Earthquake-prone buildings

By Eric Crampton 08/08/2013

Owners of earthquake-prone buildings now have a bit more time to bring them up to spec. The owners are mad because they say it isn’t long enough; people who experienced Christchurch are mad because buildings will still fall on people and kill them in another quake. They could both be right.

It is perfectly plausible that there are buildings that need never be compelled to be brought up to 33% of new building code. Imagine a building in the middle of nowhere, with no nearby pedestrian traffic, and occupied only by those who know about the risk or who are well-advised about it by a sign at the doorway. There is no reason for the government there to get involved, or at least no reason that comes from economics. People can trade off cost and beauty against risk – that’s allowed. And so a national rule that forces the owner of such a building to make costly investments to bring it up to code imposes cost in excess of benefit. The owner either will sink money into the building where it isn’t warranted, or he will demolish the building that he otherwise would prefer to keep.

On the other hand, imagine a building in downtown Wellington with an unreinforced masonry facade. Everyone in the building knows about the risk and accepts it in exchange for lower rental rates or enhanced amenities on other margins. And that’s all fine. But passers-by on the sidewalk and buses driving by on the street have uncompensated risk forced upon them. While the owner will there rightly claim that it does not pass his cost-benefit analysis quickly to bring the building up to 33% of code, he is not accounting for the costs he is imposing, probabilistically, on every passer by. It can easily be the case that, when accounting for the risk of death he is imposing on each person walking past his building, upgrading the building or demolishing it would pass cost-benefit. But he does not care about the costs imposed on others. The new rule is too lax in this case.

What then is an optimal rule? We’d need some way of accounting for the true risk that a building imposes. That risk depends not only on structural features of the building but also on the building’s surroundings. And it would be pretty hard for central government to be able to come up with a clean rule. As we saw in Christchurch after the September quake, City Council had a rule in place requiring the closure of footpaths adjacent to risky buildings; Council interpreted the rule in perverse ways. Instead of blocking busy Colombo Street, Council decided that the engineers must have meant that 605-613 Colombo imposed risk instead on tiny alleyway beside the building. And then the building fell on a bus and killed a bunch of people and left Ann Brower to work out the series of spectacular regulatory failures that led to her being the only survivor on that bus.

There’s an easier way. Honestly, we do not know when another quake will come or which buildings will collapse. Engineers can put up widely varying assessments of the true structural risk imposed by a building. What do we do when faced with this kind of uncertainty? Impose a liability rule. Instead of giving building owners 15-20 years to get their buildings up to 33% of the new building code, give them five years to get an engineering assessment, to put a safety letter grade prominently at the door, and to get liability insurance. At the end of the five year period, have every building owner liable for damages for every person killed or injured if their building falls down on passers-by. There should not be liability for deaths and injuries incurred by persons inside the building: we can and do voluntarily assume some risks, and we should not prevent people from taking on those kinds of risks. But if your building falls down and squishes a bus, you should be liable for the deaths of each of the people inside of that bus.

The Ministry of Transport currently sets the Value of a Statistical Life in New Zealand at $3.77 million. That’s arguably too low, but it’s a great benchmark: arguing about $3.77 million versus the $5 million or so you’d get from a back-of-the-envelope application of revealed-preference measures from the United States to New Zealand, accounting for the income elasticity of safety preferences and differences in income across the two countries is second order. First order is getting a consistent benchmark across different regulatory and liability sectors.

A building owner potentially liable for $3.77 million in damages paid to the estates of those who his building kills will adequately take their interests into account in deciding whether to fix up his building. It would not be that hard to require that building owners carry insurance sufficient for paying such liability claims, or to prove assets sufficient for covering the potential liability. If you’ve got a building in the middle of nowhere with no passers-by, your insurance premiums will be very small. You then will make the optimal choice and not upgrade your building. If you’ve got a brick-facade building in downtown Wellington, you’ll have to weigh up the costs of insurance against the cost of fixing the place up.

Right now, we are in the worst of all possible worlds. Building owners face neither liability for the risk their buildings impose on those outside their buildings, nor any sufficient regulatory regime to ensure that owners are making appropriate investments in ensuring that their buildings do not impose excessive risk on passers-by. This is why Wellington scares the hell out of me. It’s pretty, and I love seeing the old buildings that we no longer have, but they terrify me. I have absolutely no confidence that even really rather dodgy buildings are getting the attention they deserve.

Were the Government to have any interest in implementing a regime such as that described above, I’d recommend one further change. Flip the heritage building regulations around such that heritage boards have zero regulatory power but instead get an annual budget. Owners of risky heritage buildings should be free to demolish them if that’s what make sense, given the risk they impose and the cost of upgrading them. Heritage boards’ main role should be the payment of annual stipends to owners of heritage buildings for the provision of heritage amenities. Give them a generous budget, funded partially by local Council, partially by central government, and with ample provision for voluntary donation from the public. Let them decide, within that budget, where they can do best by spending money. And then just let go of the rest.

I will absolutely hate saying “I freaking told you so” after Wellington gets a big quake in which unreinforced masonry winds up killing a bunch of people needlessly.

And, in anticipation of the likely critique: yes, I am here absolving building owners from liability for those who chose to be inside their buildings. But current policy absolves them of that liability for those both inside and outside.


0 Responses to “Earthquake-prone buildings”

  • “If you’ve got a building in the middle of nowhere with no passers-by, your insurance premiums will be very small.”

    I’d like to see numbers to back this. My own impression (and a little experience in a different setting) is that liability insurance is expensive and I can’t help questioning if the premiums are not not based so much on the odds of someone being hurt, but the cost if someone is.

    • It has to be based on both! And I suppose if premiums are much larger than actuarial risk, then I’m in the wrong business and ought start an insurance company.

  • Hi Eric,

    Let me try again.

    Your suggestion takes as a premise that the premiums will vary according to the number of people typically around a building, etc., rather than dominantly (or entirely) being be based on an ‘average’ situation (and the dollar value of liability sought).

    (A little background. In the case of the liability insurance I have taken out, it was taken out against a dollar value of cover for particular events without much about the particular setting, number of people there, etc. My situation is different, but I still can’t help wondering if the same general principle would apply – hence my question.)

    “And I suppose if premiums are much larger than actuarial risk, then I’m in the wrong business and ought start an insurance company.”

    Why you are writing this? I didn’t say or imply that “premiums are much larger than actuarial risk”.

    What I asked for was evidence for the premise your argument rests on. These are science blogs; asking for evidence backing premises, claims or arguments is pretty normal, right? Why not just point to examples, evidence, whatnot – ?

    • Hi Grant,

      I’m not arguing that they’d be measuring the exact traffic at every location. But they’re going to have different rates for downtown Wellington than for a building in the middle of a field. The actuarial tables might be somewhat coarse, but they’re not going to be so coarse that they fail to incorporate some measure of actual pedestrian traffic.

      I’d expect in your case that they reckoned you to be part of a risk class with particular characteristics, then looked across the table to see what to charge people of your characteristics for a policy for a set amount of coverage.

      I suggested that if firms aren’t working in this kind of way and are mispricing risk, then there is a very large opportunity for someone else to come into the market and make a profit. Entry barriers being relatively low is itself evidence of that prices should be pushed towards underlying risk.

  • Eric,

    You’re simply missing my point (or perhaps avoiding it). I’m asking if this type of insurance exists or would exist in the real world, not some hypothetical what if. My example is what happens in the real world; it’s what really happens. (Your elaboration adds information to the policy; I wrote “without much about the particular setting, number of people there”.)

    While purely as academic idea I can understand why your idea appeals, but it occurs to me it’s unlikely to ever be a reality for what I thought were fairly simple reasons, so I asked for examples, evidence. Why not just back it, or, if you can’t, say that or say that it’s only hypothetical.

    Notice how in your replies you haven’t presented evidence but give generalised ‘it would be like this’ statements – which are, effectively, making more assumptions. When you write “but they’re not going to be so coarse that they fail to incorporate some measure of actual pedestrian traffic”, you presume some insurance company would do that.

    I’m also doubtful that the data needed to develop the risk profile your model would want exists.

    I could elaborate in detail, but I’d rather spare myself the effort as my impression is that you’re dealing in hypotheticals. Perhaps try looking at real policies? (Closer to what I was asking for.) Third party liability policies for, say, contractors or consultants working off site, vehicle insurance, travellers, building site liability cover, etc.? Perhaps policies used overseas? Are detailed and offer a ‘scale’ in the manner your hypothetical policy would need to function in the way you’d like? My understandably limited experience with a few of these suggests they’re not, but it’s not my patch – hence my asking. What your hypothetical model seems to want is a scaled effect, but I’m skeptical that in the real world an insurance would (or even could) do that, a few exceptional cases aside. Consider for example a rural community hall, low traffic most of year, packed out a few days each year. You’d think insurance companies would just offer a policy for ‘third-party liability for injury or death within or in the immediate proximity of a building’ across the board and that’d be it, with few or not specifics about pedestrian traffic, etc. Seems to me that reality makes your suggestion too messy for an insurance company to actually do. So I thought it appropriate to ask if policies of the kind your suggestion seems to want exist.

    Also my understanding is that people owning buildings isolated in the middle of a field or whatnot, can already apply for an exception.

    Excuse the length – what you get for me being able to touch-type reasonably quickly I guess.

  • This insurance will only exist in the real world where there is an institutional environment that requires it. In any place where building owners are not liable for deaths, there is no need for this product and it will consequently not exist.

    We do not need that pricing be perfect. We only need that is scales reasonably with risk. And all we need for that to be true is that insurance markets are competitive.

  • So it’s all hypothetical and you’re (still) not presenting examples of actual liability policies that work that way (so that the underlying premise might have some substance).

    Sorry about this, but if you don’t show examples that work in a similar fashion, why should readers think your idea plausible?

    I for one am giving up. Giving more hypotheticals is not addressing what I asked (in fact, to my reading, they just present more assumptions) and leaves the premise your suggestion rests on unsupported. Just my opinion.

    • Jeez, Grant. Look at any building insurance policy. Do your premiums for fire insurance in a wooden building change if you have a sprinkler system? The premium will depend on the value at risk (insured value of the building) and the extent of the risk (is there a sprinkler system?) Look at any life insurance policy. Do your premiums change if you’re a heavy smoker? Every existing insurance policy bases its premiums on risk. Some differences in risk categories are too trivial for insurers to care about: they don’t bother getting full details about every meal you eat for your health insurance policy. You’re asking me to prove that water flows downhill for a particular class of water.

  • If you could go easy with the tone, I’d appreciate it. Your examples aren’t third party liability policies. (You’ll see that’s what I asked after.)

  • My experience is that firms can (& do) insist consultants / contractors carry liability insurance before working on their site.

    • Different kind of liability insurance. But still, why do you think there’s no difference in premiums across individuals of different risk classes?

  • Eric,

    “Different kind of liability insurance.”

    I know they’re different types of insurance, I said as much when I brought them up. (I pointed to different types of thirty-party liability insurance earlier as possible examples of how the industry treats third-party liability that might serve to back or clarify a premise that underlies your suggestion; I was trying to politely invite you back your argument.)

    “But still, why do you think there’s no difference in premiums across individuals of different risk classes?”

    I’ve addressed this earlier. (I’d rather not go in circles if you don’t mind.)

    I’m going to leave this as it is, but one addition just as food for thought. It has since occurred to me that your suggestion [potentially, at least] gives owners the option to put money in lieu of others’ safety. They’d have the option of paying a higher premium rather than upgrade the building if they thought that financially better for them in the bigger scheme of things. Of course, that’s at higher risk (physically) to passers-by, etc.

  • That building owners could choose to pay higher insurance premiums rather than fix the building is a feature, not a bug. So long as the owner is liable for damages caused if the building collapses, then the owner has incentive to fix when efficient to do so, and not to when not.

  • I wrote about injury or worse to people, not ‘damage’ (“higher risk (physically) to passers-by”). Personally, leaving that standing would be more than just a bug; I’d like to think most would see it as defeating the main point of the strengthening programme.

  • So, Grant, if it would cost $20 million to bring some building up to code, and we knew with certainty (grant me the assumption please) that, in the event of an earthquake, only one person would be killed, you think this is worth doing?

    Note that MoT’s value of a statistical life for policy purposes in New Zealand is $3.77 million. If we’re spending well in excess of $20 million to save a statistical life in this case ($20m divided by probability of an accident = $large), while spending less than $4 million in other cases, we could save more lives by spending more in other areas than by spending it on this building. That’s the whole point of the insurance proposal: set the liability equal to the value of statistical lives and let the market sort out which investments pass cost-benefit. Without that we are wasting money.

  • Unfortunately when bad things happen (eg building collapse) we are dealing with real lives, rather than statistical lives. The real person who dies will almost certainly be part of a wider social group that will be shaken by their death, so the impact is much wider than that one person’s dying.

    • That’s true. But we don’t know which buildings will collapse or which people will be under them when it happens. Just like we don’t know who would be the person whose life would be saved by alternatively spending the money on straightening out a curve on the road from Christchurch to Kaikoura. Ex ante, all of these policies deal only with statistical lives. If it were “Spend $x to save Eric Crampton’s life (or Alison’s life, or some other named person’s life), that would be far different. But that’s not what’s proposed.

  • Eric,

    You’re suggesting it is entirely in the hands of the building owners to call if their building is to be made safe for others or not. It isn’t their call to make.

    Under your suggestion if some building owner were to think “stuff it, I’d rather just pay $2000 more p.a.” (or whatever amount), they could just thumb public safety in the nose and put their financial interests ahead of others’ safety.

    There’s a reason some things are set by regulation.

    “if it would cost $20 million to bring some building up to code, and we knew with certainty (grant me the assumption please) that, in the event of an earthquake, only one person would be killed”

    Can you see that this is a circular argument? (It circles through assume your premise that the insurance can/will scale [the underlying premise I earlier tried to get you to back], then using the assumed premise as the basis for the outcome. Before you can offer this line of thinking you need to resolve the premise. It’s why I picked up on it in the first place.)

    “Note that MoT’s value of a statistical life[…] and let the market sort out which investments pass cost-benefit. Without that we are wasting money.”

    I read this as confirming you’re happy to put money ahead of lives. Your call, but it seems immoral to me.

  • Yes, I am very happy to say that, beyond some dollar amount, it is a bad idea to spend money saving lives. And you are too, even if you won’t admit it. The Ministry of Transport puts the number at $3.77 million. I’d be happier with $5 million. But if you think there just shouldn’t be a number and we should just go with the gut on it, then we get absurdities like those found in the States where they’ll impose costs on the economy well in excess of $10 billion per statistical life saved in some environmental regulations while the cost of saving a statistical life through programmes around subsidising smoke detectors is around $20 thousand. We would save more lives by taking a fixed value of a statistical life and reallocate money from things that are really expensive per statistical life and put it instead into things that are cheap per statistical life saved. But you can’t do that if you want to say it’s just immoral to put a number on it.

    Further, the government budget would look very different if we said that it’s simply immoral to put anything ahead of lives. Money isn’t just money – it’s the things we could get with money. If we spend $10 million on something that saves one life, that’s $10 million we don’t spend on things that make life more worth living. At some point we have to draw a line. Or, do we cut all expenditures on anything that has zero chance of saving lives? I’ve never heard of a public art installation that’s saved a life, for example.

  • “And you are too, even if you won’t admit it.”

    It’s a bad idea to words in other’s mouths, Eric. You’re missing my point entirely. It’s not about the dollar amount per se. It’s about that it’s not a building owner’s call.

    “But if you think there just shouldn’t be a number and we should just go with the gut on it,”

    I’ve never said anything like this.

    I’m well aware of the larger-scale argument about funding, it’s a regular feature of public health spending issues – but you’re missing what I’ve have been saying: 1) Your argument has a logical flaw (an unsubstantiated premise), 2) making it individual building owner’s call has moral problems.

  • The government’s call is to set VSL. The current level is $3.77 million. The government’s call is to set liability equal to VSL for buildings falling on passers-by. After that, everybody optimises. Sometimes, the government decides that a roading improvement that expectationally saves a life isn’t worth the money (or makes the same call for a public building); sometimes, a building owner decides that a building improvement that expectationally saves a life isn’t worth the money.

    My assumptions:
    1) Insurance companies are profit-seeking;
    2) The insurance industry is at least somewhat competitive.

    That’s all you need to generate that insurance companies will price policies based on potential liability on the downside and risk of the downside obtaining.

    Even #2 isn’t all that needed: a monopoly insurer will charge too much for any given level of insurance, and this results in more buildings being torn down as being too expensive to insure than would be the case in a competitive market. Relaxing #2 yields too much investment in safety relative to an optimum.

    Relax #1 in odd ways and you could skew the results. Suppose the insurer is rich and loves old buildings and is happy to subsidise policies for old buildings. You then might get underinvestment in building safety if safety hurts heritage amenity features, but you also ensure that the insurer is then on the hook for damages on the downside.

    What would break things most is neither 1 nor 2, but ancillary assumptions around the regulatory regime for insurers. Imagine a case in which the government certified an insurer as sound, or at least set up a regulatory structure that people trusted ensured that insurers were sound, then insurers underpriced risk and, in the event of extreme downside case, went bankrupt rather than paying out. That’s a real worry for my scheme. Give me heck about that point if you want – my scheme does require that insurers have adequate reinsurance for large-sized downside risk and that consumers are either vigilant about that or a strong regulatory system ensures it. But this “Oh we can’t prove people charge premiums that relate to risk” stuff is pretty weak.

  • Eric –

    I’ll leave you to it. Re your parting shot –

    “But this “Oh we can’t prove people charge premiums that relate to risk” stuff is pretty weak.”

    I asked you, to try invite you to substantiate what I saw as an underlying premise. I (still) don’t have an answer and I honestly don’t mind.


  • What is happening “in the real world” is that Insurance companies have threatened to increase the premiums on any building they suspect is earthquake prone by such a degree that the owners are taking action. After the initial evaluation, if it is E.P. then the premiums ruse until work is undertaken to strengthen the building. The Governments time frame has been hijaxked by the theiving deceitful insurance companies.
    Everyones premiums have risen when their risk factors have not increased 1 iota. This is because the premiums are not held in trust or invested to cover an event, but spent in the now and wasted. Insurance companies have conned their policy holders for years.
    Now insurance holders are also being forced to pay a excess if an EQC claim is involved, regardless of whether you want to or not.

    • Agree that insurance is more likely to be providing the binding constraint. But while underlying riskiness may not have changed, our information about the true cost of that riskiness has changed. Insurers might not have reckoned on some of the costs they’re now facing in Christchurch and are now pushing back on the amount of coverage and ramping up premiums to fix things. Agree that many parts of insurance now do feel very very scammy though. We have YET to see our AMI / Southern Response guy out at our house… 3 years later. They’re not sending anybody out to look at “out of scope” stuff for a while. Another year or two to wait last we heard.