12 December 2012

Science, Politics and the "Hurricane Deductible"

The idea that coastal property owners should bear some of the risks of the exposure to hurricanes is not particularly controversial. However, implementing policies to align risks with economic incentives can be challenging, and "Hurricane" Sandy provides a vivid case study in the importance of institutions where science meets politics.

Some 18 states implement what is called a "hurricane deductible" as part of insurance policies. While a normal deductible (i.e., the amount the homeowner must pay in the event of a loss, before the insurance kicks in) for property damage might be set at $2,000, the "hurricane deductible" says that if the event causing the loss is a "hurricane" then the deductible is instead set at a much higher level, such as $25,000.

The "hurricane deductible" became important following Sandy because just about one hour before the storm made landfall, the National Hurricane Center re-categorized the storm from a "hurricane" to a "post-tropical cyclone." Because insurance is regulated at the state level, different states have different "triggers" for the application of the "hurricane deductible." Some of these triggers are tied to wind speed, some to the issuance of hurricane warnings by the NHC, some to the storm's categorization, and so on. New York, for example, does not have a single trigger for the state but allows each company to set up a trigger (the mish-mash of approaches can be seen here in PDF).

For a storm like Sandy the invocation of the "hurricane deductible" is a decision with tens of billions of dollars in consequences, as losses were spread over hundreds of thousands of homes. Either individual homeowners would bear these costs (if the deductibles were invoked) or insurance companies would (if they were not). Given the massive stakes, not surprisingly in the immediate aftermath of the storm politicians were quick to act.

In New Jersey Governor Chris Christie, a Republican, issued an executive order defining Sandy as a post-tropical cyclone, invoking the NHC re-categorization:
In light of the National Weather Service’s categorization of Sandy as a post-tropical storm, it shall be a violation of N.J.A.C. 11:2-42.7 for any insurer to apply a mandatory or optional hurricane deductible to the payment of claims for property damage attributable to Sandy.
US Senator Charles Schumer, a Democrat, sent a letter to NOAA (the parent agency of the NHC) reminding them of the political consequences of their storm categorization. He expalined why to a local radio station:
“Today, we’ve sent a letter to NOAA, the weather agency, as well as to the insurance companies that we’re looking over their shoulder. We want NOAA to keep this classified as a tropical storm and to save homeowners in New York and Long Island thousands of dollars and we don’t want the insurance companies to play any games.” 
A few years back, when politics-science issues were more fashionable, there might have been outrage from scientists and other observers at the idea of a US Senator "looking over the shoulder" of a federal science agency and telling it how to make a scientific judgment. But I digress.

Issues associated with the "hurricane deductible" likely played a role in NOAA's immediate reversal in setting up an assessment team to evaluate the agency's performanceon Sandy. NOAA initially established an assessment team, to be co-chaired by Mike Smith, a widely-respected and accomplished private sector meteorologist who has also been critical of NOAA at times in the past. Involving an outsider as co-chair made good sense from the standpoint of the credibility of the assessment. However, NOAA terminated the assessment almost as soon as it was created.

In his initial work on the assessment as co-chair, Smith had identified some key questions to investigate:
  • Was there a decision not to call Sandy a "hurricane" regardless of its meteorological characteristics? If this decision was made, was it made Friday (October 26th) or Saturday morning? If so, who made the decision and why?
  • Was this decision the reason hurricane warnings, in spite of a large and dangerous hurricane moving toward the coast, were never issued?
  • Given that an obvious large and powerful hurricane was headed for the U.S. coast, why wasn't that decision reconsidered? For example, Barry Myers, the CEO of AccuWeather, urged (on the AccuWeather.com website) the immediate issuance of hurricane warnings about eight hours before landfall. Others also urged the lack of hurricane warnings to be reconsidered.
Immediately after the termination of the assessment, Rep. Paul Broun (R-GA), Chairman of the House Subcommittee on Investigations and Oversight, sent NOAA a letter asking for a range of specific details about the curious decision (letter here in PDF, the replies from NOAA are due back to Rep. Broun this Friday).
For NOAA's part, they have since reconstituted a team to investigate their performance on Sandy comprised only of government employees, and chaired by a non-meteorologist. Smith was purged as co-chair and so too were any explicit questions about Sandy's meteorological status at landfall, prompting Smith to write yesterday:
They are ignoring the elephant in the room. Was Sandy a hurricane at landfall?
All of this still matters because the NHC still has not rendered a final determination on Sandy's actual status at landfall. Such determinations are always re-evaluated in the months after a hurricane season, when there is more time and a break from the pressures of an operational forecasting environment. The NHC explained this at the time it announced the re-categorization just before Sandy's landfall:
While I have every confidence in the scientists at NHC, can you imagine the consequences if they were to re-categorize Sandy as hurricane at landfall? The implications would be enormous and the political fallout immense.

We have here a situation where state and federal policy makers have already made judgments about what the science should say (for a longer list of state actions see this) yet the actual science is not yet completely in. This situation illustrates that the NHC is not well structured to play a key role in regulatory-type decision making. This is not the fault of NOAA of the NHC, but the policy makers who put NHC in such a position via policy. Accompanying the questions about Sandy's status at landfall is the messy prospect of NOAA appearing as if to cancel a partially-independent assessment for fear of the questions being asked, and replacing it with a plain-vanilla committee with a plain-vanilla mandate.

I have no opinion on (or much interest in, actually) the substantive judgment about whether Sandy should have triggered "hurricane deductibles" or not -- I have heard from experts arguments on both sides. The larger point that I am focused on is the process through which hurricane risks are translated into homeowner incentives as a case study in the institutional design of processes meant to bring science into policy making. Unfortunately, it appears that in this case weak policy design at the science-policy interface means that the role of science is far less than might have been hoped for, with political considerations instead driving the process. Worse still, NOAA faces the prospect of a scandal of some sorts based on its efforts to escape the political battles. All around, not good.


  1. It looks to me like NHC was forecasting post/extra-tropical way before there was a possible mid-Atlantic landfall.

  2. So, the choice was between a public (i.e. political) and a private crisis. If risk and associated costs can be selectively marginalized and redistributed through a covert shift to the private sector, generally to insurance companies, and specifically to policy holders, what risk model do they follow when the outcome is predicated on political expediency?

    It's interesting that some interests wanted the event to be classified as a hurricane, in fact a "super" hurricane, while others wanted it to be downgraded. It is further noteworthy that within each class, there are interests which change positions when the political liability changes, and they manage to escape scrutiny of this dynamic.

    As with the L'Aquila Earthquake Trial, this event seems to favor treating symptoms, while ignoring causes.

  3. Shumer went after Allstate Insurance's decision in 2006 not to sell hurricane insurance. Schumer claimed Allstate was citing "bogus hurricane risks" to boost their profits. Shumer went further stating:
    "The hurricane data hardly demonstrates a risk,” Schumer said. “It already costs a massive amount of money to buy a home in New York City, LI and Westchester. Allowing homeowner insurance prices to spike will only make that situation more unbearable. The hard working homeowners of Westchester , Long Island and New York City should not be made pawns by an rapacious insurance industry looking to manufacture a false crisis to pressure designed to pressure the federal government to create a taxpayer-financed disaster bailout system.”

    There is a FEMA report that has not been allowed to see the light of day looking at the impacts of a Cat 4 strike on Manhattan--modeled storm was called RUDY. Damages were estimated at one half trillion dollars---a value too high for political consumption

  4. Rules for triggering a politics vs. science debate

    (1) The politician says something contrary to the scientific consensus, or exerts pressure on scientists to change scientific conclusions.

    (2) The politician is a Republican (or outside the US, 'conservative' or 'classical liberal').

    Condition (2) was not triggered by Senator Schumer.

  5. Let's face facts.

    Living near the water is a symbol of economic and political influence.

    Whether it's the Kennedy Family Estate at Hyannis Port or the Bush Family estate at Kennebunkport.

    Since waterfront property tends to be owned by the politically well connected then the politics of who will cover the losses regardless of what the official policy is seems to me to be a matter of certainty.

  6. 5. Harrywr2

    This wouldn't be typical of the Jersey Shore. The way many people afford them is by using them as rental properties. Many are very small homes. The politician are motivated by not only the owners but also the renters. That's not to say there aren't some wealthy communities.

  7. President Obama proclaimed Sandy a "hurricane" at landfall. www.whitehouse.gov/the-press-office/2012/12/07/executive-order-establishing-hurricane-sandy-rebuilding-task-force

  8. "The idea that coastal property owners should bear some of the risks of the exposure to hurricanes is not particularly controversial."

    I'd like to offer a different perspective (apologies for any heart attacks/strokes given to libertarians reading this).

    Suppose the French decided the Louisiana Purchase was a big mistake, and invaded and sacked New Orleans, killing several thousand people.

    We would say that the U.S. military should defend New Orleans from the French, right?

    Sooo...if a hurricane should attack New Orleans...or New York, or Miami, or Tampa, or Charleston, or Virginia Beach, or anywhere else on the Gulf or East coasts...why not have...the Army Corps of Engineers/FEMA protect the cities?

    And how could they protect the cities from hurricanes? Well, wind and inland flooding are issues I'd rather hold off addressing, but storm surge is best handled by...a portable storm surge protection system! And what's that? Google "portable storm surge protection system", or go here:


    Virtually all of the damage to New Orleans from Hurricane Katrina was from flooding caused by storm surge (I'll guess $80 billion worth). And much of the damage caused by Superstorm Sandy was caused by storm surge (I'll guess $40 billion worth). So that's like $120 billion in storm surge damage just from those two storms.

    Developing a portable storm surge protection system that could protect any city on any coast of the U.S. in a few days' notice would probably be a very good investment for the U.S. government, in terms of avoiding future costs. One could say that the federal government shouldn't pay those costs anyway, and should just let the cities flood, without paying a penny to rebuild. But I think we know that's never going to happen. So the real choices are: 1) the federal government does nothing to develop a portable storm surge protection system, and pays big bucks in the future, or 2) the federal government develops a portable storm surge protection system and potentially saves much more money than the cost of development.

  9. It does seem rather strange that characteristics of the storm other than its actual destructive power should determine who will be liable for the damages.

    It's also the case, however, that the underlying logic is not unreasonable. In general hurricanes have the characteristic of being large storms that cause highly correlated damages, and (in the Northeast) of being reasonably rare. Historical decisions about land use and insurance are based on these facts, and the "hurricane deductible" kept the prices low for most people, at the expense of the significant but finite population who would actually be greatly harmed when the deduction was activated.

    The problem that we face now is the realistic possibility that there will be more strong(and possibly stronger) storms, with vulnerability increasing due to sea-level rise. Deciding whether or not they have to be hurricanes, or designated as hurricanes, does not change the fact that the risk profile appears to be changing, and someone is going to have to pay to insure those risks.

    People who live in frequent hurricane zones (or earthquake zones) pay extra for "physical" insurance as well - construction codes differ, windows get boarded up, etc. It's easy (to use the terms somewhat loosely) for property owners to greatly reduce the odds of getting damage over $2000 from even a pretty strong storm. But retrofitting the Northeast to those standards isn't a very easy task...

  10. To have a policy with very different outcomes based on a single threshold (below some storm level you pay $2,000 above that level $25,000) is going to bring a lot of attention when the threshold is neared. Would a policy that offered a continuous, monotonically-increasing deductible be more palatable? As a clearly-defined risk measure increases (max wind speed at nearest official site), so does the deductible. Insurance companies could compete as to the steepness of the risk-deductible curve. The incentives to fight for a specific outcome (above or below threshold) would disappear or be greatly reduced.

  11. I saw this article a couple of weeks ago. In view of your article, this proposed change certainly seems relevant:


    The National Hurricane Center confirmed Wednesday proposed changes to the definition of a hurricane warning.

    A report issued at an annual NOAA hurricane meeting in Miami explicitly stated the NHC will now be able to issue a hurricane warning on a post-tropical cyclone.

    "A proposal was raised during the NOAA Hurricane Conference last week for NWS to have the option to issue hurricane and tropical storm watches and warnings for post-tropical cyclones that threaten life and property" said Chris Vaccaro, spokesman for the National Weather Service. "This is one step in the process required before any proposed change to operational products becomes final. As part of our review of the 2012 hurricane season, including the Sandy service assessment, we will review all policies and changes through the existing and established process."

    The changes come on the heels of an investigation into Hurricane Sandy, which became a post-tropical storm before making landfall in the Northeast. Tropical storm and hurricane warnings were not issued anywhere on the East Coast north of North Carolina, and a NOAA and FEMA investigation is underway to study the handling of the events surrounding Sandy, which is a customary occurrence for major weather events.