Conversations about Risk Rating 2.0
Part III with Dave Burt
June 1, 2022
The Wharton Risk Center is undertaking a series of interviews with experts on the new pricing approach for the National Flood Insurance Program (NFIP), called Risk Rating 2.0. In today’s post, we speak with Dave Burt, Founder & CEO of DeltaTerra Capital.
DeltaTerra Capital is an investment research company providing impact assessment tools and working with policy makers to help minimize the damage done to homeowners as asset value adjustments happen.
Founded in 2019 by a team of experienced investors, DeltaTerra uses its proprietary suite of risk models called DeltaTerra Klima™ to bridge climate science and investment science. By translating newly available scientific estimates of physical risk into actionable insights for investors, lenders, and policymakers, DeltaTerra aims to accelerate an orderly transition to a climate-resilient financial system and society.
Your organization uses lots of sophisticated data and modeling to think about risk. Can you describe the ways that DeltaTerra Capital thinks about flood risk, and how it may be changing?
Dave Burt: We built our suite of risk models to look at all climate-linked hazards, but flood risk is the most concerning and has been a primary focus area for us. Flood risk is not only the largest source of property damage but also the most mispriced hazard risk in real estate capital markets. We believe that this capital markets mispricing could be just as damaging financially to homeowners as actual acute flooding events.
We performed an extensive analysis last year of flood and wildfire risk in single-family home markets and as part of that exercise, we modeled the below estimates of expected flood damages now and in 2050 under two IPCC scenarios called RCP 4.5 and RCP 8.5. This analysis utilized a variety of private and public input data including: climate-science based hazard analytics purchased from ICE Climate (formerly risQ), historical private insurance claims estimates purchased from Verisk, as well as publicly available data from FEMA, the SBA, and the Census Bureau. The table below provides our estimates of total homes, current risk, and future risk in three types of continental US communities: Special Flood Hazard Areas (SFHA) with 1-in-100-year flood risk based on FEMA flood maps, non-SFHAs where our models show high flooding risk despite being outside FEMA-designated flood zones, and non-SFHAs that we believe to be accurately classified as low-risk.
In 2020 (the most recent year with complete data availability), the NFIP insured 3.7 million single-family homes in these same communities, collecting ~$2.5 billion in premium payments. When compared to our estimates of those at risk, it shows that takeup rates are very low. In addition, they have been using an extremely outdated premium-setting system that was just updated last year after 50 years of risks increasing. In the table below, we show the number of 2020 NFIP policies created (based on data downloaded from OpenFEMA) for single-family homes in each of our three community types. We also provide an estimate of the aggregate current coverage gap relative to our current modeled risk estimates.
Beyond the challenges this underinsurance condition creates for communities when a flooding event occurs (like after Harvey when only ~20% of flooded homes in the Houston area were covered), it also subjects homeowners to dangerous repricing risks. Damages have to be paid for by someone and we argue that it isn’t fair, informative, or sustainable to socialize the costs through NFIP losses, FEMA individual assistance, and low-interest disaster loans through the SBA. The problem is that home values, which often represent a very important financial asset for ordinary families, reflect the cumulative cost of insurance payments that will be required now and in future years. This means that when expected insurance costs do rationally increase, the impact on home values and homeowner financial health are compounded beyond just one year’s increase.
The only way we see to alleviate the problem is to allow, or even encourage, the revaluation of mispriced homes while protecting households that don’t have the means to absorb the financial hit. Any other solution simply serves to kick the can down the road, which ultimately results in an even worse outcome for even more exposed households. However, policies that seek to accomplish this important fix to our financial system could have an immediate negative impact on at-risk homeowners, a large share of whom are low- or moderate-income households that were already struggling with insurance affordability.
DeltaTerra recently examined what we might see coming with Risk Rating 2.0 (see post). What insights did you find in your look at the data last fall? Do you have any updates to that analysis of how things have continued in early 2022?
Burt: In order to anticipate the potential impacts of RR 2.0 on different markets, we delved into the differences between current insurance premium collections and actual hazard risks. We discovered an unappreciated glitch in the initiative’s rollout plan: While homes with existing policies are protected from abrupt price hikes by an 18% per annum legal limit on increases, the law does nothing to protect currently uninsured homes. Due to a variety of systemic challenges around encouraging NFIP participation, this sadly accounts for a majority of homes with high flood risk. The below table from a 2018 FEMA affordability study shows the total number of households in and out of flood zones that are policyholders. Of the 5.1 million households identified as being in SFHAs, only 1.8 million were insured by the NFIP.
An even more concerning aspect of the low take-up rate is that it appears to be more pronounced for low-income households. This is intuitive since these households are less likely to be able to afford flood insurance in the first place. The median income of non-policyholders in FEMA-established flood zones was found to be just $40,000, barely more than half the $77,000 in median income for policyholders in flood zones. The uncapped rate increases for uninsured homes are almost certain to exacerbate this equity problem and also concentrate market value shocks in communities that are least able to absorb them.
Our fears that uncapped RR 2.0 rates would be unaffordable for new policy applicants (and those that previously lapsed on their policies because they couldn’t afford them) are unfortunately becoming reality, as evidenced by the alarming drop in the number of new NFIP policies created after Phase 1 of RR 2.0 began on 10/1/2021. This trend has continued into 2022. The chart below shows the percentage of new policies created at the new rates, versus renewals that benefit from protections that delayed increases until 4/1/2022. It seems reasonable to conclude from this analysis that uncapped RR 2.0 rates are significantly higher than the old rates, further discouraging take-up of flood insurance by homeowners.
Source: DeltaTerra using NFIP policy data released via Open FEMA on 3/9/2022
Do you think these changes in pricing will change homeowner behavior? In what ways?
Burt: Of course. A home is often the most important financial asset for a household and homeowners respond viscerally to value-influencing changes in policy and other market dynamics. I’ve had several people tell me recently that they decided not to purchase a home because of a high flood risk determination from First Street Foundation via one of the home buyer websites like Redfin, Realtor.com, or Estately. Millions of homes are facing increasing insurance costs since Phase 2 of Risk Rating 2.0 began on 4/1/2022. Once homeowners at risk understand that these problems are beginning to influence buyer decisions, they will probably do what they can to protect their home’s value. If they can’t afford to make their home more resilient through adaptive measures and its value declines, they may seek to move to less costly regions and in some extreme cases, if risk increases substantially and property values fall, they may default on their mortgage.
You used to work as an institutional investor and have commented publicly on how climate risk could impact the mortgage bond market. Have you seen any changes in mortgage investor behavior due to information now available because of RR 2.0?
Burt: Unfortunately, the information that is currently available from Risk Rating 2.0 is limited and investors still don’t have the data they need to explicitly measure the extent of uninsured and underinsured borrowers and the financial hardships they may face as insurance premiums rise. FEMA published a Risk Rating 2.0 methodology and data sources paper in April 2021 but the paper didn’t offer summary results, just one example calculation. They also publicly released Ratings Factors and a Premium Calculation Worksheet Example such that anyone could theoretically calculate the premium for a given home if they had all of the required location and structural information needed for the calculation. This information is not easily obtained, however, and often impossible to obtain for a mortgage investor because of borrower privacy considerations.
The only quantitative information currently available about the new RR 2.0 rates is from FEMA’s National Rate Analysis, which was also released in April 2021. This dataset (available for download at county or zip code levels) detailed the number of policyholders in each region that would see rate increases across different categories of monthly payment changes. But this exercise applied only to existing policyholders, all of whom were subject to the 18% a year statutory cap on premium increases. It said nothing about the actual new rates that would be paid by policyholders after many years of 18% hikes, or how much new policies would cost for the millions of at-risk homeowners currently without insurance.
FEMA may be reticent to announce the full magnitude of flood insurance cost increases under the new pricing method to avoid widespread voter displeasure and congressional opposition. In any event, the National Rate Analysis provides very little information about the actual price levels coming out of the new model, just a general sense of direction for regionally defined pools of existing policyholders. The National Rate Analysis would give us at least a near-term view of impact across different regions if most of the at-risk homes had a flood policy already but FEMA’s own affordability analysis indicates otherwise.
While FEMA sets the rates for NFIP, they have little influence on the mandatory purchase requirement (MPR) that is intended to boost take-up rates in flood zones. A GAO report from last summer identified probable compliance issues with the MPR and recommended greater information sharing amongst agencies. While the MPR is a lender requirement enforced by a range of federal regulators, it is highly influenced by the policies of government mortgage agencies Fannie Mae, Freddie Mac, and the Federal Home Loan Banks. The most important regulator for these entities is the Federal Housing Finance Agency (FHFA), which is currently leaning into climate risk and may someday require greater disclosure of flood zone and insurance status for government-guaranteed loans. We see this move as an important first step towards greater ability by institutional investors to measure and manage flood risk in mortgage markets.
Despite the MPR’s importance in building a climate-resilient financial system, it does require lenders to forcibly place expensive insurance policies on non-compliant borrowers who may already struggle with affordability. Like the GAO, we would like to see the FHFA and FEMA work together to overhaul the MPR while pairing reforms with affordability assistance programs that protect lower-income borrowers from financial distress during the transition to a more fully-insured housing market. The House Select Committee On The Climate Crisis recently worked to incorporate funding for such programs as part of the Infrastructure Investment and Jobs Act that provides $700 million a year in Flood Mitigation Assistance in each of the next five years.
How will uptake of flood insurance with the roll out of RR2.0 influence the type of data available?
Burt: FEMA discloses policy data on a monthly basis that includes premium rate charged and census tract location for each policy. This data source could provide a useful sampling of the extensive flood risk modeling work that went into the new pricing methodology. Unfortunately, very few homeowners are accepting the new full rates and FEMA does not have disclosure fields yet for modeled rates prior to the 18% cap being applied. As such, we may not see many actual modeled rate instances until disclosure policies and processes are updated.
Is there anything else you’d like to add?
I’d like to encourage your readers to work on these difficult issues and to advocate for transparency and compassion whenever possible. There are millions of households who have just experienced a large hike in the cost to protect themselves against flood risk. Many of these families couldn’t afford the old rates and may face limited access to credit in places where flood insurance is required to get a loan. There has been no public disclosure about the impact of these systemic changes on uninsured households which may create distrust and undermine popularity for both the Risk Rating 2.0 program and the broader climate risk mitigation effort.
Compassionate policymakers understand that, if done poorly, reform could be disastrous for our most vulnerable communities and exacerbate inequity challenges. They are also beginning to understand that reform is necessary to prevent much bigger problems down the road.
We believe that DeltaTerra Capital’s important contribution will be to supply policymakers and other systemically important capital markets agents with the measurement tools they need to achieve both aims simultaneously. It is possible to pair reforms with intentionally constructed affordability programs to set us on a path to financial resiliency without creating a crisis for low-income homeowners and communities. However, it will require better data availability and lots of work, and probably needs to happen soon.
I’d like to thank the Wharton Risk Center for all the excellent work on flood risk over the years and for giving us this opportunity to contribute to the discussion.