Insurers will soon have better analytical tools for predicting the damages caused by widespread inland floods in the U.S., allowing them to price this risk more accurately and, eventually, transfer more of this risk to private investors through insurance-linked securities.
Currently the U.S. government subsidizes residential policies through the National Flood Insurance Program (NFIP). Since the prices of these policies inadequately reflect flood risk, there is not much incentive for private insurers to compete. Private insurers’ commercial policies typically cover damages for a range of natural perils including flood.
Better models should allow insurers to better price residential and commercial flood insurance. It could also give private investors the confidence to take on some of this risk as they did with hurricanes, earthquakes, severe thunderstorms and even terrorism, following the arrival of analytical models from the three main modeling firms, particularly if the government decides to price this risk more competitively.
“Inland flooding in the U.S. was one of the last major perils that had not yet been modeled,” said Dr. Jayanta Guin, executive vice president responsible for strategic management of the research and modeling group of AIR Worldwide. “We undertook this challenge three years ago and have a large team working on the model, which will be released in the August, September timeframe.”
The two other major risk modeling firms, RMS and EQECAT, plan to release comprehensive risk-probability models for inland flooding next year, although elements of those models have already been put to use, especially in the commercial realm. More generally, Guin said, developing a model for U.S. inland flooding has taken longer since “modeling floods is more challenging than many of the other perils because you’re talking about having to model risk at a very high resolution for the entire [country], and that’s a tall order.”
Hurricane Sandy, which struck the East Coast in late October 2012, resulted in private insurers paying $10 billion in claims associated with flood losses. Tom Larsen, senior vice president and product architect for CoreLogic EQECAT, said the hurricane starkly revealed that insurers have significant exposure to flood-related losses, a realization coinciding with increasingly sophisticated catastrophic risk models.
“In the past, flood losses may have appeared more uninsurable,” Larsen said. “But now there’s greater confidence that today’s modeling can enable insurers to offer policies that are priced accurately so they’ll have the resources to cover claims.”
The probabilistic inland-flood models are arriving as demand for exposure to catastrophic risk is increasing, reflected by aggressive spreads both for reinsurance and ILS as well as record ILS volume so far this year. Catastrophe risk in general is highly prized among capital market investors because it is viewed as uncorrelated with the financial markets. Modeling the probability and cost of inland flooding more accurately will improve understanding of the peril for a wider variety of market participants.
“Models are useful tools when underwriting [natural catastrophe risk]. Most investors already take a range of natural catastrophe risk which is not modeled or not modeled properly, and any closing of this gap helps,” said Luca Albertini, chief executive of Leadenhall Capital Partners, a fund specializing in catastrophic risk investments.
William Dubinsky, head of ILS at Willis Capital Markets & Advisory, a unit of insurance broker Willis Group, noted that commercial insurers have typically sought all-natural-perils coverage from their reinsurers, whereas as capital-market investors in catastrophic risk have focused on more specific “named” perils and sometimes single perils. That’s been changing, however, as cat bonds have increasingly covered a wider variety of perils.
Long-time cat bond issuer USAA, for example, recently launched the Residential Reinsurance 2014 Ltd. (Series 2014-1), which covers a variety of perils including earthquakes, severe thunderstorms, and wildfires. It also covers volcanic eruption and meteorite impact, in what appears to be the first time those perils have appeared in a cat bond.
“That kind of unmodeled risk has moved into deals at the margin”, Dubinsky said, adding that introducing broader inland-flood models should facilitate further migration. “It should benefit the commercial insurers in the U.S. because they will have more quantifiable risks to pass on to their reinsurers in the capital markets, and that should also help insurers price policies more appropriately on the front end.”
Dubinsky noted that the final models could end up showing that there’s either a lot more or a lot less risk related to inland flooding than market participants have realized. If the models instead arrive at similar results to insurers’ current analysis, the quantification should make the market more efficient and coverage more affordable and available to organizations seeking insurance.
“If an insurance company has a [flood] sublimit now of 25% of its overall limit, that sublimit could grow and the insurer will be able to get more coverage, and that’s got to be a good thing for insurers and potentially for policyholders,” he said.
Insurers providing policies to homeowners for hurricane related damages have found enormous success in reinsuring this risk via the cat bond market. Florida’s Citizens Property Insurance, for example, recently issued a record $1.5 billion in bonds through its Everglades Re special purpose vehicle, and the single-B rated bonds priced for a 7.5% coupon.
That success may not readily transfer to inland-flood risk, at least right away. The Biggert-Waters Reform Act of 2012 established a schedule for allowing premiums on flood policies to increase and more closely reflect their actual risk, in part to incentivize private insurers to take on part of that risk. Concerns about the impact on homeowners in a still-weak economy, however, prompted Congress to pass the Homeowner Flood Insurance Affordability Act in March, delaying many of the first law’s provisions.
“Given the ridiculous punting by our federal government on the new flood rate regime, a lot of the anticipated private-market demand is now unlikely to materialize,” said the principal at one of the largest cat-bond funds.
Dubinsky said the law doesn’t eliminate the potential for insurers and catastrophic-risk investors to take on some of this risk, but it likely slows the process. He added that the NFIP, tasked with finding ways to transfer more risk to the private market, may come up with other solutions to access reinsurance and capital markets capacity for its own exposures.
Guin said the team developing AIR Worldwide’s the inland-flood model comprises meteorologists, hydrologists, and a variety of other experts who have mapped 1.4 million miles of river networks and created a digital terrain model for the country to determine how water would flow in a flood event. The model also incorporated variables such as ground saturation, which can turn even relatively minor rain storms into flood events; and accumulations of snow in the upper Midwest, for example, that can impact states thousands of miles down the Mississippi River months later.
The calculations must be precise, not only to gauge where water is likely to flow but to estimate damage. Guin noted that if the wind speed used in a hurricane model is a few miles an hour less than the hurricane’s actual wind speed, the estimated damage will resemble the actual damage. No so for flood damage, since whether water reaches the property or not makes all the difference from a loss perspective.
“Whether one foot or three feet of water gets into the house, that causes very substantial damage,” Guin said.
Late last year EQECAT was acquired by Corelogic, which has a longstanding hazard-based inland flood model that estimates flood levels and attaches a risk score to specific properties. The new model, incorporating EQECAT’s probabilistic methodology, will enable determining the probability of flooding over multiple properties and quantify an average annual loss.
Following Hurricane Sandy, “There was the realization that the risk could be a lot larger than anticipated, and if insurers wanted to be more aggressive offering these types of flood policies, they needed to establish management controls, and that’s what the catastrophe model supports,” Larsen said. “If this peril takes the path of others, first a credible model will establish reasonable prices, then the likely progression will be toward ILS such collateralized reinsurance and Rule 144A cat bonds.”