12 November 2009

AIR Responds to My Critique of the ABI Report

Peter Dailey of AIR Worldwide, who led the analysis in the ABI report that I critiqued earlier today, has been kind enough to provide a response to my comments. Here they are in full, and I'll offer some reactions in the comment. But first, here is Peter, who I'd like to thank very much for providing a quick and thoughtful response:

Thanks for your recent blog comments on the ABI study. To avoid confusion, let me clarify a few things for you and your readers.

GDP growth was used as a proxy for increases in the number and value of properties insured, and the period of 10 years was selected to illustrate in a simple way how a relatively small annual growth rates (say from 2.5 to 6%) compound over time. We did not project 35 or 65 years of GDP because over that long of a period, uncertainties attributed to changing exposure concentrations in hazardous areas, changes in building codes and vulnerability, new or updated man-made defenses, and other factors including GDP itself will be significant and cannot be predicted over such an extended period. Also, public adaptation could also change, such as reversing migration patterns which currently trend toward areas of risk. All of these factors, and the interactions amongst them, are largely uncertain.

The report clearly points out that the time element of predicting climate change was removed by instead associating changes in risk to changes in global temperatures. Of course, the natural question arises, “When will these temperature changes occur?” There is an appendix specifically devoted to answering this and other related questions. The time frames associated with a 2°C (4°C) rise are quite wide, and by no means does the report conclude that 35-years (65-years) is the consensus. If we had computed a 35-year or 65-year growth rate based on current changes in GDP, some would have argued that the resulting changes are misleading for many reasons, not the least of which is the reality of the current economic crisis where many GDP rates are negative.

Perhaps it would have been more precise to say:

Over a 10 year period of 2.5% growth (as an estimate of GDP in the UK), the compounded increase is 28%. The sensitivity of the catastrophe model results to this level of change in insured values for the 4°C scenario is as follows:

Insured 100-year (1.0% annual probability) Great Britain flood losses would rise by 38%, compared to 30% without GDP growth considered

Over a 10 year period of 6.0% growth (as an estimate of GDP in China), the compounded increase is 80%. The sensitivity of the catastrophe model results to this level of change in insured values for the 4°C scenario is as follows:

Insured 100-year (1.0% annual probability) China typhoon losses would rise by 16%, compared to 9% without GDP growth considered

Even here, with all the percentages and probabilities involved, it can be confusing and the main point can be lost in the translation. To be clear, there was no intention to mislead. In fact, I hope you will find that the climate model and catastrophe model results are quite sensible. As you well know, bringing economic conditions into the analysis leads to a great deal of uncertainty. For example, regional growth changes will certainly not follow the UK average. While this study does address several socio-economic factors related to climate change, it will naturally lead to more questions, and to more research, both of which fuel a healthy debate.

One of the key goals of this study was to dig into these complex interactions between climate science and risk management more deeply than in the 2005 study, but trying to predict how all of the moving parts will change over many decades, and moreover how individuals will respond to these changes is beyond its scope. Certainly, I hope you agree that exposure trends which exist today cannot be extrapolated very far into the future just as exposure trends of the early to mid-1900s do not apply today.

We greatly appreciate your comments, and are always looking for ways to better communicate these highly complex but extremely important issues for industry and society.

2 comments:

  1. Mr. Pielke:

    Will have to agree with you that the report is apples and oranges.

    Plus the report suffers frailty from too many Known-knowns, Unknown-knowns, Known-unknowns, and Unknown-unknowns. It a nice attempt, but likely you would have to heavily discount the study and its results.

    A comprehensive maybe, kind of, sort of study. With comprehensive maybe, kind of, sort of results.

    ReplyDelete
  2. Hmmm...I'm still confused and agree that we're still dealing with apples and oranges. Pielke et al 2008 and even AIR-Worldwide has generally contended that based on societal factors in the U.S., hurricane damages will likely double every 10 years. How can AIR-Worldwide contend that growth of 28% in the U.K. economy only has an 8% effect on the 1.0% annual probability of loss? Or that growth of 80% in the Chinese economy only has a 7% effect on the 1.0% annual probability of loss?

    ReplyDelete