Critique of Current Social Vulnerability Indices and Opportunities for Improvement

Between 1980 and 2013, the United States (U.S.) experienced 151 weather related disasters causing approximately $1 billion in overall damages with total costs exceeding $1 trillion. Social vulnerability (SV) is a widely used concept that aims to assess the differences in the susceptibility to disasters, losses, and coping and recovery abilities of communities. The SV of populations at risk of disasters in the majority of cases is expressed as an index (SVI) which has the potential to be used for deriving proactive plans that will protect communities and assist them to rebound from emergency situations. The majority of indices aiming to assess SV are derived with a composite model based on principal component analysis or percentile ranks. Only a few studies have attempted to assess existing SVI in terms of their relation to potential losses from disasters; these assessments found a limited predictive performance in terms of identifying potentially high risk areas. We argue and demonstrate that the current methodologies for deriving SVI may not capture the qualitatively differentiating nature of vulnerability of communities in geographic areas and do not provide a practical and reliable planning tool. Our study proposes a paradigm shift by considering SV to disasters as a classification issue and, consequently, by introducing classification modeling and performance assessment techniques which are likely to provide a different perspective on attributes influencing SV as well as a reliable approach to identify potentially high risk areas. To demonstrate the potentials of this approach historical U.S. Census and hurricane loss data from the FEMA Hazus® program were used for the Houston metropolitan area.