Buying Elections? An Analysis of Clientelism in South African Local Municipal Elections

Abstract

Clientelism in South Africa’s political system is widely recognized in the existing literature but understudied at the municipal level, particularly regarding its pervasiveness and efficacy for winning votes. This leaves a noticeable gap in the literature, especially since municipal elections determine who will be in control of municipal-level government funds, which in turn affect the quality of goods and services made available for South Africans. This paper seeks to fill these gaps in the literature by asking whether clientelism by the incumbent party in the year prior to an election - measured by proxy through three categories of local municipal spending - impacts how the South African electorate votes for the incumbent party. Additionally, I ask whether the efficacy of clientelism at the local municipal level for winning votes varies for different incumbent parties. In order to investigate these questions, I used local municipal election data, local municipal finance data, and estimates of the average constituent’s poverty level to conduct two sets of multiple linear regression analyses. The local municipal finance data, namely annual measures of the money spent on Contracted Services, Grants and Subsidies, and Other Expenditures, was selected as a proxy measurement for clientelism due to its potential for clientelistic exchanges between politicians and voters. The two sets of regression analyses were conducted with the goal of testing the relationships between (1) clientelism in each of the three municipal financial categories and percent of the vote for the incumbent party; and (2) which party was the incumbent and how much of the total vote share was gained or lost for every additional rand spent in each financial category. I controlled for inflation and population in the data, and accounted for differences in municipalities’ constituent wealth by including a proxy measure in the regression analyses. For the first set of regressions, I found that only municipal spending on Contracted Services negatively and significantly predicted the percent of the vote won by the incumbent party. For the second set of regressions, I once again found that only Contracted Services could significantly explain vote share based on the incumbent party and spending. However, whether or not that relationship was positive or negative varied by incumbent party. The resulting linear model predicted that Contracted Services spending by the ANC would decrease their vote share, while identical spending by the DA would actually increase their vote share. Taken together, the results of both sets of regressions suggest that Contracted Services spending in local municipalities contains some level of clientelism, and that it works better at winning votes for the DA than the ANC. The findings of this study begin to fill the gap in the literature by pointing to what categories statistically significant levels of clientelism occur in, how well clientelism works to win votes, and which parties it works for. These results are of particular importance in South Africa’s ongoing fight against corruption and mishandling of public funds. The more that one knows about the problem, the better they can combat it. Lastly, this study hopefully provides a foundation for future research on the nature and efficacy of clientelism in South African municipalities. Based on my findings, future research could further investigate Contracted Services as a clientelistic avenue, as well as other, similar financial categories. The steps taken in this study also provide a possible format to replicate for metropolitan and district municipal spending in order to further explore clientelism at the municipal level.