Mapping relative social vulnerability in six mostly urban municipalities in South Africa

Abstract

Urban decision-makers in South Africa face growing challenges related to rapidly expanding populations and a changing climate. To help target limited resources, municipalities have begun to conduct climate change vulnerability assessments. Many of these assessments take a holistic approach that combines both physical hazard exposure and the underlying socio-economic conditions that predispose populations to harm (i.e., social vulnerability). Given the increasing use of socio-economic conditions in climate change vulnerability analyses, this paper seeks to explore two key research questions: 1) can the spatial distribution of relative social vulnerability be estimated in six mostly urban South African municipalities, and if so, 2) how sensitive are the results to a range of subjective methodological choices often required when implementing this type of analysis. Here, social vulnerability is estimated using socio-economic and demographic data from the 2001 and 2011 South African censuses. In all six municipalities, social vulnerability varies spatially, driven primarily by differences in income, assets, wealth, employment and education, and secondarily by differences in access to services and demographics. Even though social vulnerability is estimated from a wide array of population characteristics, the spatial distribution is surprising similar to that of the percent of working-age individuals making less than 800 rand per month. Areas with high percentages of previously disadvantaged, extended family, and informal households tend to display relatively higher levels of social vulnerability. In fact, demographics (e.g., race, language, age) are often highly correlated with other characteristics that have direct ties to social vulnerability (e.g., income, employment, education). The spatial patterns of relative social vulnerability are similar in 2001 and 2011. However, there is some evidence social vulnerability is relatively lower in 2011. While the choice of input data and aggregation method can affect the spatial distribution of social vulnerability, the general spatial patterns appear to be fairly robust across a number of subjective choices related to methodological and aggregation approach, spatial resolution, and input data.