EXTRA SESSION: War, Polarization, and Partisanship (on Tuesday 12.30)

Title: War, Polarization, and Partisanship (Pauline Grosjean, Saumitra Jha, Michael Vlassopoulos and Yves Zenou)

Abstract: With people increasingly mobile and inter-connected, one might expect that increased social contact would also lead to homogenization and convergence in political attitudes. Instead, partisanship and polarization are on the rise to the point of threatening the political consensus that underlie liberal democratic institutions. Solving this paradox and understanding the dynamics of contact, partisanship, and polarization is challenging because of endogenous sorting, either through residential choice or media echo chambers. This paper uses a valuable historical context where peers were exogenously assigned to demonstrate how political homogenisation at the local level may displace political tensions to new frontiers, where they can degenerate into civil war.

We first develop a network model of polarization and partisanship in which partisan choices are modelled as strategic complements, so that each agent chooses an action as long as at least a simple majority of her neighbours choose the same action. The model predicts that each network will coordinate on one partisan equilibrium and will thus become locally homogeneous; however, in the presence of multiple partisanship coordination equilibria (e.g. Left, Right), overall polarization may increase. 

We then study this dynamic process of homogenisation and polarization empirically, exploiting the large-scale exogenous assignment of individuals to peer groups in the context of World War I in France. Our historical context is particularly well suited to this empirical exercise. First, the circumstances of military recruitment local exogeneity in regimental assignment. Second, legislative elections were held just before (May 1914) and just after (May 1919) the war. The historical context of universal male suffrage combined with universal male military conscription in World War I ensures that our voting data, at the highly granular level of 35,000 French municipalities, captures the preferences of the population exposed to war service.

Our empirical results show that fighting together in WWI made people more politically homogeneous. This convergence is strong and persistent, since municipalities that were initially further away from the average converge faster to the majority of the votes, across the interwar elections of 1919, 1924, 1932, and 1936. Partisanship is determined by variation in the initial partisanship of different military units, with communism a stronger attractive force. Our results are consistent with the theoretical model, since we find strong homogenisation locally and polarization at the national level. This is because the local homogenisation, which is partisan (either to the left or to the right), leads to sharp local discontinuities across military boundaries.

Beyond electoral results, we also illustrate some of the real political costs of polarization. We show how such polarization in interwar France sowed the seeds of the country’s “long civil war” (Jackson, 2003) and predicts either local Nazi collaboration or resistance against Nazi occupation during World War II.