How Income Inequality Influenced Affordability of Homeowner's Insurance Among Disaster Victims: Evidence from Hurricane Maria in Puerto Rico

Background/Purpose: In September 2017 Hurricane Maria ravaged Puerto Rico (PR) with sustained winds of 160 mph that damaged 1/3 of all occupied homes leaving these households in need of recovery and at-risk for downward mobility. Existing studies suggest Homeowners’ Insurance (HI) plays a critical role in achieving speedy housing recovery, but low-income households likely have less access to affordable HI than their counterparts in PR (Ma and Baker, 2020). Further, social epidemiologists speculate that “there is a cross level interaction between community income inequality and household income, such that the most adverse effect of income inequality is on poor individuals, while income inequality may not have any impact on rich individuals (or may even be beneficial)” (Subramanian et al, 2003). Income inequality is deeply rooted in PR, with its Gini-Index among the highest in the world (Hernandez and Schmidt, 2018). However, whether or not income inequalities at the local community level might play a role on individual access to homeowner insurance remains missing from the literature.

We focus on homeowners who suffered damage from Hurricane Maria in Puerto Rico with the following research questions:

1. To what extent was the community level of income inequality for these homeowners related to their decision to adopt homeowners insurance?
2. To what extent did the community level of income inequality for homeowners influence the relationship between their incomes and decisions to adopt homeowners insurance?

Methods: Using Individual Assistance Housing Registrant data collected by FEMA after Hurricane Maria (N=265,807), we employed four nested multi-level logistic regressions to estimate household probabilities of adopting HI, given the conditions of sequentially added predictors of i) null; ii) household income and other housing characteristics; iii). Gini-index of the Census-tracts where the households were located; iv) the interaction term of income and Gini-index.

Results: First, our simplest random-intercept model found significant variation in HI adoption probabilities across 881 census tracts (Median Odds Ratio [MOR] = 2.28, 95% Confidence Interval [CI]: 2.17, 2.38). Second, when adding individual household and housing characteristics, households with higher incomes were significantly more likely to adopt HI (Odds Ratio [OR] = 1.55; 95% CI: 1.53, 1.47). Third, the addition of income inequalities in terms of Gini quartile ranges showed that relative to the lowest range, households in the three higher-range census tracts were less likely to adopt HI (OR = 0.85/ 0.71/ 0.73 respectively; 95% CI: 0.74, 0.96/ 0.62, 0.81/ 0.73, 0.83 respectively). Finally, the addition of a possible interaction between income and income inequality showed that income inequality significantly influenced the relationship between income and HI adoption behavior (OR = 1.04/ 1.06/ 1.12 respectively; 95% CI: 1.01, 1.07/ 1.03, 1.09/ 1.09, 1.16 respectively).

Conclusions and Implications: Income inequality played different roles for poor and rich: impeding HI adoption behavior for low-income households while facilitating such a behavior for high-income households. Policy formulation should consider public-private partnerships to address HI affordability for those low-income households most vulnerable to housing damage.