The economics of Japanese earthquake insurance

Here is one interesting bit:

…very few people in that region of Japan held earthquake insurance, and also because of strict loss limits imposed by the Japanese government. For instance, residential buildings and furniture can be covered, but very expensive jewellery and artwork cannot, and there are rules that ban people from taking out insurance once an earthquake warning has been issued.

Mr. McGillivray said the Japanese government protected domestic insurers by limiting foreign participation in the system and, to keep the risks manageable, limited the payouts.

Are the capital requirements on Japanese insurance companies well-designed?  Probably we will find out.  Can anyone recommend readings on this and related questions?

Comments

I'm not sure of how Japanese insurers manage catastrophe risk, but can give a few comments on usual western practice:

- a direct insurer will typically only retain a small part of its risk from catastrophes;

- above this level it will hedge the risk via reinsurance to the global reinsurance market;

- the 'global reinsurance market' can be thought of as including the use of catastrophe bonds and the like (these are, however, typically very small compared to the hedging done with reinsurers);

- in Japan's case much of this risk is actually taken by a government authority;

... which means that, one important caveat aside, the risk 'capital' to back catastrophe risks isn't supplied by insurers. It mainly comes from reinsurers and (for Japan and NZ) the government.

The caveat is that the hedging is only done to a certain level, typically specified by a likelihood of exhaustion. That is, for example, a Japanese insurer may hedge its earthquake risk to, say, 1 in 200 years. Meaning that its models see a 1 in 200 annual chance of an event exhausting its hedging / reinsurance. Given the small amount of capital typically held by direct insurers, exhausting ('blowing out the top') your reinsurance program would likely cause insolvency. An annual 1 in 200 chance of exhaustion is not small - it means, for example, a 1 in 20 chance over a 10 year period. And the models will, of course, be wrong...

Finally, the key players in the global reinsurance market are European, Munich Re & Swiss Re in particular. Certain events would cause them to become insolvent. These are systemically important institutions for the insurance industry. A severe earthquake in Tokyo is a potential insolvency event. Thankfully, the initial numbers indicate that the insurance losses - while still very high - are well below this level. They may, however, be high enough to sufficiently deplete capital so as to move us into a hard reinsurance market.

A couple of other comments, maybe getting too technical:

- The catastrophe models are created by independent (ish) companies such as AIR and RMS. You may see their press releases in the media.

- There models do not capture tsunami risk well, or sometimes at all. It is far harder more complex to model than losses coming from ground shaking / fires i.e. directly from the earthquake.

- When something isn't modelled, insurers often assume it is zero. It wouldn't, for example, surprise me if some insurers had purchased, say, 1 in 100 years worth of earthquake catastrophe protection, completing ignoring tsunami. Bluntly, this means they were, in reality, only reinsured to 1 in (say) 50 years - tsunami losses, of course, being completely correlated to the associated earthquake.

- Insurers also often ignore demand surge risks. This is the increase in rebuilding costs in the aftermath of a catastrophe i.e. basic economics: almost infinite demand (vast number of properties to fix) while supply is limited. Again, very hard to model or get data on, so it is assumed to be zero.

All that said, the size of the event may be such that the insurance industry escapes without too much damage. The area isn't that densely populated.

The Japanese government, however, may be a different matter - it takes the nuclear risk, for example...

This is a big insurance event but not likely to severely strain the insurance market. Katrina, for example, topped $40B insured losses. Observers have said the property/casualty industry has $70B excess capital worldwide.

Early insuarnce estimates of quake damage - ignoring tsunami - top out at $35B, according to a leading cat modeling firm, AIR. Tsunami estimates are being worked now. The $35B is about what models show a repeat of the 1923 Great Kanto Earthquake would generate.

My impression is that Japanese insurers are very careful in managing their catastrophe risk, so much of the exposure is ceded to the reinsurance market, through the process that James Goodchild describes above. Reinsurers have become much better at managing their exposure to catastrophes in the past decade.

Lloyd's, for example, requires its underwriters to know their losses from each of a set of realistic disaster scenarios, including hurricanes, terrorism, airliner crashes, etc. Specifically, Lloyd's expects underwriters to know their losses from a repeat of the 1923 quake.

In the p/c industry, the tragedy won't trigger massive insolvencies. It might trigger a worldwide increase in rates. Insurance markets are essentially priced through supply of and demand for insurer capital. Demand doesn't change much, but an event like this could reduce supply enough to cause rates to rise (mainly in commercial insurance - in U.S. regulation mitigates wide swings in rates).

I can't comment directly on Japan's solvency standards, but will note that Japan will likely be one of the first non-European nations to adhere to Solvency II, the emerging EU standard (similar to Basel's banking standards). In that regard, Japan is a couple of years ahead of the U.S. (whose solvency standards are quite high).

Generally speaking, government bodies only afford property and casualty insurance for adverse selection risk e.g. hurricane, flood, earth quake. The adverse selection comes about as the risk can not be spread across homogenous exposure units over a wide geographic area i.e. only those risks situated in the hurricane, flood, earth quake exposure area will consider buying insurance.

When coverage is in fact available from government as well as private insurers e.g. NC Beach Plan area, the government insurance is under priced. Hence coverage flows to the under priced coverage until the exposure balloons past the ability of the government body to pay for the mounting exposure at which time coverage is rationed (classic price fixing scheme results).

Meanwhile, many governments do not reserve the risk correctly and depend on the taxation authority of the government body to make up any reserves for losses (related to the under pricing phenomena). Finally, the under pricing causes the government body to purchase woefully low amounts of reinsurance.

There is a reinsurance question, perforce, but I never worked on Japan while in that industry and I have to wonder who would write the kind of cat cover that this would blow through. I'm just guessing here, but my bar-stool opinion at this point is that they are sovereignly effectively self-insured with little or no reinsurance sitting on top of them. Would like to know the facts.

My main question for all the fine economic minds visiting this website is have we witnessed the coup de grace on Japan as we know it? Will this be seen as the Big Bang of a precipitous decline in Japan as a force in the world? I think the odds are likely. History happens.

There's clearly an incomplete market here.

Earthquake insurance here would double our annual insurance premiums. There would be a 15% deductible on both the house AND all our belongings which would cost a fortune.

The probability of a total loss is minuscule. The loss given an event is mitigated by the fact we can walk away from our mortgage. The lender can repossess a pile of rubble in a flooded neighborhood.

Talk about an "under water" mortgage!

For every incident, we scramble to raise international aid. How about setting up an International Hazard Insurance Fund and offer everyone on the planet coverage? That would spread the premiums over billions of people, few of whom are going to have a disaster event, EVER, in their lives, so the premiums would be very small.

Never knew their was earthquake insurance but it certainly makes sense in Japan. Earthquakes -- the relatively light ones that only serve to remind you that there's the ground below your feet -- are a common occurrence in that country.

When in doubt on insurance-related matters, look to London where approximately 50% of all ins co biz gets done: http://www.telegraph.co.uk/finance/newsbysector/banksandfinance/8380422/Lloyds-insurers-fall-on-22bn-claims-bill-estimate.html.

This event is, of course, why we have reinsurance, so that even the worst cat losses can be absorbed by the global market, thus mitigating the damage to any single economy. Reins is a gray and arcane space in the otherwise heavily regulated ins industry, and let's hope it stays that way, as it is quite rational and super-efficient.

Japan is, as one might expect, an odd man out in this arena. I now recall Tokyo Re as being a rather prissy, stiff-necked outfit that we never really considered when going out to the markets to place a cat facility.

As for my earlier question, O Wizards of Oikos, would anyone care to venture an opinion on the effects of this event on Japan's long-term prospects. Seems like a tipping point to me, but what do I know?

The caveat is that the hedging is only done to a certain level, typically specified by a likelihood of exhaustion. That is, for example, a Japanese insurer may hedge its earthquake risk to, say, 1 in 200 years. Meaning that its models see a 1 in 200 annual chance of an event exhausting its hedging / reinsurance. Given the small amount of capital typically held by direct insurers, exhausting (‘blowing out the top’) your reinsurance program would likely cause insolvency.

As someone with reasonably good knowledge of global reinsurance markets, the most interesting observation about Japan is that there is so little insurance overall. For example, as I'm sure most have read, homeowners don't often carry earthquake insurance even though everyone knows they're exposed to it. Two potential explanations for this outcome:

1. All of Japan shares very high earthquake risk reasonably uniformly. This is very different from say the US where the Pacific Coast has high quake risk, East Coast has low risk, and the middle of the country (New Madrid fault) somewhat ignores their risk. This uniform distribution across Japan means the average citizen is more willing to let the government absorb the risk. As a (Northern) New Jersey resident, I oppose every national program to cover earthquakes and hurricanes. (Similarly, the California Earthquake Authority serves as a bulk buyer of earthquake reinsurance for all of California.)

2. Japan's incredibly high savings rate seems to allow them to more easily bear the risk. This is no different than individual households in the US: Wealthy people tend to have much higher insurance deductibles because they can absorb the risk out of savings. Japan's wealthy economy allows them to absorb much greater losses without going to (what one might perceive as) very expensive capital sources in the form of global reinsurers.

Obviously, Japan has done a phenomenal job of risk mitigation that to a large degree is unseen, as the media focuses so much attention on nuclear reactors. For example, in California reinsurance markets, one of the more costly components of total coverage is termed "fire following", as in the fires that will follow from a substantial earthquake. California has natural gas lines that break and burst into flames. Japan, it seems, has natural gas canisters that scare guys with video cameras.

buildings and furniture can be covered, but very expensive jewellery and artwork cannot, and there are rules that ban people from taking out insurance once an earthquake warning has been issued.

Comments for this post are closed