Spatial Econometric Approach to Modeling Election Results in Russia: Municipal Level

Abstract

In this article we assess the role of mutual influence of voters living in neighboring territories and the influence of socio-economic factors on the example of voting results for the main candidate in the 2018 elections in Russia. We claim that spatial factors (neighboring of municipalities, regions and belonging of municipalities to the same region) significantly affect the results of voting for the main candidate in each municipality. To confirm this hypothesis, we evaluated several different specifications of the Durbin model, which include dummy variables for the region and other spatial factors, and compared the results with the specifications of the model without taking into account spatial factors. We confirmed main hypothesis: the results of voting depend on the region in which the municipality is included, and, in addition, there is a positive spatial autocorrelation (the results of voting in neighboring municipalities depend on each other). The absence of consideration of spatial factors reduces the quality of regression fitting, there coefficient estimates are biased, and the qualitative picture of the results obtained is distorted. We also showed that the economic situation of the region also affects the results of the voting: economically stronger the municipality received higher share of votes for the main candidate.