June 17, 2021
Can private insurance encourage homeowners to adapt their properties in order to reduce escalating flood losses from climate change? In a new research study with my coauthors, Howard Kunreuther and Carolyn Kousky, we find substantial demand for a novel supplementary flood insurance product that funds low-cost adaptations to reduce future damage when rebuilding after a flood—and we show that auto-enrollment increases that demand.
Flooding is the most common and costly natural disaster risk facing American households. Moreover, as flood-related losses are expected to continue escalating due to climate change, the need for investment in risk mitigation is growing. For the millions of residents exposed to significant flood risk, increasing their property’s flood resilience offers an attractive and underutilized risk management strategy that can save money in the long run. But many homeowners lack sufficient savings or financing to cover the upfront costs. As a result, they cannot simply adapt their properties and recoup their investments over time through reduced flood damage. In fact, while estimated costs of relatively modest adaptations like replacing building materials with flood-resilient ones are upwards of $9,000 for an average home, 47% of American households would struggle to cover even a $400 unexpected expense.
Potential Financing Solutions
One approach to financing residential adaptation is through public programs like low-interest loans or grants to homeowners. To date, however, public funding opportunities have been limited and cannot plausibly scale to cover all the necessary investments. Another approach to broaden access to funds for flood mitigation would be to utilize private flood insurance to cover adaptive repairs. The repair and rebuilding process after a flood presents an opportunity to make cost-effective retrofits that lower future flood losses and also reduce any risk-based insurance premiums that homeowners pay. In theory, insurers could charge slightly higher premiums on flood insurance and cover the extra costs of building back more flood-resiliently.
In practice, there have been few opportunities for the insurance sector to experiment with such strategies. The vast majority of flood insurance policies in the US are purchased through the federal National Flood Insurance Program (NFIP), housed in the Federal Emergency Management Agency (FEMA). The standard flood insurance policy is the same for everyone in that program. While the NFIP has covered limited mitigation measures for some homeowners through Increased Costs of Compliance (ICC) coverage since 1994, it falls far short of funding all cost-effective adaptation.
Roughly 6% of all residential flood insurance policies are written by the private market, and that number is growing. In general, private flood insurance firms are able to innovate with respect to the type of policies they offer.** The Flood Insurance Agency (TFIA)—one of the first firms to sell private market residential flood insurance policies in the US—introduced a new supplemental product called FloodReady. It enables homeowners to cover up to $10,000 of the additional costs for rebuilding their damaged property using flood-resilient materials for a small increase ($35) in their annual insurance premiums. In a new research project, my coauthors and I studied the demand for this supplemental insurance among homeowners buying standard flood insurance coverage.
Role of Behaviorally Informed Design
Behavioral economics has found that defaults are generally a powerful and low-cost tool that can encourage people to make good choices. Specifically, automatically enrolling consumers in an option–even when the process of unenrolling is as simple as clicking a button–has been shown to increase take-up of that default option across a wide range of contexts, including retirement saving, organ donor lists, and debt repayment plans.
Yet, there is little field evidence on the efficacy of auto-enrollment for insurance decisions and there have been no tests of whether defaults influence real flood insurance choices. In addition to studying demand for FloodReady, our research aims to fill this gap in evidence and show whether auto-enrollment works in the context of property insurance.
Our Field Study
In collaboration with TFIA, we conducted a research study examining homeowner demand for FloodReady and testing whether auto-enrollment into FloodReady could increase demand.
Comparison Groups: Opt-In vs. Auto-Enrollment to FloodReady
From June 2019 through February 2020, TFIA assigned residential flood insurance applicants to either an auto-enrollment process where FloodReady coverage was included in a residential flood insurance application by default, or to an opt-in process for FloodReady, based on the region where the applicant lived and the timing of their application. The table below describes the assignment process in the columns labeled Period 1 and Period 2. Overall, 30% of applicants during these nine months were assigned to auto-enrollment and 70% were assigned to opt-in enrollment. As this is not a randomized assignment, to isolate the causal effect of auto-enrollment from potential differences over time and between the regions, our analyses also used data from April 2018 through April 2019 prior to any treatment when applicants from both regions received the “control” opt-in process as opposed to the “treatment” auto-enrollment process (labeled Period 0 in the table).
We found that being assigned to the auto-enrollment treatment dramatically increases the likelihood of purchasing FloodReady relative to the opt-in control. In terms of raw means, 32% of consumers who purchased their policy with the opt-out FloodReady process purchased this coverage relative to baseline take-up under the opt-in process ranging from 10%-16% over each region and period of the study.
These results are displayed in the figure below, which plots the 7-day moving average share of total policies with FloodReady separately for the subgroups in each cell of the table. The dashed lines indicate the opt-in process while the solid lines indicate the auto-enrollment process (AE denotes auto-enrollment, shaded areas indicate 95% confidence intervals, with Region 1 in orange and Region 2 in blue).
This figure reveals the similarity in pre-treatment take-up of FloodReady for the two regions during Period 0 and the large jumps in take-up for the subgroups experiencing auto-enrollment in Periods 1 and 2, relative to those still receiving the opt-in process. Consistent with the large impact of auto-enrollment apparent in this plot, the main causal estimates in our regression model show a 23 percentage point predicted increase in take-up of FloodReady–or a 150% increase from an average predicted probability of 12% under opt-in to an average predicted probability of 35% with auto-enrollment.
There are three main lessons from our findings. First, there is consumer demand for products like FloodReady. Second, as a result, flood-resilience insurance offers a promising strategy for encouraging property-owners to finance adaptation measures after experiencing flood-related damage. Third, auto-enrollment is likely critical to maximize participation in such insurance programs. Incorporating auto-enrollment and bundling supplemental resilience coverage with standard coverage may be a crucial piece of designing financing arrangements to save costs in the long run for individual consumers in flood-prone areas.
**Based on authors’ calculations of 2019 and 2018 private market share of total residential policies in force at year-end using statistics from the NAICS 2020 Private Flood Data Call results and NFIP policies in force.