31 December 2012

Updated: Normalized Hurricane Losses 1900-2012

The graph above shows an updated estimate of the 1900 to 2012 normalized hurricane losses for the United States. The normalization methodology is described in Pielke et al. 2008 (here in PDF) and the data presented in the graph comes from the ICAT Damage Estimator, which extends the analysis of Pielke at al. through 2011.

Today Willis Re released a report (here in PDF) on the state of the reinsurance industry, and presents an estimated $20-25 billion in insured losses for Sandy. As is conventionally done, to arrive at total losses for 2012 I have doubled the $25B figure to arrive at an estimated $50 billion total loss for Sandy. Please note that the final loss estimate, apples-to-apples with the normalized loss database (kept by NOAA/NHC) may wind up being higher or lower. In addition, I have added in 3 placeholders of $5 billion in losses for the 3 other storms (besides Sandy) which made landfall as post-tropical cyclones of hurricane strength in 1904, 1924 and 1925 -- the losses from these four storms are depicted in grey in the graph above). In 2013 we will develop a rigorous basis for estimating these losses, which I'd guess have a good chance of winding up larger than the placeholders I have entered for now.

In case you are curious there is no trend in the normalized data, which makes sense as there is also no trend in hurricane frequency or intensity at landfall in the United States over the same period, and the lack of trend is insensitive to the removal of the 4 post-tropical cyclones.

What does the reinsurance market say about all this? Willis Re explains:
"most reinsurers are still within their annual catastrophe budgets for 2012 and not facing any capital impact... In the absence of Superstorm Sandy, reinsurers would have found it difficult to resist buyer pressure for further concessions. As such, Sandy’s impact has helped to stabilize market pricing on an overall basis and reinsurers have largely delivered to their clients in terms of capacity and continuity."
In other words, thank goodness for Sandy. 

8 comments:

manicbeancounter.com said...

Assuming
1. That a 2012 will be higher than average for claims
2. 2013 will have a surge in premium income (through higher premiums, and people increasing their cover)
3. 2013 has no >$5bn disasters
would you expect the industry to more than make good the 2012 losses in 2013?

In a similar way, did the insurance industry make good the 2005 Katrina losses in the two years following?

Roger Pielke, Jr. said...

-1-manicbeancounter

Thanks ... I'll take issues with your premises:-)

1. Depends on your geographical scope, but globally for reinsurance? I don't think so. In some sectors, yes, like marine.

2. No. As Willis Re reports reinsurance cover will be stable to lower in 2013.

3. Highly unlikely.

With respect to your bottom line question, the reinsurance industry did very well in the years following Katrina. The market has seen a steady increase in reserves, such that there is actually a shortage of big disasters. I've seen figures such as "2 Katrinas" mentioned as the scale of disaster needed to harden up the market.

Thanks!

manicbeancounter.com said...

Thank you Roger for the quick and fulsome reply.

Have just been checking out the ICAT damage estimator tool. It is a great tool for comparing Sandy with previous storms. The previous most costly storm within 50 miles of New York was Diane in 1955 with current damage estimate of $24bn. This storm had wind speeds of "just" 65mph. On the other hand Storm 2 of 1934 with current damage estimate of $0.54bn had wind speeds of 120mph. Part of the reason is that Diane ploughed right through the middle of New York, whereas Storm 2 of 1934 tracked 10 miles south of the city. (It might be that Storm 2 of 1934 had slowed by the time it hit New York). It seems to demonstrate that even normalized losses may be a poor indicator of storm trends.
Another aspect is that storm damage costs might be inversely related to the expectation of the storms occurring. So a tropical storm hitting Florida will have lower damage costs than a similar storm hitting New England due to better defensive measures taken in Florida.

http://www.icatdamageestimator.com

All the best for the coming year.
Kevin Marshall

Roger Pielke, Jr. said...

-3-manicbeancounter

You are correct that a lot hinges on exactly where a storm hits ...

Happy New Year to you as well!!

EliRabett said...

How have you accounted for improved observation and forecasting of landfall, improved housing construction, etc. Given that these factors are not in your data, lack of a trend means that the storms hitting land are worse.

Roger Pielke, Jr. said...

-5-Eli Rabett

You write: "lack of a trend means that the storms hitting land are worse"

Fortunately, there is no need to engage in speculation -- to assess whether storms are getting worse we can look directly at data on ... storms.

The data show that there are no long-term trends in hurricane intensity or frequency at landfall -- in the US or elsewhere -- over the period of record.

This serves as a nice independent check on the fidelity of our normalization methods, which as would be expected, also find no trend.

Thanks.

EliRabett said...

Sorry, you can't have it both ways, your methods of adjustment (from the linked PDF)
--------------------
adjusting for inflation, wealth, and population updated to 2005, called PL05; and 2 the methodology used by Collins and Lowe 2001 , adjusting for inflation, wealth, and housing units updated to 2005, called CL05
------------------

Do not adjust for improved forecasting, and weather observation which give time to harden buildings and evacuate endangered expensive things such as people, airplanes and cars, nor for improvements in construction including hurricane straps, etc.

Therefore you have merely pointed out an inconsistency, that either your estimates are wrong in correcting for the factors given, or the estimates of land falling intensity/ frequency are wrong. Frankly Eli has no horse in that race.

Mark B. said...

Your conclusions disagree with my prejudices, therefore you must be wrong. ;-)

Post a Comment