The Flood/Wind Divide

Solving the Flood/Wind ProblemAfter Hurricane Losses:
The Case for a Standardized Loss Allocation System

By Scott Richardson and Eli Lehrer

Introduction

There isn’t enough data or experience relating to hurricane losses to make any certain determinations as to whether a given loss resulted from wind or water.In the aftermath of a hurricane landfall, there often isn’t enough data or experience relating to the losses to make known for certain whether a given loss resulted from wind or water. Each time an indeterminate loss situation (total home destruction) takes place, protracted legal battles and custom-made agreements are currently the only way to resolve disputes over who should pay for losses.

This paper describes the problem of indeterminate losses, outlines the shortcomings implicit in two other frequently proposed solutions, and then describes a better solution: a standardized loss allocation system. Under such a system, federal and state regulators and private parties would agree in advance to distribute losses in specific ways. Rather than adjusting claims individually, a standardized loss allocation system would, in the case of a total loss, distribute claims based on a formula between wind insurers and the National Flood Insurance Program (NFIP).

1.  The Problem of Indeterminate Loss

Hurricanes rank among nature’s most destructive phenomena. High winds, wind-driven debris, and the storm surges that follow hurricane landfall can do enormous damage to everything in their paths. In many cases, over areas of several miles, all above-ground traces of buildings can be torn apart, leaving nothing but foundation slabs in place.

The current U.S. system for paying insurance claims can cause enormous problems for consumers, insurers, and regulators.Because they combine high winds with storm surge, hurricanes cause both wind and water-related losses in very close proximity to one another. When nothing above remains of a structure and both wind and storm surge were sufficient to cause its total destruction, claims adjusters, consumers, and insurers currently have no practical way to determine how to distribute losses between NFIP, which provides nearly all flood coverage, and private insurers, who cover wind and related damage. (Adjusters can make such claims on the basis of flood-lines, the point to which water rises during a flood, when a structure remains essentially intact.)

When a home is totally destroyed, there’s no simple way to resolve disputes over who should pay for losses. When this happens, the current U.S. system for paying insurance claims can cause enormous problems for consumers, insurers, and regulators.

The problems are most serious when people do not have flood insurance and therefore try to get insurers that sell wind insurance policies to cover damages that may have resulted from flood. Insurers often refuse these claims and say flooding (which they do not cover) caused the damage, leaving consumers with little recourse besides litigation. This, in turn, can lead to enormous and protracted legal battles among insurers, consumers, and regulators. 

The country’s largest homeowners’ insurer spent about 18 months in court wrangling with Mississippi officials and consumers over the payment of claims related to a single major hurricane before reaching a settlement that left few of the involved parties satisfied. Even in cases where a flood-line exists, homeowners who experience flood and wind damage but lack flood insurance will understandably attribute as much damage to wind as possible. Furthermore, since different insurers determine losses differently (even for houses right next door to one another), many individuals feel cheated.

Insurers lose out, too: Those that are less generous than their peers, even when they have good actuarial evidence to support their decisions, can take a major reputation hit. Likewise, insurers spend significant amounts on court costs even when they are fully within their rights to deny claims for flood damage they specifically excluded. Regulators, finally, end up mediating disputes where nobody can win: Some consumers will almost always feel cheated and some insurers will almost always end up paying far more than they believe actual loss data justify. In short, the system doesn’t work right now.

2.  Unsatisfactory Solutions to Indeterminate Loss

Two solutions frequently proposed to address the problem of indeterminate loss – a government takeover of the wind insurance business and privatization of flood insurance – have attracted support from certain quarters. Neither proposed solution, however, appears practicable at this time.

A.  Government Takeover of Wind Insurance

A proposed government takeover of the private wind insurance system would be very expensive and violates sound insurance principles.A proposed government takeover of the private wind insurance system, advocated by former Rep. Gene Taylor (D-MS), has passed the U.S. House of Representatives but would be very expensive and violates sound insurance principles.

The case for federal government assumption of wind risk is that residual insurance markets already take it over to some extent in the most hurricane-prone states, and unifying responsibility for flood and wind risk with the federal government would end any dispute over who would pay for damage. In the wake of Hurricane Katrina, when several members of Congress lost their own homes, a federal wind insurance takeover passed the House of Representatives in 2007 by a significant margin despite concerns that it would cost several hundred million dollars in order to be “actuarially adequate.” The legislation faced a veto threat from President George W. Bush and went down to an overwhelming 73–19 defeat in the Senate.

Whatever its political prospects, a federal government takeover of wind insurance remainsunworkable without a very significant taxpayer subsidy. Here’s why: “Primary” insurance companies such as Allstate, Nationwide, and CNA buy insurance of their own, called reinsurance, to manage and pool risk. The reinsurers that sell this type of coverage invariably operate around the globe. A large reinsurer might simultaneously underwrite the risk of an industrial accident in Japan, a flood in the United Kingdom, a hurricane in Florida, and a cyclone in Australia. Since there’s almost no chance that of these events will happen at the same time, the reinsurer can profit from the premiums it earns on one type of coverage even when it pays out big claims on another.

All other things being equal, a broader, more diversified pool results in lower premiums. Consolidating wind insurance in a federal entity would narrow the pool by concentrating risk in the United States under the aegis of federal guarantees. To break even, then, the government would have to charge higher premiums than the private sector for any coverage or guarantee it provides. If it hoped to sell any coverage at all, the government would have to underprice the coverage and thus leave taxpayers with very large liabilities.

Given the enormous liabilities such a program would impose on taxpayers around the nation for the sake of the relatively few who live near the coasts, in addition to the displacement of private insurers, having the government take over the wind insurance market would be terribly unwise.

B.  Privatization of Flood Insurance

No private primary insurers are willing to write flood insurance without sweeping, permanent, state-level modifications to the overall regulatory structure in the United States.Unfortunately, outright privatization is not a viable option at this time, either.

In principle, we favor having the private sector take over the provision of flood insurance and ending the National Flood Insurance Program. Doing so would leave all risk with private carriers and end all separation of flood and wind claims. Private companies in the United States already write flood coverage for businesses and individual homes above the coverage limits of the National Flood Insurance Program. Many other developed countries, including Germany and the United Kingdom, manage flood risk almost entirely through the private sector.

That said, we do not believe full privatization in the United States is achievable at the current time. This has been made clear in serious conversations with top-level representatives of nearly all of the nation’s large primary insurers and many reinsurers over the past year, in which one of us found that no private primary insurers are willing to write flood insurance without sweeping, permanent, state-level modifications to the overall regulatory structure in the United States. Some say it is not worth the investment, and others express concerns about the quality of underlying flood maps. Still others say state regulators would not allow them to charge actuarially adequate rates for such insurance. Even companies writing flood insurance in other countries have made it clear they have no desire to do so in the United States.

With no primary insurance companies willing to write flood insurance for ordinary houses in the United States, we regretfully conclude it’s simply not credible or politically possible to propose that flood coverage be turned entirely over to the private sector at this time, although we believe that should remain the long-term goal.

C.  Bottom Line

Thus we reach a simple bottom line: Adding wind coverage to the National Flood Insurance Program may be politically practical in some circumstances but is not wise, while dropping the NFIP program entirely may be wise but is not practical. Thus, it is necessary to consider another solution: a standardized loss allocation system.

3.  Standardized Loss Allocation

A Standardized Loss Allocation approach would solve almost all of the problems implicit in the current system and could potentially save money for both NFIP and private insurers. Under a Standardized Loss Allocation (SLA) system, insurers, NFIP, and state-run or state-mandated residual insurance markets such as wind pools and Fair Access to Insurance Requirements (FAIR) plans would agree in advance to distribute liability based on standardized loss allocations models. The sizes of the distributions would be based on extrinsic, meteorological evidence.

This approach would solve almost all of the problems implicit in the current system and could potentially save money for both NFIP and private insurers. Implementing it properly, however, will require the collection of more meteorological data. The next few paragraphs describe how it would work, how it would solve existing problems, what public policy changes will be needed to develop it, and how it could be made both universal and voluntary.

A.  How an SLA System Would Work

An SLA system would begin by establishing a model for assessing the damage resulting from each major storm based on four major inputs: direct observations, information about building characteristics, indirect aerial observations of storm surge, and widely used storm models.
Direct observations are what they sound like: direct information, collected by scientific instruments about wind speed and storm surge. Information about building characteristics – particularly elevation information that would provide input as to the consequences of storm surge – is not always included in current insurance underwriting standards but is necessary in certain cases to determine the exact damage resulting from storm surge.

Aerial observations collected during and immediately after a storm (as well as satellite imagery collected during a storm) could provide additional information on storm surge since few instruments can withstand the most severe storm surge. Finally, a wide range of storm modeling techniques would work together to build a model that could, ideally, provide a very good estimate of the extent to which wind or rain caused the damage to a given house. The National Oceanic and Atmospheric Administration and several private-sector modeling firms already have created such models; building a standardized national one would not require additional work.

For those participating in the system, the model essentially would end litigation, assure payments to consumers, shore up NFIP finances, and provide certainty to NFIP itself.

B.  An SLA System Would Work Better

An SLA system would work better than the current system or other proposed solutions, too. For those participating in the system, the model essentially would end litigation, assure payments to consumers, shore up NFIP finances, and provide certainty to NFIP itself. With payments determined by formula alone, there would be little to litigate: The use of the model would be written into policy language for both NFIP and private-sector coverage, and anyone challenging it in court would have to present expert scientific testimony indicating why the model was invalid. If anyone went through the trouble to do this, it would probably help improve the system by pointing out flaws in the models.

Customers, even those who live in flooded areas but lack flood insurance, generally would be assured of at least some payments rather than outright claims denial. NFIP itself, likewise, would rarely if ever face the prospect of having to pay the full cost of damage caused in part by wind. Insurers, finally, would have the same assurance, that they would never have to pay total losses when flood has done part of the damage.

C.  Improving Data Collection and Public Policy

Although the scientific instruments and mathematical techniques needed to develop a standardized loss allocation model already exist, the data aren’t all there. The current Aviation Weather Observation System and network of Automated Surface Observing Stations (known jointly as AWOS/ASOS) are not extensive or robust enough to deliver the data necessary to create an unimpeachably reliable and accurate model.

Even at wind speeds of just over 40 miles per hour (near the minimum needed to be considered a major “named” storm), nearly a quarter of all observing points fail. As wind speeds increase, even more stations fail: 58 percent once wind speeds hit 55 miles per hour, and almost three-quarters at 60 mph. The current data collection network just isn’t good enough to provide the information needed to create an unimpeachable model. 

For an SLA system to work, data collection will have to be improved. One piece of pending legislation, the Consumer Option for an Alternative System to Allocate Losses (COASTAL) Act, introduced by Sen. Roger Wicker (R-MS), would have the administrator of the National Oceanographic and Atmospheric Administration form partnerships with private entities and other federal agencies to place new sensors on federal government property. It also envisions paying some or all of the costs for keeping an expanded network in place by leasing space on new observation posts (many of them on existing federal property) to mobile phone providers, other weather-related businesses, and anyone else who has need for such access.

This particular funding approach certainly has promise, but Congress, the Federal Emergency Management Agency (which oversees the flood program), NOAA, and state agencies should keep their minds open to other approaches as well. Some specific options that may be worth considering include assessing fees on insurers that want to take part in the system, adding fees to NFIP policies that take part in the system, imposing a onetime assessment on NFIP policies to pay for the costs of developing the network, and even directly appropriating existing funds to pay for the deployment of the network, although doing so would be very difficult in the current budget climate.

Although we’re convinced that an SLA system will work better for most people involved, there’s no advantage in forcing everyone to change right away the way they do business.

D.  Making the System Universal but Voluntary

Once the data are available and a model is developed, participation in the system should be voluntary for all parties and available to all parties – nobody should be required to participate, but everyone who wants to should be able to. Although we’re convinced that an SLA system will work better for most people involved, there’s no advantage in forcing everyone to change right away the way they do business.

To facilitate universal but optional participation, NFIP should participate in an SLA model for all policies written under the Write Your Own program and not participate in it for policies written under NFIP Direct. This will provide a choice to all consumers, insurers, and state residual insurance market mechanisms. Individuals who want to opt out of the SLA program can purchase a direct policy, since nearly all independent insurance agents in the country will write them, and thereby get the same flood coverage at the same price as they would in the private program. Private insurers and residual insurance market mechanisms that don’t want to take part can simply leave (or never join) the Write Your Own program. In short, it seems possible to make SLA-based policies an available option for almost everyone without mandating them for anyone,

4.  Conclusion

A Standardized Loss Allocation model could solve the indeterminate loss problem implicit in the current loss allocation system for flood insurance in the United States. It is not a final or perfect solution to the problem. In the long term, we believe private companies should underwrite all or almost all flood risk. Given the current unwillingness of primary insurers to do so, however, it seems likely that any system will ultimately rely on reinsurance provided either by the government or by the private sector.

In a system of mixed insurance providers, risk distribution between wind and non-wind losses would still be required, and an SLA system allows that while being consistent with the desired movement toward a greater private-sector role. Even if one desires to retain the overall flood insurance status quo, an SLA system promises to be less expensive, more certain, and more efficient than any currently available alternative.

###

Scott H. Richardson is the former director of the South Carolina Department of Insurance and currently a partner at Richardson and Ritchie Consulting. During his tenure at the department from 2007 until 2011, Richardson oversaw the implementation of a successful effort to bring private insurers back to South Carolina’s coastal market.

Eli Lehrer is vice president for Washington, DC operations at The Heartland Institute, overseeing Heartland’s Washington, DC office and field offices in Tallahassee, Florida and Austin, Texas. He is also national director of Heartland’s Center on Finance, Insurance, and Real Estate, which coordinates Heartland’s work on issues relating to insurance, risk, and credit markets.