based elections. We further develop an algorithm to computationally simulate how candidates adjust their spatial positions to maximize their vote shares under RCV and first-past-the-post (FPTP). Applying the model to ethnically divided societies, we find that RCV may reduce ideological polarization among candidates more than FPTP across different electoral contexts and voter distributions. We also show that RCV may yield lower levels of ethnic polarization among voters, especially when candidates seek to maximize up to their third-choice votes. Our work has several implications for RCV adoption and research, highlighting the importance of studying when (in addition to whether) RCV reduces polarization. Our computational model is available via our open-source software neodowns.