Conversations about Risk Rating 2.0

Part II with Nancy Watkins

May 24, 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. This new approach to pricing harnesses catastrophe models to help better understand flood risk. In this second post of the series, we take a deeper look at these models with Nancy Watkins.

Nancy P. Watkins is a principal and consulting actuary in Milliman’s San Francisco office, leading a consulting practice that specializes in climate resilience, insurtech, and catastrophic property risk. Her team provides state-of-the-art tools, technology, and analysis to insurers, reinsurers, and stakeholders in the flood insurance space. She serves as the client relationship manager for NFIP Risk Rating 2.0 and provides strategic advice and support to FEMA. Nancy leads the global Milliman Climate Resilience Initiative. She serves on the United Nations Capital Development Fund’s Climate Insurance Linked Resilient Infrastructure Finance Working Group to pilot climate adaptation financing for emerging markets and least developed countries.  She was awarded the American Academy of Actuaries Outstanding Volunteerism Award for her participation in the Flood Insurance Work Group. 

Thanks for talking with us, Nancy. Can you begin by explaining why the new NFIP rates are such a leap forward in flood risk assessment?

Watkins: The new NFIP rating system harnesses modern catastrophe models, as well as the program’s historical data and experience over the past twenty-five years. To create Risk Rating 2.0, FEMA is using flood catastrophe models from three commercial model vendors as well as a proprietary model derived from FEMA’s Flood Insurance Rate Maps and other data sources. This comprehensive approach provides a much better depiction of the variation of flood risk across the entire United States, and not just within the Special Flood Hazard Areas.

Milliman was engaged starting in 2017 to collaborate with FEMA’s actuaries and staff on the redesign of the NFIP’s pricing approach. This analysis took over three years and synthesized more data and models than we’ve ever seen for a flood rating plan; it is probably the most sophisticated catastrophic insurance rating plan developed to date.

NFIP’s new pricing now includes not just major perils, such as riverine flooding and storm surge, but also more localized flood perils including Great Lakes flooding, tsunamis, and flash flooding, as well as coastal erosion. This type of modeling provides granular local risk differentiation throughout the entire U.S. and territories. This is a leap forward from the coarse flood zones used in the old rating system and provides much more accurate indications of flood risk at each property.

Most importantly, the final plan provides greater quantification of the factors that drive flood risk, and not just “the answer.”

What does Risk Rating 2.0 mean for the cost of flood insurance going forward?

Watkins: It is important to note that Risk Rating 2.0 isn’t in-and-of-itself raising overall insurance premiums. Insurance premiums are high when the risk is high. NFIP premiums were already increasing under the old rating methodology because the premiums in total have been demonstrably insufficient to keep the program from being heavily in debt. The need to increase rates for NFIP policies existed before Risk Rating 2.0 and was not created by Risk Rating 2.0. In part, the need arose because the pricing under the old methodology was unsound and created incentives to increase risk. By better aligning price with risk, Risk Rating 2.0 effectively helps us as a society make better decisions that might lead to lower expected losses which would result in lower insurance premiums over the long term. But the risk has to be reduced first, we can’t just pretend it’s not there.

Why was it important for FEMA to start using catastrophe models in their rate setting?

Watkins: Catastrophe models are computer simulations of thousands of hypothetical, but realistic, events including the losses from those events and then what that means for insurance claims. This enables us to estimate crucial metrics such as the Average Annual Loss and a curve of losses corresponding to different exceedance probabilities.

FEMA needed to use catastrophe models in addition to the historical NFIP experience because basing rates solely on historical events does not adequately represent all plausible events and drastically changes the risk view from year to year, even for a portfolio as large as the NFIP. This new modeling also makes it possible to show risk everywhere in U.S., not just in the flood zones previously mapped by FEMA.

What is the difference between a catastrophe model and a climate model? Can catastrophe models help us think about how flood risk might change as the planet continues to warm?

Watkins: Historical experience is a reliable predictor of future losses only when the historical experience adequately captures the full range of possible events, and when the frequency and severity of events are not changing with time. Neither of these conditions is true for flood. The first condition requires the use of cat models, and the second one will eventually require those models to be calibrated to reflect changing climate conditions since the climate that produced the historical events is different from the future climate.

Although it is certainly true that cat models are needed to account for climate change, that is not central to Risk Rating 2.0 due to the time scales involved.  Risk Rating 2.0 rates are set for the near-term future in which the premiums will be charged and the flood risk will be accepted by the NFIP, which we believe reflects the current state of the climate.  That said, the design of Risk Rating 2.0 facilitates extremely valuable testing of future flood risk under climate-informed conditions.

Can catastrophe models help us think about risk reduction, as well as insurance prices?

Watkins: Yes to both. Models can help us better understand the benefits that risk reduction can provide in terms of lower losses. When this is reflected in insurance prices, it can promote risk reduction by encouraging people to build in less risky locations and to build structures that are less susceptible to flooding. When we build in a safer way, insurance premiums will be lower. In the long run this creates both a more resilient society and a more robust flood insurance program.

Catastrophe models have sometimes been controversial with state insurance regulators. Why is that?

Watkins: Many flood cat models are relatively new and may produce widely divergent views of risk.  It is important to note that divergence does not necessarily indicate that less mature models are not fit for use, even if they disagree in material ways. Using flood models effectively may require extensive evaluation and comparison, a deeper analysis of how they are built, and consideration of model adjustments to be suitable for the intended purpose.

Regulators often don’t have the resources or in-house expertise to review multiple models, and consequently may believe that cat models are a “black box.” When we do rate filings for private flood insurance coverage, we have found that the NFIP’s use of cat models helps legitimize their prominence in flood ratemaking in the eyes of regulators.

Also, the way that the NFIP incorporated individual geographic and structural risk descriptors helps regulators understand more about what is inside the “black boxes” and make the analysis more transparent and understandable.  As an example, to quantify storm surge risk the new Risk Rating 2.0 algorithm incorporates geographic factors such as distance to coast and elevation relative to sea level.  To quantify inland flood risk, the algorithm incorporates different geographic factors such as distance to a river, elevation relative to a river, and elevation relative to the area around the structure. For both storm surge and inland flood risk, the algorithm incorporates structural factors such as first floor height and foundation type.

Does FEMA’s use of these models change how the broader insurance market is thinking about them?

Watkins: Definitely! The NFIP tested several models, used its own data to validate the models, blended outputs from multiple models, and “smoothed” or eliminated anomalous results to produce the Risk Rating 2.0 rates.  This synthesis of rich and granular data from multiple sources provides a valuable benchmark for companies, regulators, and the cat model vendors themselves.

Do you think Risk Rating 2.0 will have any impact on the emerging private residential market for flood insurance?

Watkins: The NFIP has been the dominant provider of flood insurance in the U.S. for over 50 years, but private flood insurers historically didn’t trust that the NFIP’s rates properly measured the risk being transferred. The common wisdom was that the NFIP was generally underpriced, so most insurers didn’t want to compete with them or didn’t believe that they could. Those who did write flood insurance typically had no better benchmark than the NFIP, even though they didn’t believe its rates to be sound.

Risk Rating 2.0 establishes rates that private insurers are likely to find more credible and therefore will be more willing to use as benchmarks.  They may be more willing to enter the private flood market if they believe that the NFIP isn’t offering artificially low rates with which they have to compete.

Is there anything else you’d like to share?

Watkins: This is a situation in which a rising tide lifts all boats; that is, private flood insurance should not be viewed as a competitor to the NFIP.  The market is so large and so underserved that there’s virtually limitless potential for growth and competition.  If Risk Rating 2.0 encourages more insurers to believe that they can manage and measure flood risk, and NFIP’s consumer outreach brings greater awareness of flood risk, the result should be increased take-up of flood insurance – both public and private.  As the NFIP’s investment in Risk Rating 2.0 helps to close the flood protection gap, it will be an enormous benefit to consumers, communities and businesses across the country.