Too big to fail: large companies influence on electoral results in non-democratic settings

This paper attempts to look into the influence of proximity to large companies on electoral results in local electoral committees during federal parliamentary and presidential elections in Russia. It is proven that companies of certain types are used as a platform for mobilisation of voters through various incentives. According to Frye et al. (2014, 2019) – firms and their employers are extremely important for providing the ruling party with necessary support in Russia. As was discovered (Frye et al. 2014), if the company is a part of the state-supported or state-owned industry, or the one with immobile assets or monopoly on providing population in locality with working places – it is more likely to be a broker in the clientelistic exchange. This work tries going a little further and checks if the presence of large companies in general can be considered a factor influencing the party performance or the electoral results of the president. Some media reported a well-established scheme of brokers being assigned to the large local companies and working with its employees to make sure they will attend the elections and vote accordingly (Kozlov 2018, Kuhmar’ 2020). The data for analysis was collected from the Central Electoral Committee and a company register. This data was then matched by the geographically nearest distance between local electoral committees and large companies calculated in metres using the vantage-point tree approach, resulting in a dataset containing 92,875 observations with 14,300 unique companies. The distribution of observations shows that the density of electoral committees is higher and the coverage is better, while large firms are more likely to be concentrated around economically better performing territories. A logistic regression and linear regression was conducted to study the relationship between the distance to the nearest large company and the electoral results registered at local electoral committees. The results from the logistic regression showed that for federal parliamentary elections, the probability of the local electoral committee to show the lowest share steadily increases as the distance between the voting place and the large company increases. For presidential elections of 2018, the probability of getting into the first group with the lowest share of votes gets up to three times higher with the increase in distance. The linear regression models also showed negative coefficients for distance and number of registered voters when predicting turnout and share of votes in both federal parliamentary and presidential elections.