The importance of accounting for equity in disaster risk models

The study addresses the gap between evidence of disproportionate disaster impacts on vulnerable groups and prevailing disaster risk modeling practices that emphasize aggregate, asset-based losses. The authors note that from 2005–2015 over 1.5 billion people were affected by disasters with losses around US$1.3 trillion, yet conventional models often fail to capture full human impacts. Differential burdens have been documented across income, social status, age, race, gender, and disability. The research is guided by two questions: (i) in disaster risk assessments produced by major international development organizations, to what extent and how are equity concerns included? and (ii) what are the implications at the assessment level of not including equity? To explore this, the authors develop a typology of equity integration in risk models and apply it to 69 public risk assessments (2010–2021), followed by a re-analysis of a Nepal earthquake assessment to compare results with and without equity considerations. The purpose is to reveal how excluding equity can obscure differential risks and potentially misdirect disaster risk management policies.

The paper situates disaster risk as a function of hazard, exposure, and vulnerability and reviews how risk assessments typically quantify impacts, often in monetary asset-loss terms to guide interventions and investments. Prior work shows that reliance on asset losses can bias policies toward protecting asset-rich areas, neglecting vulnerable populations who have fewer assets but face higher welfare impacts and slower recovery. Ex-post literature documents disparities by gender, race, ethnicity, age, disability, and wealth, but ex-ante models used for decision-making rarely quantify these disparities. The authors synthesize approaches into a typology (Types 0–5): Type 0 (no equity), Type 1 (qualitative/descriptive discussion), Type 2 (index-based social vulnerability indices), Type 3 (risk disaggregated by exposure), Type 4 (risk disaggregated by exposure and differential vulnerability), and Type 5 (welfare-loss approaches using equity weights, after Hallegatte and Vogt-Schilb). They note the paucity of comparative evidence on prevalence and effects of these modeling approaches, motivating their empirical review and case re-analysis.

The study comprises two parts. - Part I: Risk assessment review. The authors searched repositories of major international organizations: World Bank Open Knowledge Repository, Asian Development Bank, Inter-American Development Bank, and PreventionWeb. Search terms included variations of “disaster risk assessment”, “climate risk assessment”, and “climate vulnerability assessment”, complemented by repository category filters. From 901 unique documents (World Bank: 424; ADB: 285; PreventionWeb: 125; IDB: 67) within 2010–2021, they applied inclusion criteria: multisectoral risk assessments intended for international development planning, with a primarily quantitative, primary analysis. Sixty-nine documents met criteria. For the 30 that mentioned equity, the team classified how equity and differential risk (race, income/class, gender, disability, age) were represented, developing a typology (Types 0–5). A QA/QC process included spot-checking ~26% of included items (N=18) and ~10% of excluded items (N=83), confirming original decisions and harmonizing reporting. - Part II: Nepal re-assessment. Building on a district-level multi-hazard risk assessment for Nepal (focusing on a 1-in-500-year earthquake), the authors implemented equity-informed analyses: - Type 3 (differentiated exposure): Disaggregated previously modeled fatalities by group proportions (gender, age, disability, caste, income) using census-based group shares by district to apportion total fatalities. - Type 4 (differentiated exposure and vulnerability): Where literature provided group-specific relative fatality risks (odds ratios) for gender, age, and disability, they adjusted fatality allocations to reflect differential vulnerability while preserving total district fatalities. - Type 5 (welfare-loss using equity weights): They estimated asset losses by combining earthquake shaking intensity, district building-type distributions, fragility curves, and replacement costs. They then applied equity weights to redistribute losses by district proportional to income, using an elasticity of marginal utility of income (gamma=1.2) and district mean incomes to calculate equity-weighted losses. Due to data limitations, weighting was applied at district-level (not household-level). Data sources included Nepal census, building inventories, fragility curves, and published replacement cost data. Code for figures and analysis was implemented in Python and made openly available.

- Prevalence of equity in assessments: Of 69 risk assessments (2010–2021), only 30 mentioned equity; most did so qualitatively. Typology frequencies: Type 0: 39; Type 1: 11; Type 2: 8; Type 3: 8; Type 4: 0; Type 5: 3. Only ~28% attempted quantitative evaluation of differential impacts. No assessments implemented Type 4, and only three (all World Bank-supported) used Type 5. - Equity parameters included (non-mutually exclusive): Class/Income: 29; Gender: 16; Age: 15; Race: 7; Ability/Disability: 6. - Nepal re-analysis results: Disaggregating fatalities revealed large disparities among groups not visible in the original Type 0 assessment. The bottom income quantile exhibited approximately double the fatality rate of the top quantile when vulnerability was considered. Adults aged 5–60 had sharply lower mortality rates than <5 and >60 groups; gender and disability differences were also evident when vulnerability adjustments were applied. Caste-related disparities emerged largely through spatially differentiated exposure. - Welfare vs total losses: Unweighted (asset) losses increased with district average income, concentrating in wealthier urban areas (e.g., around Kathmandu). After applying income-based equity weights (Type 5), the pattern reversed: welfare losses were greatest in poorer, rural districts, indicating higher welfare impacts despite lower absolute asset losses.

The study shows that contemporary disaster risk assessments rarely incorporate quantitative equity considerations, potentially biasing risk reduction investments toward asset-rich areas and obscuring the higher welfare impacts borne by marginalized populations. Incorporating equity via higher-order approaches (Types 3–5) can identify at-risk groups and illuminate mechanisms (differential exposure and vulnerability), improving targeting and design of interventions (e.g., gender- or disability-informed measures) and informing decisions across the disaster cycle. Data limitations are often cited; however, the Nepal case demonstrates that meaningful equity-aware analyses can be conducted with publicly available district-level data. The authors caution that disaggregating exposure alone (Type 3) may be misleading without group-specific vulnerability information (Type 4), underscoring the need to develop context-appropriate vulnerability models. Beyond distributional equity, the discussion highlights the importance of procedural equity (participation of affected communities and transparent processes) and recognition equity (capturing non-economic losses such as psychosocial impacts, cultural heritage, education disruption, and displacement). The paper also notes potential trade-offs between efficiency (reducing total risk) and fairness (equitable distribution), which should be explicitly addressed in risk management.

Few contemporary risk assessments explicitly or quantitatively incorporate equity, and conventional asset-focused methods can conceal substantial disparities in disaster impacts. The authors’ typology provides a framework to assess and improve equity integration. Re-assessment of Nepal demonstrates that equity-aware approaches reveal different priority areas and affected groups: welfare losses concentrate in poorer districts, and fatality risks differ markedly by income, age, gender, disability, and caste. International development institutions are well positioned to drive adoption of equity-aware modeling. Further research should advance methodologies for differential vulnerability modeling, welfare-based evaluation with richer household-level data, and systematic inclusion and reporting of procedural and recognition equity. While models alone cannot resolve structural inequities, equity-aware models are an important tool to guide more just disaster risk management.

- Scope of document review: Focused on ex-ante, multisectoral, quantitatively oriented assessments from selected international repositories (2010–2021), potentially omitting other relevant assessments or gray literature; procedural equity activities were rarely reported and may be inconsistently documented. - Equity focus: The analysis centers on distributional equity; procedural and recognition equity are discussed but not systematically measured. - Data constraints in the Nepal case: Household-level data were unavailable; equity weighting was applied at district level rather than household level, and group-specific vulnerability adjustments used literature-derived odds ratios only for gender, age, and disability. Vulnerability models for other groups (e.g., caste, income) were not available; thus, Type 3 results may overemphasize exposure differences and underrepresent social vulnerability. - Welfare-loss implementation: The equity weights reflect only one component of welfare loss and do not capture other household financial characteristics (e.g., savings, insurance, social protection), which would be considered in full welfare-based models.