On 9 July 2021, civil unrest began in the South African province of Kwa-Zulu Natal. The protest action was sparked by the arrest of former president Jacob Zuma for contempt of court.[1] The resulting protests triggered wider looting and rioting, further fuelled by historical economic inequality and job losses exacerbated by the third pandemic-induced lockdown. By 11 July, civil unrest had spread to the province of Gauteng.[2]
A study by the IMF showed that, while major shocks like a pandemic can set off an economic downturn, the risk of civil disorder is much higher in countries that already face high levels of poverty, inequality, and low economic growth[3].
South Africa was already experiencing an economic recession when the pandemic struck.[4] In the years leading up to the pandemic, the country was also labelled one of the world’s most income-inequal economies, with a Gini coefficient of 0.65 in 2015[5]. Greater inequality coupled with low growth in the midst of a pandemic supported the manifestation of such violent social unrest. However, as a local reported correctly points out:
“… in the current discourse on the uprising of the impoverished in South Africa, consideration of the inequality deepening the impact of the pandemic is largely absent.” [6]
Based on our own initial analysis, we found a high degree of correlation between income inequality and the number of cases of unrest per KZN municipality. This is evidenced by the mapping below:
Figure 1: Choropleth map of KZN by Gini coefficient, overlayed with riot incidence
Source: Own analysis of data from (Quantec, 2021) and (PolicyLab, 2021)
As some respondents to our original post [7]rightly pointed out, there were likely other factors at play that may have influenced the number of cases of civil unrest in the province. This point is confirmed in the literature – looking beyond income inequality, there are several other factors that contribute to civil unrest. These other indicators are summarised as follows:
Figure 2: Contributing factors to the occurrence of civil unrest
Source:(Arde & Wicks, 2021); (Garcia-Murillo, Almeida, & Zaber, 2017); (Gracia-Murillo & Almeida, 2017); (Shelton, 2016); (Marks, 2016); (Wilson Center, 2007)
Based on this literature, we have identified the following factors that likely contributed to the incidence of unrest in KwaZulu-Natal (KZN) during July of 2021:
1. Income inequality,
2. The concentration of non-residential buildings (shops, warehouses, and factories),
3. Population density,
4. The degree of political interest[8],
5. The degree of access to mobile devices or landlines, and
6. The degree of indigence within each municipality
The team used regression analysis to distil the relationship between the prevalence of civil unrest and these factors. Regression analysis is a statistical technique used to assess the degree to which each of the contributing factors impacted the incidence of civil unrest within each municipality of KZN.
The regression equation the team estimated is as follows, using data across all 44 municipalities in KZN:
Where:
- U is the total number of cases of unrest experienced in each municipality,
- GC is the Gini coefficient
- NRB is the number of non-residential buildings in each municipality,
- PD is the population density for each municipality,
- PI is the degree of political interest in the unrest,
- TD is the proportion of households with access to telephonic devices within each municipality, and
- IH is the proportion of indigent households within each municipality.
The regression technique of preference was a negative binomial regression, which was most appropriate given the data structure.[9] Specifically, a negative binomial regression model works well when the dependant variable is a count of some occurrence.
This negative binomial regression provided the following results:
Figure 3: Summary of negative binomial regression results
Source: Own analysis of data from (Quantec, 2021) and (PolicyLab, 2021)
Note 1: These are results are from a cross-sectional negative binomial regression. Each cross-section represents one municipality in KZN.
Note 2: Regression coefficients and confidence intervals were exponentiated and standardised. Standardisation is necessary because the variables of interest are measured using different units. For instance, a Gini coefficient is on a scale between 0 and 1, while the number of indigent households is simply a number. Thus, standardisation is critical to speak to which variable of interest was the largest contributor to the unrest, keeping other things constant.
Note 3: Grey dots indicate a statistically insignificant relationship. Statistical significance simply means that there is a high likelihood that a result from data generated by testing is not likely to occur randomly. Irrespective of statistical significance, it is possible that statistically insignificant variables still played a role in increasing the number of unrest cases.
As expected, a key contributing factor to the number of unrests was income inequality. A standard deviation increase in the Gini coefficient is associated with 1.3 times more unrest cases. In short, a higher degree of inequality is associated with more instances of looting and riots at the municipal level.
Other indicators also contributed to the increased occurrence of unrest in KZN, namely:
- An increase in the number of non-residential buildings in a municipality. The higher the number of businesses (and potential looting sites) in the area – the more likely the municipality was to be targeted (a standard deviation increase in the number of non-residential buildings was linked to 1.7 times more cases of civil unrest in a municipality).
- An increase in the population density within a municipality. The more densely packed a population is, the easier it is to mobilise more individuals towards a specific cause.
- The degree of political interest that individuals showed in the cause. Former president Jacob Zuma still has strong ties with the African National Congress. Therefore, if more individuals voted for the ANC in 2016, the likelihood that some of those individuals would show interest in the imprisonment of former president Jacob Zuma would increase. Because the unrest was attributed, in part, to the former president’s imprisonment in July of 2021, an increase in his supporter base would see increased mobilisation for these individuals’ political cause (i.e., disagreement with the imprisonment).
As far as the team can tell, this is the first attempt to quantify the drivers behind the unrest in KZN during July 2021. While some of the variables we have used are imperfect proxies and some modelling issues still exist[10], our evidence is at least relatively robust. We have shown that, beyond obvious factors like population density, political interest and a concentration of non-residential buildings, something deeper was at play in inciting civil unrest.
Income inequality, which is often glossed over as a critical factor for civil unrest in South Africa, is a key factor that increased the prevalence of riots in KZN. Relative disenfranchisement, coupled with the effects of the COVID-19 pandemic and high unemployment rates in the country, are drivers of emotional response. These emotional responses breed civil and political unrest, and often increase the proclivity for individuals to turn to crime to meet their needs. To ensure that such cases of civil unrest are avoided in the future, targeted policy should aim to move the country away from such a high degree of income inequality.
A higher degree of social unrest, in turn, may result in lowering output and increasing inequality, pointing to a vicious cycle. Without proactive policy measures, this cycle is expected to continue. In such circumstances, social safety measures can play an important part in providing a minimum level of income security and reducing social tensions[11]. In South Africa, the social welfare system is not a panacea for resolving the country’s deep structural and economic challenges. It can however play a critical role in bridging the gap between those that have and those that cannot afford to have.
[1] (CNBC, 2021; IFRC, 2021)
[2] (Steinhauser & Patel, 2021)
[3] (Sedik & Xu, 2020); (Shelton, 2016)
[4] (Stoddard, 2021)
[5] The closer the Gini coefficient, a measure of income inequality, is to 1, the greater the degree of income inequality in a country (Galal, 2021).
[6] (Cottle, 2021)
[7] See the original LinkedIn post here: https://www.linkedin.com/feed/update/urn:li:activity:6825354782807482368/?commentUrn=urn%3Ali%3Acomment%3A(activity%3A6823536841220747264%2C6825354671725535232)
[8]This was proxied for by using the proportion of individuals in each municipality that voted for the ANC in the 2016 municipal election.
[9] Ordinary Least Squares (OLS) overstates the significance of variables if the errors are non-normally distributed or heteroskedastic. Because of this, OLS regression is not appropriate. Similarly, Poisson regressions on count data would also be inefficient because the variance of the Y variable (or the number of unrests) was far greater than the average number of unrests.
[10] There are inter-relationships between some variables of interest, or “multicollinearity”. For example, the number of indigent households is related to the Gini coefficient within a municipality. In most standard practice, this problem is either ignored, or robustness checks are done. In our case, the team conducted robustness checks, removing collinear variables (the number of indigent households and in a separate case, the number of non-residential buildings). The results in these analyses remained largely the same, highlighting income inequality as an important contributing factor to the incidence of unrest in KZN.
[11] Petrovski, Alharbi, Alhomairi, & Morozov, 2020
Bibliography
Arde, G., & Wicks, J. (2021, July 13). Zuma Unrest: Chaos reigns in Durban as looters target warehouses. Retrieved from News24: (Shelton, 2016);
CNBC. (2021). South Africa unrest and Covid measures are starting to derail its economic recovery. Retrieved from CNBC: https://www.cnbc.com/2021/08/02/south-africa-unrest-and-covid-measures-are-starting-to-derail-its-economic-recovery.html
Cottle, E. (2021, August 2). Potent brew that led to South Africa’s urban uprising: Pandemic, recession and a crisis of social reproduction. Retrieved from The Daily Maverick: https://www.dailymaverick.co.za/opinionista/2021-08-02-potent-brew-that-led-to-south-africas-urban-uprising-pandemic-recession-and-a-crisis-of-social-reproduction/
Galal, S. (2021). Gini coefficient in South Africa 2006-2015, by area. Retrieved from Statista: https://www.statista.com/statistics/1127890/gini-coefficient-in-south-africa-by-area/
Garcia-Murillo, M., Almeida, M., & Zaber, M. (2017). Information and communication technologies as drivers of social unrest. Retrieved from The Research Gate: https://www.ippapublicpolicy.org/file/paper/1433972107.pdf
Gracia-Murillo, M., & Almeida, M. W. (2017). Information and Communication Technologies as Drivers of Social Unrest. Research Gate.
IFRC. (2021). South Africa: Urban violence – Emergency Plan of Action (EPoA) DREF Operation n° MDRZA010. Retrieved from Relief Web: https://reliefweb.int/report/south-africa/south-africa-urban-violence-emergency-plan-action-epoa-dref-operation-n-mdrza010
Marks, Z. (2016). Poverty in Conflict. Edinburgh: GSDRC Applied Knowledge Services.
Petrovski, S., Alharbi, W., Alhomairi, A., & Morozov, A. (2020). Modelling Population Dynamics of Social Protests in. E Mathematics.
PolicyLab. (2021, July 30). PolicyLab. Retrieved from Application: https://www.policylab.app/
Quantec. (2021, April 6). EasyData Regional Service. Retrieved from EasyData: https://www.quantec.co.za/easydata/regional-subscription/whats-included/
Sedik, T. S., & Xu, R. (2020). IMF Working Paper: A Vicious Cycle: How Pandemics Lead to Economic Despair and Social Unrest. IMF.
Shelton, S. (2016). Effects of Poverty on Society. Retrieved from Brogen Project: https://borgenproject.org/how-poverty-effects-society-children-and-violence/
Steinhauser , G., & Patel, A. D. (2021). South Africa’s Looting, Violence Reflect Inequalities Exacerbated by Covid-19 Pandemic. Retrieved from The Wall Street Journal: https://www.wsj.com/articles/south-africas-looting-violence-reflect-inequalities-exacerbated-by-covid-19-pandemic-11626189253
Stoddard, E. (2021). IMF study suggests civil unrest will have a significant economic impact on South Africa. Retrieved from Business Maverick: https://www.dailymaverick.co.za/article/2021-07-20-imf-study-suggests-civil-unrest-will-have-a-significant-economic-impact-on-south-africa/
Wilson Center. (2007). Demography and Conflict: How Population Pressure and Youth Bulges Affect the Risk of Civil War. Retrieved from Wilson Center: https://www.wilsoncenter.org/event/demography-and-conflict-how-population-pressure-and-youth-bulges-affect-the-risk-civil-war
Zurich. (2015). The risks of rapid urbanization in developing countries. Retrieved from Zurich: https://www.zurich.com/en/knowledge/topics/global-risks/the-risks-of-rapid-urbanization-in-developing-countries