Abstract: Are We Equally Prepared for a Disaster? an Affordability Study of Homeowners Insurance and Flood Insurance in Puerto Rico (Society for Social Work and Research 24th Annual Conference - Reducing Racial and Economic Inequality)

Are We Equally Prepared for a Disaster? an Affordability Study of Homeowners Insurance and Flood Insurance in Puerto Rico

Saturday, January 18, 2020
Marquis BR Salon 13, ML 2 (Marriott Marquis Washington DC)
* noted as presenting author
Chenyi Ma, PhD, Postdoctoral Fellow, University of Pennsylvania, Philadelphia, PA
Amy Baker, PhD, Assistant Professor, University of Pennsylvania, Philadelphia, PA
Background: Climate change poses significant risks for communities pre-disposed to accumulated disadvantage (Kemp & Palinkas, 2015). While a growing body of literature demonstrates how storms produce displacement, disaster’s impact on housing type and access to insurance is underdeveloped leaving unanswered questions about who is at risk of new forms of inequality exacerbated by climate change.  We begin addressing this using the FEMA IA dataset on damage produced by Hurricane Maria in Puerto Rico (PR).  Maria was responsible for the losses of 3,000 people and $90 billion in infrastructural damage. In response, FEMA inspected affected homes; recorded damage (no, minor, major, or destroyed); identified damage vectors (wind or flood), and confirmed if Homeowners Insurance (HI) and/or Flood Insurance (FI) covered the property. However, the severity of individual level property damage often determines a community’s recovery pace and if they are in a more vulnerable economic position post-storm. Thus far, how the take up of FI and HI interacts with Maria’s damage characteristics has not been investigated. We ask: Were low-income homeowners more or less likely to have HI and FI before Maria? Were homes covered by HI and FI more or less likely to be damaged?

Methods:  Our sample consists of all owner-occupied homes in PR inspected by FEMA post-Maria (N=522,079). First, we examined the prevalence of HI and FI, cross-tabulated with damage characteristics, residential types, and low-income households (annual household income below the median level of PR), and damage vectors. Then, we employed multiple regressions in binary logistic models to predict the probabilities of a home taking-up a).HI, and b). FI; and a generalized ordered logistic model to estimate the relative risks of a home damaged at different severities.

Results:  Only 20% were covered by HI, 4% by FI. The extremely low prevalent rates of HI and FI were found among the 3,372 destroyed homes, at 1.25% and 0.3% respectively.  11% of the wind-damaged homes were covered by HI; 10% of the flood-damaged homes by FI.  Damage caused by winds was more likely at the destroyed level than floods (OR=3.82, p<0.001). Mobile homes had the lowest insurance coverage rate, at 5% (HI) and 0.33% (FI). Low-income households were significantly less likely than their counterparts to take-up HI and FI (OR=0.26, p<0.001; OR=0.25, p<0.001) before Maria. Homes without HI and FI in each had significant higher risks than their counterparts to be damaged at the major (OR=2.54, p<0.001; OR=6.34, p<0.001 respectively), and the destroyed (OR=4.51, p<0.001; OR=10.64, p<0.001 respectively) levels. In contrast, low-income households were more likely to live in the mobile homes (OR=1.28, p<0.001), to have damage by winds (OR=1.88, p<0.001), and to suffer from home damage at all levels (p<0.001).

Implications:  These findings indicate that entrenched economic inequality present before disaster likely produces deeper forms of housing inequality based on access to insurance, potentially eroding the accumulated capital of entire communities.  Nonetheless, existing policy focuses on post-storm efforts that are not tailored to differentiations in groups pre-storm leaving disadvantaged groups at risk of a slow or non-existent recovery.