Tuesday 3 May 2016

In search of Arrow-Debreu worlds

I remain a bit puzzled as to why I have been utterly unable to convince any insurer to provide me a quote for earthquake insurance for Wellington.

I've mooted the product before; I've stripped it down here to what I think is its simplest form.
In short, I want insurance against a large Wellington earthquake. If the risk is on the order of 1/10 for a large one sometime over the next century, that’s about 1/1000 annual risk. I’d like to purchase a contract that provides a large lump-sum payment if a sufficiently large earthquake hits Wellington. I’m sure there’s a way of specifying a set of legal conditions that would effectively say “If something at least as substantial as the 2011 Christchurch event happens in Wellington, this contract pays out.”

As first cut, I’d suggest basing it on the Modified Mercalli reading for downtown Wellington, with a trigger at MM 9 or higher. The 1855 earthquake was MM10, but nothing else higher than 9 has been recorded in Wellington since colonisation. The Christchurch February quake was MM 9. We would need to check that MM ratings applicable to downtown Wellington are reported for larger events farther from downtown.

I’ll explain why I want this contract, and in doing so potentially explain the size of the potential market.

I was in Christchurch for the 2011 event. That event resulted in substantial uninsurable losses not just from the event, but from the post-earthquake experience. Businesses who had continuation insurance found themselves out of luck when Council barred their entry to their premises: acts of Council are not covered. Homeowners who thought they were insured to a rebuild-as-new standard found themselves instead with revised standards for reinstatement methods that left them substantially worse than prior to the event.* And the process, involving large-scale coordination failures between EQC and private insurers, let things drag out for years. I never want to go through that again, and I suspect that many who experienced Canterbury would appreciate a different kind of contract.

A large lump-sum pay-out that comes if the insured event happens requires no lengthy claims adjudication process: the MM9 quake either happened or it did not. No assessors need argue about whether anything on the house were pre-existing damage. There’s no interface between EQC and anyone else arguing about whether something is over or under-cap.

Instead, I hike out of town with my family as best I can, start my life over somewhere else in the world with the resources to do so comfortably, and have a real estate agent pack out our house and sell it as-is, with all earthquake claims transferring to the new owner to deal with.

There has been sufficient press around the problems in Canterbury post-quake insurance that many owners in the Wellington area would be aware of the problem. Simply announcing the existence of the new insurance product would undoubtedly lead to press coverage that would help to attract new customers.

Moreover, I expect that this is something you could and should take worldwide. California and the rest of the Pacific Northwest, Japan, and other places offer a bundle of offsetting uncorrelated risks that could build a pool for purchasing reinsurance against claims. If you look at the Pacific Northwest, many homes are uninsured because earthquake insurance for natural disaster is too costly, but it’s too costly at least in part because insurers face costly assessment and dispute processes after an insured event, where homeowners will be tempted to pass off pre-existing damage as being due to the event. Insurers then face very uncertain overall liability. With my proposed product, the pay-out is known with certainty: if the event does not happen, claims are zero; if it does happen, the pay-out is the total sum. It is then more like life insurance than like any standard homeowner insurance.

Please let me know if this is a product you think could be developed, or if there’s something obvious I’m missing explaining why this cannot easily be offered.
The only explanation that makes any sense to me thus far is that contracting costs are non-trivial and fixed, that the potential number of customers for such a product is smaller than I would anticipate, and so it is worth nobody's while to develop the contract and set up the reinsurance. The alternative is that insurers are leaving dollars on sidewalks because they're too conservative; in a world with insurance against your celebrity endorser's disgrace, that doesn't seem immediately plausible.

If the actuarially fair price for a $1m payout for a 1/1000 annual event is $1000, I'm happy to pay the standard insurance multiple over the actuarially fair rate for the contract.

* Since then, the High Court has - five years after the earthquake - issued a declaratory judgement that repair to an "as new when new" standard specified in insurance contracts actually means that the repair has to be to that standard, updated to meet current building code. EQC had been instead rebuilding to an alternative standard that MBIE came up with, in which notched bearers, jack and packed piles, and floors up to 5 cm out of level counted as good enough. And those with the patience to go through the whole rebuild process again can now presumably go back to EQC for a do-over. This Press editorial is also good.

1 comment:

  1. Good idea, though the risk might be rather higher than you suggest. The Christchurch Feb 2011 quake was magnitude 6.3 and led to MM8 shaking. According to Geonet, the annual risk of a magnitude 7 quake or greater in the Wellington region is 1%: http://info.geonet.org.nz/display/quake/Cook+Strait+aftershocks+and+forecast+probabilities

    So depending on criteria, it may cost more like $10,000/year (plus profit margin) for $1M cover.