|Home About Us Our Authors Us In The Media Links to Data Our Values Contact Us|
Disasters that Come and Go—But They Will be Back
Who hasn’t heard the phrase “in today’s dollars”? We all know that this refers to the price of goods being adjusted to reflect the passage of time.
For example, if we want to know whether cars have become more expensive since the 1950s, it would make very little sense to compare the sticker price of Holden‘s classic HR in 1966 ($2,167) to today’s Commodore (approximately $36,000) without adjusting for inflation. In fact, the 1966 HR would have cost $24,091 using 2010 dollars, the latest year that the Reserve Bank Inflation Calculator will handle.
Still, that’s about $10,000 less than today’s Commodore, so have cars become more expensive?
Comparing prices and costs across time is therefore not entirely trivial and it is easy to be led astray, in particular because any such adjustment is affected by a number of considerations that are not always spelled out. The preceding example adjusted for inflation but it did not adjust for “how much car” you are getting now, as opposed to 45 years ago. And when you consider the evolution of the quality of products over time, even a low-end Barina at $13,990 probably offers more value now (USB and iPod compatibility, wow!) than an HR did in 1966.
But how would one quantify the value of that USB and iPod compatibility? How much “more car” can you purchase now compared to 1966? This is a tricky issue and there are some ingenious solutions but I do not believe anyone would claim that it has been resolved altogether.
And that was the easy part.
Now let’s consider an issue that has caused considerable public and scientific debate, namely whether there is a “signal” from climate change in the recent acceleration of extreme weather events.
Are we experiencing climate change and is the number of recent freak weather events a sign of this increasing climate instability?
This question has no easy answer at the moment.
Let’s begin by counting the number of weather-related disasters around the globe during the last 30 years: They are represented by the increasing function in the figure below (taken from Neumayer & Bartel, 2011). There is no doubt that this number is increasing—in fact, the frequency of weather-related disasters has nearly tripled in the last 30 years.
The figure also contains a (virtually) flat line, which represents the number of geophysical disasters—that is, things like earthquakes and volcanic eruptions that are not affected by a changing climate. The fact that this line is nearly flat tells us that the increasing disaster count isn’t an uninteresting consequence of better accounting or reporting. If geophysical disasters had also been increasing dramatically, then that would suggest that the trend was likely due to factors other than climate change, such as better detection of events as technology improved or better reporting as more developing countries had the funds to keep accurate records.
Case closed? Climate change obviosuly led to an increasing number of weather disasters around the globe?
It is tempting to draw this conclusion but that would be premature. There is another way of looking at these data.
Partialing out Wealth: The Intriguing World of Normalization
Rather than merely counting events, insurance companies and other professional risk analysts are interested in finding out whether insured losses from disasters have increased over time. Considering losses has the additional advantage that unlike counting of disasters, there is no argument over what constitutes a “disaster”—whereas if you count events you need to know what counts as an event; how many uprooted trees does it take for an event to be a bona fide disaster?
However, when we turn to measuring losses, this is where the problem of “today’s dollars” and how to make comparisons between the qualities of a 1966 HR and today’s Commodore rears its ugly head in full complexity.
Not only do we have to use today’s dollars to look at disaster-related damage over time, but we have to account for other factors that might increase insurance losses for reasons unrelated to climate change. For example, more people may move into attractive areas—such as floodplanes or coastal strips—that are also disaster prone. Property values may rise faster than inflation, and as people get wealthier, more precious and costly objects are put in harm’s way. Insured disaster losses may increase over time for those reasons alone and without any additional contribution from climate change.
The challenge is to account for those other factors by statistical means, thereby removing them from the analysis to permit identification of the net effect of climate change on insured losses. This process is known as normalization.
It turns out that after normalization, most of the time—but not always; Barthel and Neumayer (2012)—there is no longer an increase in disaster-associated losses over time. If we adjust the earlier losses to today’s figures, similar to the way in which we expressed the price of a 1966 Holden HR in today’s dollars, then most researchers have thus far failed to find a climate-associated trend in the data (see Bouwer, 2011, for a review).
This is illustrated in the figure below, also taken from Neumayer and Bartel (2011), which shows temporal trends for Global losses from non-geophysical disasters after normalization, using two different normalization techniques. Neither trend line is significantly different from zero.
This is good news, of course, because it means that thus far the insurance industry does not have to cope with additional losses from climate change. (And we don’t have to cope with higher premiums.)
But is it the whole story?
No. And this is where things get to be particularly interesting.
We Don't Just Get Wealthier—We Also Advance Technologically
Because the moment one begins to “normalize” temporal trends by adjusting for relevant variables, intellectual rigor demands that one include all such relevant variables. That is, it is insufficient to include only variables related to increasing risk exposure (e.g., growing wealth, people moving into disaster-prone areas): One must also consider variables that decrease exposure to risk over time.
For example, today’s building standards render homes far more resistant to wind damage: Your grandfather’s fibro shed may have easily collapsed under a hefty breeze but your beach-front double brick structure is likely to withstand anything short of Hurricane Rambo. Likewise, your great-aunt couldn’t call in the cavalry when her farm was threatened by bush fire, whereas today you might get assistance from a pair of helicopters shortly after reporting a fire.
Most normalization research to date has not accounted for those variables because they are extremely difficult to quantify. (And most researchers have been at pains to point that out; e.g., Neumayer & Barthel, 2011, pp. 23-24.)
This failure to include relevant variables has drastic implications, because as noted by Nicholls (2011), it means that the absence of a loss trend rests on the assumption that there have been no improvements—zero, none, zip—in our ability to issue advanced warning of weather hazards, in our building codes, in our fire fighting abilities and so on. (Anthes et al., 2006, and Trenberth, 2010, have made similar points; though see Pielke et al., 2006).
In effect, normalization research to date largely rests on the oddly inconsistent pair of assumptions that (a) we have built up enormous wealth during the 20th century but (b) did so without any technological advance whatsoever.
So where does this leave us?
The issue of normalization of disaster losses is tricky and cannot be considered resolved. We urgently need research to quantify and account for the effects of better mitigation technology that has—so far—kept losses from disasters manageable compared to historical precedents.
Absent such quantification we can be sure of only one thing: If those variables had been accounted for during normalization, the observed loss trends would have been greater—how much greater is unknown at the present time, but greater for sure.
We also know that the raw numbers show a dramatic increase over the last few decades, and although it would be inadvisable to attribute all of that to climate change (after all, a hidden variable such as the number of leprechauns that are born on 17 March every year might also be responsible), it seems equally inadvisable to dismiss that trend altogether.
Evaluating Risk and Thinking About the Future
As usual, therefore, we are left with some uncertainty. The normalized trend of disaster losses is certainly underestimating the true state of affairs, and equally, the raw number of disasters comes with its own set of problems that may, if anything, overestimate the trend.
In light of that uncertainty, what could (and should) we conclude?
First, we must acknowledge that risk judgments and tradeoffs are inherently subjective and subject to preferences. Some people don't get on planes because the risk is too great (for them) while others not only get on planes but jump out of them because (for them) that's a risk worth taking.
However, that doesn't mean that evaluating risks is a free-for-all game without any rules: For example, just because there is uncertainty doesn't entitle one to be certain that there isn't a problem.
This is a crucial point, so let's revisit it without the double negatives: uncertainty cannot imply that a problem is certainly absent. So appealing to uncertainty about trends in disasters to conclude that we don't have a climate problem constitutes a fallacy of reasoning.
And in the context of weather-related disasters, a balanced assessment of risk would probably consider the following additional facts:
It would be unwise to conclude that we definitely don't have a problem.
Anthes, R. B.; Corell, R. W.; Holland, G.; Hurrell, J. W.; MacCracken, M. C. & Trenberth, K. E. (2006). Hurricanes and Global Warming—Potential Linkages and Consequences. Bulletin of the American Meteorological Society, 87, 623-628.
Barthel, F. & Neumayer, E. (2012, in press). A trend analysis of normalized insured damage from natural disasters. Climatic Change.
Bouwer, L. M. (2011). Have Disaster Losses Increased Due To Anthropogenic Climate Change? Bulletin of the American Meteorological Society, 92, 39-46.
Neumayer, E. & Barthel, F. (2011). Normalizing economic loss from natural disasters: A global analysis Global Environmental Change, 21, 13-24.
Nicholls, N. (2011). Comments on “Have disaster losses increased due to anthropogenic climate change?” Bulletin of the American Meteorological Society, 92, 791-793.
Pielke, R. A. J.; Landsea, C.; Mayfield, M.; Laver, J. & Pasch, R. (2006). Reply to ``Hurricanes and Global Warming—Potential Linkages and Consequences.'' Bulletin of the American Meteorological Society, 87, 628-631
Trenberth, K. E. (2010). Fixing the Planet? Science, 330, 1178-1179.
Comments 1 to 14:
Post a Comment
You need to be logged in to post a comment. Login via the left margin or register a new account.