Voting and protest tendencies associated with changes in service delivery

ABSTRACT

Existing literature highlights the complexities of the relationship between voting, protest action and service delivery within Southern African democracies. We explore the relationship between these variables for South Africa by generating a new dataset from various sources. Although this relationship has been explored before, the existing evidence is mixed. The results from this study therefore bring a fresh perspective, especially at a time during which the political power of the incumbent African National Congress has been declining. We find that improvements in service delivery are not significantly associated with an increase in votes for the incumbent party, indicating that accountability is not the main motivation for South Africans’ voting decisions. However, we find a statistically significant negative correlation between protest action and service delivery, providing evidence of the increasing importance of protest action as a form of political accountability in South Africa.

KEYWORDS:

• Accountability‌
• elections
• political participation
• protest action
• public service delivery

JEL CODES:

• D72
• H11
• H41
• H50

Acknowledgements

The authors wish to thank Derek Yu, Servaas van der Berg, and participants at the Research on Socio-Economic Policy (ReSEP) seminar series for valuable inputs. Fransman acknowledges funding support from the National Research Foundation (NRF) and Research on Socio-Economic Policy (ReSEP).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 In South Africa, more than 14 protests occurred daily, on average, over the period 1997 to 2013. (Bekker, 2022:232). There is also evidence to suggest that this number increased since 2013 (see: Runciman et al., 2016; Powell et al., 2015)

2 The inert and predictable nature of voting outcomes, accompanied by electoral imbalances in the party system have raised concerns that elections fail to act as an accountability mechanism, which evokes deeper concern about the quality of democracy (Schulz-Herzenberg, 2009:2).

3 The ANC has not always been the dominant party incumbent in all municipalities for local elections.

4 Poisson regressions provide a standard framework in the analysis of count data. In reality however, it is often found that count data has a higher incidence of zero counts than is expected for the Poisson distribution.