This is an old revision of the document!


Development of indicators of Social Vulnerability

Introduction

Social vulnerability is defined within the context of this work as characteristics or qualities within social systems that create the potential for loss or harm. Since it is difficult to measure social vulnerability in relative terms, one of the most common approaches for performing a social vulnerability assessment is the use of a set of composite proxy indicators. Here, there is a general consensus within the literature about some of the factors that influence the social vulnerability of populations. These include pre-existing characteristics of populations and economies such as age, gender, access to resources, and the distribution of wealth. Other characteristics include access to education, governance and institutional capacities, healthcare resources and access, and elements of the built environment such as the density of building infrastructure.

Methods

Social vulnerability helps to explain why some areas, such as a country’s sub-national parishes or city neighborhoods, will experience the consequences of an earthquake differentially, even though they are subject to similar levels of ground shaking. Understanding the differential impacts of an earthquake as a product of social vulnerability is a critical element to earthquake risk reduction, for the enactment of mitigation plans and the development of sound public policies to reduce earthquake risk. To measure social vulnerability, the starting point is to capture contextual conditions within a study area’s social fabric. The social fabric includes pre-existing socioeconomic characteristics related to the overall capacity of populations to prepare for, respond to, and recover from damaging events (Cutter et al. 2000). Within this context, there is a rich tradition of research focused on those factors that increase or decrease the impact of hazard events on populations. These characteristics include age, gender, access to resources, and the distribution of income (Tierney et al. 2001; Heinz Center 2002; Cutter et al. 2003, National Research Council 2006). Others characteristics include access to education, governance, institutional capacities, healthcare access, and elements of the built environment such as the density of residential infrastructure (Cutter et al. 2003). It is these inherent characteristics of populations that help to redistribute risk before an earthquake event occurs and after an event in terms of the distribution of losses.

One of the most common methods to measure social vulnerability is through the construction and application of composite indicators. Social vulnerability indicators are arguably powerful tools because they summarize complexity and provide quantitative metrics to compare places and track progress (Tate 2012). Similar to how companies benchmark their performance against industry standards, governments and hazards practitioners are finding it useful to evaluate communities in terms of their comparative social vulnerability. The latter is partially to attract public interest in disaster loss reduction, to uncover hotspots of management concern and to provide metrics to set priorities, measure progress, and aid in decision-making processes. The literature on composite indicators is well-defined with many methodological approaches for index construction. Most of the literature highlights the need for a number of specific steps. These steps include: 1) the identification of relevant variables; 2) multivariate analyses; 3) weighting and aggregation; 4) the ability to convolute or link variables; and 5) dissemination or visualization of results.

To accomplish the steps for the development of social vulnerability and integrated risk indices, we utilized an Open source software platform known as the OpenQuake Integrated Risk Modelling Toolkit (IRMT) (Burton and Tormene 2015; Khazai et al. 2014). The IRMT is comprised of a suite of geospatial-modeling tools supporting the meaningful combination of estimates of physical earthquake risk with social vulnerability indicators. The IRMT leverages a Quantum Geographic Information System (QGIS) platform Social vulnerability and integrated risk (SVIR) QGIS plugin and expands the effectiveness of the SARA programme through its ability to allow a number of GEM’s open-source tools to operate seamlessly through an intuitive workflow. The latter includes the OpenQuake-engine OQ-engine that can export the results of physical risk calculations that are compatible with the toolkit and the OpenQuake-platform GEM Platform that can be used to facilitate project sharing and project visualization over the web. Each step in the process is outlined in the subsections below.

Social Vulnerability Index: Identification of relevant variables

The selection, collection, and pre-processing of data is the primary step in composite indicator development. It is within this context that the starting point for measuring social vulnerability in South America was to capture pre-existing social conditions related to the overall capacity of populations to prepare for, respond to, and recover from damaging events. Figure 1 is a generalized workflow for the construction of Integrated Risk Indicators using the OpenQuake suite of tools. This workflow was originally presented in Integrated risk modelling within the global earthquake model (GEM): Test case application for Portugal (Burton et al 2014) and it is inspired by theoretical constructs designed to guide the convolution of assessments of a natural hazard threat, potential economic losses, and social vulnerability (Cutter 1996; Cardona 2005). The workflow entails: 1) the seismic hazard (Figure 1A); 2) the geographical context of a study area that includes its exposure and physical vulnerability (Figure 1B); 3) physical risk estimations (Figure 1C); 4) the pre-existing social fabric of populations (Figure 1D); 5) social vulnerability estimations (Figure 1E); and 6) the integration of physical earthquake risk estimations with social vulnerability metrics (Figure 1F).

Figure 1

To develop the social vulnerability indicators, a total of 430 variables at the sub-national level (P3) for Argentina, Bolivia, Chile, Colombia, Ecuador, Perú, and Venezuela were collected, pre-processed, categorized, and compiled into a database of social vulnerability indicators for each respective country. The data was obtained from the most recently available national censuses on population in each country. Each country database contains approximately 50 to 70 indicators. Variations in the number of indicators and the enumeration units, as well as the data source for each country are outlined in Table 1 below.

Table 1

Country Sub-national division Subdivision count Indicators collected Data Source
Argentina Departamento, Partido, Comuna 527 57 Argentina Instituto Nacional de Estadisticas y Censo (INDEC) - Censo 2010
Bolivia Municipio 341 68 Instituto Nacional de Estadistica (INE) de Bolivia - Censo 2012
Chile Comuna 342 68 Instituto Nacional de Estadistica de Chile (INE) - Censo 2002
Colombia Municipio 1114 60 Colombia Departamento Administrativo Nacional de Estadistica (DANE) - Censo 2005
Ecuador Parroquia 1024 56 Instituto Nacional de Estadística y Censos (INEC) - Censo 2010
Peru Distritos 1833 65 Instituto Nacional de Estadistica e Informatica (INEI) - Censo 2007
Venezuela Parroquia 1130 47 Instituto Nacional de Estadística (INE), Censo 2011

Since it is difficult to measure the social vulnerability of populations relatively, variables were collected as proxy measures to represent the concept. Here, a step was taken to ensure the relevance of the data within the domain of social and economic vulnerability research. A literature review exceeding 400 articles was conducted to ensure the relevance of all data that was collected and compiled into databases. It is within this context that we collected variables within the population, economy, infrastructure, health, and education dimensions by adhering to the taxonomic classification developed in Risk and resiliency indicators, EMI topical report (Khazai et al. 2011) for the selection of socio-economic indicators typically used in social vulnerability assessments. A hierarchical approach (see Figure 2) was utilized in which variables were collected within components, yet classified into their corresponding sub-components (e.g. Population variables were collected and subclassified into corresponding population structure and vulnerable populations sub-components).


Figure 2

Many of the indicators were easily classifiable into the broad groups such as population, economy, health etc. with a large degree of overlap, so this hierarchical system of relevant categories and subcategories was used to classify the social vulnerability data so that these classifications and subclassifications could be explored more in depth throughout the study. The SARA social vulnerability database does not include variables related to the environment or the governance and institutional capacity themes; this is due to the fact that data related to these themes is not broadly available at P3 level of sub national division. Table 2 is an example of some of the SARA variables within the main themes and subthemes.

During the first stage of the project a statistically robust, representative, and comprehensive spatially enabled database was developed for the measurement of social and economic vulnerability at the national level for the South American continent. The primary stage in developing the database was the collection, harmonization, and analysis of relevant socio-economic variables from open and publically available sources as well as the development of a comprehensive taxonomy for social and economic vulnerability that can be used to structure relevant indicators at the different scales of geography, i.e. national and sub-national.kazai et al., 2001

In the second period, data was collected from the most recent census available for a given country and classified according to the indicators taxonomy suggested in Power et al. (2013). Also, an appraisal of national censuses in South America, at various levels of geography, to understand data availability and scalability in order to guide the collection of data was performed. The following table is an example of the themes, subthemes and main variables used in the SARA project. For a complete set of variables per country please refer to the country tabs below.
Table 2

Social Vulnerability Index: Weighting and Aggregation

Central to the construction of composite indicators is the need to meaningfully combine different data dimensions in a manner in which consideration is given to weighting and aggregation procedures. The method of aggregation that we employed is the summation of equally weighted sub-index scores. Here, variable scores in each sub-index (e.g. population, economy, etc.) were averaged to reduce the influence of the varying number of variables in each sub-index. Each sub-component was then summed to derive a final composite score. Since there are between three and five sub-components, depending on the data available for each country, the summed score of the composite indicator ranges between 0 and 5 (0 being the least socially vulnerable and 5 being the most). As a subsequent step, the composite social vulnerability scores were MIN-MAX rescaled to produce a final composite score between zero and one (0 being the least socially vulnerable and 1 being the most vulnerable).

An aggregation method using equal weights was applied due to the lack of theoretical justification for weighting one variable over another in all countries. Subsequently, we chose an additive approach for aggregation since it is transparent and easy to understand, a criteria we deemed important for potential users of the integrated risk modelling software. It should be noted that a web-based methodology for the development of weights using the Analytical Hierarchy Process (AHP) was developed and administered to stakeholders and experts within each country. Results were only obtained for Colombia, Venezuela, and Peru, and will be available in the future

Integrated Risk Index

Preliminary evaluations of indicators of socioeconomic vulnerability

Indicators of social vulnerability of Peru

By using the database already available from the Census and indicators of quality of Life, a preliminary assessment of the social vulnerability of the districts of Peru was performed. The indicators selected for the analysis are presented in the following table:

Dimension Indicator
Social Illiteracy rate
Population in school age (6-16) that do not attend school an is illiterate
Population that does not have any health insurance
Households in dwellings with overcrowding
Households in dwellings with overcrowding
Households without service of information or communication
Economic Self-employment rate and employment in micro business
Households with a high economic dependence

The next Figure presents the districts of Peru by ranges of the normalized indicators. From this Figure it is possible to identify different conditions of the socioeconomic context of Peru

 Illiteracy  Overcrowding
 Self Employment Economic dependence
 Comunication  Education
 Health  Population

By using the data available, a composite indicator of the social vulnerability of the districts of Peru is presented in the next Figure

This work is presented as a test case and proof of concept. The methodology used to construct the composite index of social vulnerability was adopted from Carreño et al. (2007) This evaluation requires the normalization/transformation of the indicators and their aggregation. For simplicity, a linear transformation was performed on the indicators by using the maximum and minimum values for each indicator (i.e. MIN/MAX rescaling). Min-Max rescaling rescales each variable into an identical range between 0 and 1 (a score of 1 being the worst rank for an indicator score and 0 being the best rank). Indices were aggregated by considering equal weights for this particular case.

A rigorous analysis of this index is still required. It is necessary to perform a statistical analysis of the database, select the most representative indices and perform robustness and sensitivity analysis. Once calculated, the indicators of socioeconomic vulnerability will be convoluted with the estimations of direct losses in order to obtain a comprehensive view of the seismic risk for this country and the others. An example of this assesment by using the OpenQuake-engine is presented in Burton and Silva (2014).

Integrated Risk Index per SARA countries

The evaluation of integrated risk is composed of the modeling of losses for each county (or first order impacts) and the models of social vulnerability (described in section above) that can be described as a potential aggravation coefficient of the first order impacts. Here, a physical risk index utilizes the probabilistic seismic hazard, the exposure, and vulnerability models; it combines them using the OpenQuake-engine to calculate numerous risk metrics: average annual losses, loss exceedance curves, risk maps for different return periods and loss curves at various spatial resolutions (Yepes et al. 2017). This methodology is explained in detail in… Modelling the Residential Building Inventory in South America for Seismic Risk Assessment (Yepes et al. 2015), and the probabilistic seismic risk assessment of the residential building stock in South America (Yepes et al. 2016). To derive an estimate of integrated risk, a total risk index was constructed via the mathematical combination of the social vulnerability index with the estimates of average annual loss. To accomplish the latter, the average annual loss values rescaled using the MIN-MAX method. Carreño et al. (2007; 2012) provide the aggregation method due to its simplicity, its successful use-case applications within literature (see Carreño et al. 2007; 2012; Fernandez et al. 2007; Khazai et al. 2008; Khazai and Bendimerad, 2011), and it’s method of treating the social vulnerability as an aggravation coefficient to the physical risk estimate. Here, the direct potential impact of an earthquake is denoted as Rt=Rf(1+F) where Rt is a total earthquake risk index, Rf is a physical risk index, which in this case is the average annual loss estimates for each country, and F is a social fragility index that was modeled here using social vulnerability. The aggregation scheme is a method to derive a total risk index or the potential impact of an earthquake that is obtained from the compounding of the physical risk index by an impact factor based on the socioeconomic characteristics within the country’s social systems (e.g. social vulnerability index). The integrated risk indices are outlined in the section below.

Back to the SARA Project main page

References

Burton, C., Silva, V. (2014) “Integrated risk modeling within the Global Earthquake Model (GEM): test case application for Portugal” Second Conference on Earthquake Engineering and Seismology. Istanbul Aug. 25-29 2014

Carreño ML, Cardona OD, Barbat AH (2007) “Urban seismic risk evaluation: a holistic approach”. Natural Hazards 40(1): 132–137.

Power, C., Daniell, J., Khazai, B., Burton, C., Oberacker, C. (2013) “National level Socio-Economic Vulnerability Database – Data Collection, Harmonisation and Analysis” Socio Economic Vulnerability and Integrated Risk Project. CEDIM. Global Earthquake Model

  • Audemard, F.A., Machette, M., Cox, J., Dart, R., Haller, K. (2000). Map and Database of Quaternary faults in Venezuela, U.S. Geological Survey Open-File Report 00-108. US Open-File Report
  • Khazai B and Bendimerad F. (2011) Risk and Resiliency Indicators, EMI Topical Report TR-1 1-03 link
  • development_of_indicators_of_social_vulnerability.1475674373.txt.gz
  • Last modified: 2016/10/05 15:32
  • by Miguel Toquica