The Long Shadow of Agrarian Conflict: Agrarian Inequality and Voting in Spain

Jordi Domènech,Ignacio Sánchez-Cuenca

doi:10.1017/s0007123421000387

Jordi Domènech, Ignacio Sánchez-Cuenca

Open Access

https://doi.org/10.1017/s0007123421000387

Copy DOI

Abstract

AbstractThis article studies the persistent effects of past agrarian inequality on contemporary voting preferences. Although Western European countries became industrial (and later post-industrial) economies, the political effects of the agrarian cleavage are still visible in those countries in which the agrarian issue was dominant in the interwar period (the industrial laggards). Looking at the spatial variation in voting patterns in the fifteen elections held in Spain since 1977, we show through mediation analysis that areas with high historical agrarian inequality have higher levels of leftist vote. We examine two transmission channels: one economic (related to backwardness); the other political (related to family transmission of political allegiances). A survey analysis provides evidence in favour of family transmission. A brief exploration of other cases confirms the general argument: a similar effect is found in Italy (an industrial laggard), but not in England (an early industrializer).

Highlights

The historical and geographical persistence of patterns of political behaviour has been well documented
Since the socio-structural element of the agrarian cleavage plays a small role in the contemporary politics of Western countries, we argue that the political dimension of the agrarian cleavage can be analysed as a historical legacy of the original socio-structural element regarding landownership
Conclusions industrialization downplayed the relevance of the socio-structural element of the agrarian cleavage, the political allegiances that were part of the cleavage have survived until the contemporary period in some countries

Summary

Leftist vote

Level Provincial controls Municipal longitude, latitude, altitude Gender, age, education, religion Elections Election year fixed effects Std errors clustered by: Model N R2. Historical agrarian inequality has both an effect on the mediators (voting in the 1936 elections and Civil War violence) and on the intermediate confounders (the contemporary economic conditions of the provinces). The graph specifies the pre-treatment variables (geographical factors and old regime characteristics), the treatment (historical agrarian inequality), the mediator variables in the political channel (the 1936 elections and Civil War repression) and the post-treatment variables that are contemporaneous to the outcome variable (levels of industrialization, education and unemployment at the time of the elections during 1977–2019). Moving to the sequential g-estimate, we have applied the Stata code from Acharya, Blackwell and Sen (2016a) to estimate the controlled direct effects of the treatment variable This procedure de-means the main variable to keep the intermediate confounders or the mediator constant, re-estimates the regression without post-treatment variables and bootstraps the standard errors of the coefficient of interest.

Findings

Leftist vote Sequential g-estimate Channel

Conclusions