Light week last week, probably be the same way for the next couple.
Late Thursday, AIR Worldwide tightened its estimate of Japan insurance losses.
Old estimate: $15B to $35B.
New estimate: $20B to $30B.
These line up a bit better with Eqecat’s estimate of $12B to $25B, which I recapped here – at least on the surface. When you dig down, AIR still projects losses twice as high as Eqecat.
There’s a bit of an apples-to-oranges problem here:
I’ve tried to line up things up below, using the midpoint of each modeler’s range (in U.S. dollars) to ease the comparison:
AIR’s property estimate is more than twice as high as Eqecat’s ($25B vs. $11.7B). Even so, Eqecat includes two items that AIR does not – business interruption and losses ceded to the government – that I’d estimate sum up to $1B.
Whose estimate is better? That’s way beyond my expertise. I do know that Japan’s government estimates the economic loss between $198B and $309B. The midpoint between those is $253.5B.
So AIR’s estimate is about 10% of the economic loss, and Eqecat’s property estimate is about 5%. The Kobe earthquake in 1995 had total loss of $100B and insured loss of 3% of that, or $3B.
It’s the eighth consecutive quarter (actually a 1% decline).
Five lines – commercial property, professional liability, D&O liability and EPLI all showed declines for the fifth quarter in a row.
Small and mid-market accounts were flat; large and specialty accounts showed declines.
The survey also indicated AY10 loss ratios are 5 points higher than in 2009, which were themselves 2 points higher than 2008.
Claim cost inflation was 4%, in line with long-term averages. (Last year claim cost inflation was -2%.)
Quoth Bruce Fell, director of Towers P&C practice in the Americas: “Our survey results for 2010 support the contention that the economic recovery will be accompanied by higher cost trends, and those estimates could increase if additional ‘catch-up’ from 2009 negative trends — beyond a return to long-term averages — would occur with rebounding economic conditions.”
Again, Japan vs. the world
Sat in on Thursday’s Eqecat briefing regarding the Japan quake. Highlights:
First, Eqecat, as I tweeted earlier, has labeled this a “seismological black swan.” The particular plates that shifted were considered too old and too slow moving to generate anything bigger than 8.2. The graph below, cribbed from Eqecat’s presentation helps explain:
Each dot in the chart represents a major quake zone, labeled. The position of each dot is determined by the age of the zone (x-axis, which notice counts from right to left) and the average slip rate per year (y-axis). The number represents the biggest earthquake thought possible from a given fault. The arrow points to the zone that slipped March 11.
So you can see NE Japan was thought to max out at an 8.2. In addition the probability of that happening in the next 30 years was thought to be 20%. And the biggest earthquake modeled anywhere in Japan was 8.75.
Part of the reason this one was so bad: Usually an earthquake shakes only one segment of a fault. This quake shook five. The slip zone – the faults that shifted against each other was enormous – about the size of Indiana. Now, Eqecat says, modelers will have to re-think what might happen in zones like this one, particularly zones along the Pacific Rim.
Or, as they dryly put it in their lecture: “Earthquake hazard cat models in megathrust regions need to account for the possibility of rare multisegment ruptures resulting in unprecedented high magnitudes – seismological Black Swans.”
You can also see that many Pac Rim zones (Alaska, Aleutians, Chile) were already considered candidates for 9.0 quakes. Still, in terms of catastrophes, the world just got a little scarier.
However, this doesn’t change the situation for California, where the maximum quake remains approximately 8.0. That’s because San Andreas and surrounding faults shift by rubbing side by side. The dangerous quakes – like the March 11 quake – occur where one fault slides beneath the other, which geologists call megathrust subduction boundaries.
The range of insured losses, $12B to $25B, breaks down this way:
Eqecat believes the tsunami drove losses much more than the quake, in no small part because Japan’s building codes are so good. They note this event differed from the 1995 Kobe quake, where much of the loss came from conflagrations spreading from building to building.
This time out, there were fires, but they didn’t spread.
Property estimates broke this way:
Dwelling risks are covered by the Japan Earthquake Reinsurance Co., which should be able to handle this level of losses easily. Kyosais are the Japanese equivalent of mutual insurers. The relatively light losses – given the enormity of the event – is because Japan is relatively lightly insured for earthquake.
Eqecat’s estimate includes tsunami losses, but their cat models – like much of the industry – don’t consider tsunami risk, as National Underwriter wrote earlier this week.
It’s also critical to note that this estimate includes business interruption – about 20% of the total – but it doesn’t include indirect business interruption. And there’s a lot of disruptions, because of power shortages, areas evacuated because of the nuclear disaster, and severely damaged roads and rail.
This morning, I blogged at I.I.I. about how earthquake insurance works in Japan. I want to add a bit to that, mainly to say residential insurance claims won’t hit the private market too hard, if at all.
But if losses get big enough, things could get dicey for the quasi-governmental Japan Earthquake Reinsurance.
As I noted, Japan Earthquake Reinsurance holds the vast majority of Japan’s residential exposure. Much of that is retroceded through treaties placed April 1. Based on JER’s 2010 annual report (pdf), here is the 2009 structure, spread over two treaties: Continue reading
EIOPA summarizes the results of QIS5 in press release here (pdf).
Compared to the calculation under Solvency I standards, insurance groups have €86 billion less surplus capital available, which is a reduction of 44%. However, the QIS5 exercise demonstrated that this effect would be largely absorbed if insurance groups apply internal models and transitional measures to calculate the capital requirements under Solvency II. This would limit the reduction of the surplus to €3 billion, which represents roughly 1%.
That €83 billion difference translates to US$116 billion of capital that could be freed up.
An ocean away, it sounds like every actuarial modeler in Europe just got a lifetime job.
Report here (pdf).
Annexes to the report here (pdf).
I’ve separated Japan quake/tsunami tweets from the rest.