The flood insurance protection gap in the United States is massive, but the take-up rate of private flood insurance continues to be low. The relative inability of private insurers to penetrate the flood insurance market has been blamed on the lack of risk models, low consumer demand, high private premiums and, most of all, on the availability of relatively inexpensive government-sponsored insurance.
Some of those systemic issues have been resolved. For example, flood risk models have been developed. But most remain.
“The private flood insurance market is small, but the [flood protection] gap is huge,” affirmed Robert Muir-Wood, chief research officer for RMS, the Newark, Calif.-based catastrophe modeling company. “In the U.S. at present, there are 5.1 million households in the so-called ‘100-year flood zone’ as defined by the Federal Emergency Management Agency (FEMA),” he added. “Of those households, 65 percent do not have flood insurance,” which shows the extent of the potential business opportunities for the insurance industry, said Muir-Wood during a panel discussion at the reinsurance Rendez-Vous de Septembre (RVS).
During Hurricanes Harvey and Sandy, less than 20 percent of the houses that were flooded had flood insurance, he said, noting that these catastrophes have highlighted the enormous protection gap around U.S. flood peril. The insurance protection gap—or underinsurance—is defined as the value of assets at risk that are not covered by insurance when a catastrophic event occurs, either natural or manmade.
The National Association of Insurance Commissioners has found that half of U.S. flood losses occur outside the designated high-risk areas, yet only 1 percent of properties outside of the defined flood zones have flood insurance, according to a report on the protection gap published by Lloyd’s of London.
Part of the problem is that the government’s National Flood Insurance Program (NFIP), which is managed by FEMA, dominates the flood insurance market.
Indeed, Muir-Wood said that 95.5-96.5 percent of the flood policies are offered by the NFIP, which means that only 3.5-4.5 percent of all the residential flood policies in the U.S. are now covered in the private market.
Muir-Wood was joined on the RVS panel by Karl Jones, managing director, head of Catastrophe Analytics at Willis Re. They discussed the state of the private flood insurance market and some of the potential remedies to improve growth. The panel was moderated by Ben Brookes, vice president at RMS.
While a robust private insurance market has yet to develop in the U.S., there is growing interest on the part of insurers and reinsurers, looking to diversify their books by offering cover for the flood peril, agreed the panelists.
In his comments, Muir-Wood began with an historical perspective of catastrophe models that were developed after Hurricane Andrew in 1992. “Through the 1990s Andrew was the big poster child of what a hurricane looks like. Yet, we see in retrospect that Andrew was actually pretty unusual,” he said, explaining that the storm was dominated by wind impacts and had a low proportion of flood damage. “Over time, it’s become completely clear that flood causes a much more significant part of hurricane losses,” he added.
Muir-Wood pointed to events like Hurricane Katrina in 2005, Hurricane Sandy in 2012, Hurricane Harvey in 2016 and Hurricane Florence in 2018, where flooding caused significantly more than half the economic loss. (See related article, “Hurricane Florence Demonstrates the Perils of Being Underinsured/Uninsured, Says RMS.”)
“Actually, what this means from a modeling perspective is that we need tools that combine and integrate the wind loss, the storm surge loss, the inland flood loss and even the wave impacts on offshore platforms,” Muir-Wood said, stressing that by addressing these four hurricane perils, insurers can be more confident that all of the risk is being covered.
“If you write flood risk, deep inland in the U.S., it doesn’t have a strong correlation with hurricane. But if you write flood business in any of the coastal Gulf or eastern states, you will find that many of the top flood events are hurricanes.”
Robert Muir-Wood, RMS
As a result, an insurer needs to manage its book for the wind, storm surge and inland flooding all together, he said, noting that modeling technology has only recently advanced to the point where it’s able to combine all the perils. “Previously, [an insurer was] a bit in the dark about how to diversify its risk.”
Jones at Willis Re agreed, saying that flood has moved from being an unmodeled peril to one for which models are available—greatly assisted by advances in computing and technology.
Both Jones and Muir-Wood affirmed that flood is actually a highly predictable peril and that profitable risk selection is possible for insurers and reinsurers as long as they understand the models and determine the correct price for the risk.
“There are more elegant ways of putting this, but if your property sits on top of a hill, you’re not at risk. If you’re in lower-lying areas, there’s a flood risk,” said Jones. “You can delineate the flood plain and you can deal with that. We have the tools and the accuracy to be able to identify the hazard in a very detailed manner.”
Further, he added, the flood events of the last few years are helping insurers, reinsurers and modeling companies to be able to validate the models against the losses.
In addition, the information about the building exposure at risk is much more accurate than it has been in the past, said Jones, pointing to detailed data on properties, captured by insurers and brokers, which is brought together by modeling companies.
Muir-Wood emphasized how flood risk is arguably more of a definable peril than even hurricane and earthquake. He provided a technical explanation of modeling, which faces two classes of uncertainty: “epistemic” uncertainty and “aleatoric” uncertainty. Epistemic uncertainty is what you don’t know, while aleatoric uncertainty is the basic randomness associated with the risk, he said.
“When you think about hurricanes or earthquakes, there’s actually a lot of aleatoric uncertainty, because wind is a chaotic process. One house is hit by strong gusts, while the next house is missed by strong gusts. The peril can be very variable on a small scale, and it’s not possible to model that full variability,” Muir-Wood continued. He noted that the same applies to earthquake ground motion, which is chaotic and complicated.
“And actually flood…has a much lower level of aleatoric uncertainty, a much lower level of intrinsic variability, because flood heights are pretty consistent from one patch of land to the next,” he explained.
He said that means flood models have to contend with mostly epistemic uncertainty, for which information is needed in three areas in order to model flood effectively:
- Elevation data of the property.
- The existence of flood defenses.
- Information on what is happening in that property below the ground floor—i.e., does the property have a basement? If so, what is the basement used for? Is it occupied? Is it full of expensive artwork or something valuable?
By obtaining this data, flood risk modeling can become more accurate than even hurricane wind or earthquake loss modeling, Muir-Wood affirmed. “We’re in a better state to be able to really quantify the flood risk, and an insurer can write that business with confidence.”
So, if flood is such an insurable peril, then why is insurance penetration so low?
Muir-Wood said the slow growth of the private flood insurance market in the U.S. is mainly due to the existence of the NFIP. Jones agreed by saying that the U.S. government is “offering subsidized, cheaper insurance.”
“The rating schemes from the National Flood Insurance Program are proven not to be sufficient to build up enough reserves to pay for losses, which is why the scheme has had to borrow more than $30 billion altogether from the government to pay for its claims from Katrina, Sandy, from the Louisiana floods in 2016 and Harvey in 2017,” Muir-Wood added.
The coverage provided by NFIP creates a big challenge for the private market because if insurers have to compete against NFIP rates, then there are only certain niches in which a company can operate, “which is why the private market currently is so small in the U.S.,” he went on to say.
One such niche is available because the NFIP only offers coverage up to $250,000, “so if you want to buy excess of that, you need to go to the private market to buy coverage,” Muir-Wood said.
Insurers also are offering coverage outside the defined 100-year flood plain, where it’s easier to find modeled risk prices that are lower than the NFIP’s pricing of the flood-risk costs, he said.
The NFIP has indicated it plans to move to risk-based pricing, which is made possible for the first time because of the new modeling capabilities. However, Muir-Wood said that such a move would likely cause protests among people living in high-risk areas such as southern Florida.
“Until the NFIP’s pricing actually reflects the risk, it will be very hard for insurers to compete against it,” he continued.
Jones predicted that consumer demand will be driven when people feel the governmental safety net is reducing. Governments will look for support from the private sector to develop insurance solutions, he noted. “If you’re seeing a situation where that is under pressure, you’re going to see a natural opportunity in the market.”
How to Increase Penetration
The Lloyd’s report on the insurance protection gap noted that in the countries where flood risks are automatically included in insurance policies for households and businesses, penetration rates are generally higher. “For example, in the U.K., take-up rates for residential property insurance are more than 90 percent, while for home contents (for which flood damage insurance is not required by mortgage lenders), insurance penetration can drop to just 44 percent,” said the Lloyd’s report titled “A world at risk—Closing the insurance gap.”
Penetration rates are also high in other countries where flood risk is included in standard coverage, such as France and New Zealand, said the report. On the other hand, in countries where flood coverage is optional, such as Portugal and the U.S., insurance penetration is generally much lower, it continued.
In the U.S., flood insurance is only required by banks when you take out a mortgage if you live in a designated “flood zone,” according to Lloyd’s. However, in recent years, areas that haven’t been classified as being at risk have been hit hard by floods, which shows “the need for new analysis and mapping of risk,” said the report.
“In the U.K., Flood Re (established 2016) is not primarily ‘state-backed’ but instead risks are pooled among all insurers offering flood policies,” said the Lloyd’s report. “In an extreme case, the government would step in as the final guarantor. The U.K. also seeks, as many countries do, to implement an engineering solution to flooding,” the report added.
In his comments at the RVS, Jones noted that the U.K. has committed to risk-based pricing by 2039 for its Flood Re program. In addition, any property built from 2009 cannot be included in the program, so there is a disincentive to build properties in places where there is a potential flood risk. “So, the U.K. is starting to combine the forces that influence the market.”
Karl Jones, Willis Re
Brookes, vice president at RMS, the moderator of the panel, asked Muir-Wood and Jones if they think there are ways in which flood insurance might be mandated or required as part of a mortgage.
“In the U.S., there is a requirement that if you take out a mortgage, then you have to have to have flood insurance if you’re within the so-called 100-year flood zone. But one year after you take out a mortgage, your mortgage has probably gone to Fannie Mae or Freddie Mac to be packaged into a mortgage-backed security,” said Muir-Wood. As a result, the bank that issued the mortgage is no longer interested in the question of whether the property is protected by flood insurance, “and so, the lapse rate for flood insurance policies is about 25 percent per year,” he said.
There actually is no mechanism to ensure the flood insurance is persistent through time, he noted.
If the property is located just one meter outside the 100-year flood zone, insurance isn’t required “even though your flood risk may be pretty much the same,” said Muir-Wood.
“In California, you have this ridiculous situation where if you take out a mortgage you have to have flood insurance, but you don’t have to have earthquake insurance, so most people don’t bother with the earthquake insurance,” Muir-Wood continued.
Policymakers could actually take away the freedom not to have coverage and require people to purchase it, he suggested. Another possibility would be to change policy design, he affirmed, pointing to the example of Japan, where property insurance products for fire, wind and flood are offered that run with the life of the mortgage.
“Another reason for underinsurance against flooding is the prevalence of adverse selection in this market,” said the Lloyd’s report, explaining that people who buy flood insurance are those who know they are most at risk, which drives up price and reduces uptake in the coverage.
“It is also the case that many homeowners believe their property is insured by their regular home insurance policies, which is rarely the case,” it continued. “Awareness-raising is an important initiative to encourage insurance uptake.”