Thursday, 6 March 2014

EQC and flood risk

As we're still awaiting the Treasury review of the EQC legislation, which will be followed by proposed legislation, now seems the time to be thinking about premiums, risk adjustment, and flood damage.

As reminder, EQC provides cover for natural disasters like flooding or earthquakes. EQC coverage is compulsory when you're taking out home insurance; it's a levy attached to your insurance policy providing natural disaster coverage.

EQC coverage, as it stands now, has a few issues apart from the one about which I bang on continually here. The main one for present purposes: failure to risk-adjust premiums.

In a perfect world, we'd have perfect risk adjustment of EQC premiums, in the way I'd set out here a couple months ago:
Wouldn't it make more sense simply to have EQC set risk adjusted insurance premiums? Zoning changes mandating higher floor levels only really apply on new builds or, potentially, on substantial-enough building redevelopments. While new buildings will then get the higher floor levels, some older ones will have replacement or refurbishment delayed because the relative cost of a new building's gone up. If EQC set actuarially fair rates for their disaster insurance, the distortion would be gone. We'd also then avoid all of the fights and rent-seeking that will result when Council starts deciding where people will be allowed to build in future due to flood risk, arguments about how seriously we should take the upper-limit projections on sea level changes, and the like. Let the insurers set the premiums, then let individuals sort out whether they like current beachside property. And let EQC's premiums basically reflect the incremental effects of insuring different types of properties on EQC's reinsurance costs. If Swiss Re won't provide reinsurance for my house except at additional charge, I should be paying the costs of that. Where EQC's trying to minimise its reinsurance costs and where the international reinsurance markets are at least somewhat competitive, this knocks the political fights around global warming out of the mix.

Tim Harford comments usefully on distortions caused when government subsidises living on flood plains. EQC premiums are scaled by value at risk, but not by the likelihood of adverse events. The country varies in seismic and flood risk; buildings vary in robustness to those risks. But nobody pays more than $150 per year for their EQC cover. While you might think the distortion can't be that big as EQC only covers the first $100,000 in damage to your house, with insurers charging actuarially fair premiums taking on the bigger part, EQC also covers land remediation.
Now suppose we add in that the Government could be worried that actuarially fair rates might have some folks abandon taking out insurance altogether. For some now very low-lying parts of Christchurch, subject to regular flooding, it's not inconceivable that risk-adjusted premiums would be sufficiently high that everybody in the neighbourhood drops their insurance coverage* and hopes for a bailout should a flood happen. Then the government's got all of the payout risk but none of the premium revenues.

Now that could be an argument for setting some upper bound on the potential risk-adjustment, and especially for lower-value homes in lower-income neighbourhoods, but it isn't really an argument for having no risk adjustment. And, as I understand things, EQC premiums scale only with the value insured, not with the risk.
EQCover costs 15 cents (+ GST) for every $100 of home or contents fire insurance that you have. You pay this amount to your private insurance company, who pass it on to EQC.
The most you can pay, per year, for one home and its contents is $180 (+ GST). This would give you the maximum cover of $100,000 (+ GST) for your home, $20,000 (+ GST) for contents, and cover for your insured residential land. This amount of insurance is available for each event of natural disaster damage.
This has a couple of effects:

  • Because the maximum EQC payout is capped at $100,000, the maximum premium is capped at $180. But consider two houses. The first is worth $100,000; the second is worth $1,000,000. Both purchase insurance and both pay the maximum levy to EQC. An earthquake or flood hits both of them; repair costs are proportionate to the value of the house. Say it costs half the value of the house to fix it. So the first one gets $50k in repairs; the second gets $100k, with the rest paid for by the private insurer. The insurance scheme is then somewhat regressive: where we expect greater repair or replacement costs for more expensive homes for any given-sized event, and where the premiums cap out in the way they do, more expensive homes effectively get more coverage for a premium of a given size.
  • Because there is zero risk adjustment other than the bit that comes through the value of the house, people who live in safe areas provide a large subsidy to those who live in dangerous areas. And then all of the distortions Harford pointed out apply.
Why not start risk-adjusting the EQC premiums? Even just putting a small surcharge on spots known to be flood-prone would be a start.

* If their mortgages are paid off: the banks kinda make you have insurance if you have a mortgage.


  1. In the example you give ($100,000 house versus Million dollar house) the bulk of the total payments come from private insurer - and I have always assumed the private insurance payments were risk adjusted. To the extent that if what I heard on TV last night about the amount of flooding now happening in parts of ChCh are correct, insurance companies will soon stop insuring certain parts of the city.

    As an aside, governments of both hues have refused to give any substantial rebuild type assistance to uninsured victims of recent floods. People on the Manawatu flood plane found this out not so long ago. Of course, natural disasters generate large amounts of sympathy - and donated cash for victims. But the money is not always all spent - because the insured don't really need much and the govt does not want to encourage non-insuring, so they dont get any (except from private charities). Often the money raised is not all spent. I have often wondered were the surplus goes, and did all the money raised for ChCh relief get spent?

  2. When we screw up the pricing of insurance, we wind up with a bunch of second-best regulatory hacks to try to push things back to what people would likely have done if they'd faced fair insurance premiums when they were deciding where to build. We get it here, we get it in all the dumb social cost stuff around the public health system. Where we can push towards getting rational insurance premiums first, that's far better.

  3. Good questions, Rob.

    I tend to like a variant on 2, though with a bit of 1 built-in. It isn't a fairness argument though.

    We all have been paying the EQC levy for insurance that includes land remediation. If your house is now a half-meter lower than it used to be, and as consequence suffers greater flood risk, they really should be fixing this or paying you out for the reduced value of the house.

    If they put your house to its ex ante level, then your flood risk is no higher than before and you don't face differentially higher premiums in a risk-adjusted premiums world.

    If they give you the cash lump sum instead, because the reduction in value is less than the cost of fixing it, and you then decide that it's too expensive to live there because of the insurance cost, the value of that site drops (for which you've been compensated in the cash settlement) and somebody else buys the land, deeming that the ongoing higher premiums to live there are worth it.

  4. I wonder RobS, if there are not 2 things that need thinking about. One is the compensation for the flooding - and here I see that there is a need for compensation as per the insurance contract. But, as you say, it appears the topography (or the structure of near the surface aquifers) has changed - in which case should they go as well. Things like streets, drains, etc are not covered by householder insurance yet householder decisions may affect the costs loaded on to city ratepayers. Not sure that fairness is involved, but we are generally willing to help people when really un-foretold events occur (like the original earthquakes). But the flooding risk appears to a predictable event as things now are - so compensate them once but if they ask for more they are just asking to be compensated for their silliness!

  5. Two things
    First, thanks for the answers. It looks like general agreement that at least some of the flooding is a new reality and should be compensated once. Some of the flooding is predictable and the pricing should have worked itself out by now.

    Second, on what basis can it be said that the recent weather was a 1 in (some number bigger than 50 year event)?
    Christchurch has existed with weather records for at most 200 years. I can't believe that it is reasonable to build a mean and variance for rainfall that takes no account of time of year so that would mean that there had been substantially fewer than 4 such events in the entire historical record in order to reasonably claim a 1 in 50 year probability. today's press has indications that Christchurch/Canterbury has high variance weather so it would be risky to assume that no such events in the historical record meant that it was "rare".

  6. I've got no clue about the sourcing of the "1 in x" stuff. Council's working out some new flood levels given the earthquake-consequent drops in some parts of town too.

  7. The Hutterites are indeed a fascinating example of implausible communist success. Phil Keefer and I investigated the sources of their success. A highly sophisticated
    voting system appeared one of the keys, with God playing an important governance

    LEE BENHAM, PHILIP KEEFER Economic Inquiry October

  8. I never ever heard of such things in Manitoba. If anything, it went the other way. But it's possible that I just didn't know about it.

  9. Not that I know of, but there could be.

  10. Apparently the google is "extreme value statistics" or "extreme value theory" so approximately you take an ordered set of your worst event in each year and then try to fit that to the tail of a distribution. That gives you an estimate of the probability of bigger events than you have ever seen. As always, you should never trust a statistic without the confidence intervals.

  11. Concur with those who say Hutterites interesting case to look at. Very high exit costs for those who choose to leave. And colonies are on their own, for the most part. I remember a news story in Alberta during the 1990s (hard times for Alberta's economy) when some colonies went bankrupt and their members were refused by other colonies; because they were not in the social welfare system, some were on point of starvation.

  12. I don't remember that Alberta story. Also agree that exit costs are very high - I expect that part of the approach they take in their education system, whether for that purpose or not, effectively increases those exit costs. But, a lot of Colonies seem to be opening up on that front - more Hutterites are going to, for example, teachers' colleges.

  13. Is that how you're saying it should be done, or how you're saying they actually do it?

  14. That is the description of the typical process used for flood risk, storm risk etc for the purposes of civil engineering. I would be surprised if NZ practice turned out to be very different. That said, the UK confidence intervals (long time series available even if some of the early measurements are dubious) should be much smaller than NZ intervals. The wikipedia article on "100 year floods" also specifies the assumptions on lack of correlation between years that ought to be tested.

  15. The land in Christchurch has taken on new undulations . One of the most interesting was the resealed Fitzgerald avenue , where asphalt covered over a 50 drop in the road. You would think wouldn't you, that the goblins would understand that if you leave a hole, there water will collect.
    The Earthquakes were a National disaster $NZ 50 billion, thats $35,000 for every taxpayer, and we have no plan for the future.
    Gerry was right NZ Nat Govt have dropped the ball.
    It was a national disaster and should have been declared so, that means we all have the same liabliity.
    But no, they cordoned off Christchurch and left us with the EQC.
    To Gerry everything East of Colombo street is goon territory.
    I am beginning to get angry,

  16. Does anyone know if they use gmos for there vegetables?