08 February 2011

Two New World Bank Papers on Climate Extremes

The World Bank has a new paper out by Robert Mendelsohn and Gokay Saher (here in PDF) that focuses on:
the global impact of climate change from several extreme events: local storms, heat waves, cold spells, floods, and droughts.
They conclude that with respect to these events:
Given empirical evidence about the link between climate and damages, climate change is calculated to increase the damages from these five extreme events by between $11 and $16 billion a year by 2100. There is little supporting evidence that climate affects deaths from these events (except for the possibility of local storm deaths increasing).
Note that work that I have been involved in on floods assumes that damage increases proportionally to increases in population and wealth, whereas they find a relationship that is less than proportional. Their findings are of a very small impact (0.015% of global annual GDP when combined with the tropical cyclone results discussed below).  They contrast this result with that of the 2006 UK Stern Review (which I have also criticized here in PDF):
These values are completely consistent with estimates in the literature per extreme event. However, they are completely inconsistent with values stated by Stern (2006) who suggests that extreme event damages could be 0.5 to 1.0 percent of GWP by 2050. Oral statements by Lord Stern even suggest values as high as 5 percent of GWP by 2200. The Stern analysis has been criticized because it confuses changes caused by what is in harms’ way (baseline changes) with what is caused by climate change (Pielke 2007b). But even this mistake cannot justify the estimates by Lord Stern. The hypothesized damages quoted by Lord Stern are completely inconsistent with empirical evidence.
A second study, by Medelsohn, Kerry Emanuel and Shun Chonabayashi (here in PDF), focuses on the future impact from tropical cyclones. They conclude that
Using the minimum pressure damage model, the estimated impact of climate change on tropical storm damages ranges from $28 to $68 billion USD/yr (0.005 to 0.012 percent of GWP) by 2100. This represents an increase of between 50 percent and 122 percent over future baseline levels. Climate change is expected to double the damages from tropical cyclones by 2100 by $54 billion USD/yr. The findings confirm the results of earlier tropical cyclone studies that relied on cruder methods.
Score one for cruder methods (here is one such study in PDF).  They also project that deaths from tropical cyclones will fall, due to projected changes in Myanmar and Bangladesh. 

How much confidence should we place in the projections of the future from these studies? Make no mistake, they are quality research projects from top scholars, but the answer is essentially zero confidence. These studies tell us something about what is possible, not probable. Both papers are very up front about inherent uncertainties -- as the first paper notes:
All the estimates presented in this analysis are inherently uncertain. Each facet of the integrated assessment model is uncertain. The emissions scenario, the climate scenario, the change in extreme events, and how damages might change are all uncertain. A sensitivity analysis is performed for the five events in this study. The results tend to be very sensitive to the climate scenario and assumptions about the damage function. Unfortunately, the analysis could not test the importance of the emission trajectory or the link between climate and extreme events.
What this inherent uncertainty -- perhaps better characterized as inherent ignorance -- means of course is that strategies of robust decision making make good sense. That is to say, with respect to preparing for future extreme events, we should emphasize those strategies that are insensitive to uncertainties. Perhaps the most robust finding of each of these studies (and the broader literature) is that future damages will increase regardless of the the effects of human-caused climate change. Thus, improving adaptive capacity is a no-regrets course of action.

Interestingly, the findings of the paper on tropical cyclones project a doubling of damages by 2100, a scenario that we explored in a recent paper, finding that under such a projection, detection of a human-caused climate signal would take many, many decades at best, perhaps centuries. These studies underscore the fact that efforts to try to pin claims of attribution of recent events to greenhouse gas emissions are empirically groundless, even if symbolically and emotionally satisfying. We are going to have to proceed into the future without knowing the influence of greenhouse gas emissions on extremes. That uncertainty need not stand in the way of making effective decisions about adaptation and mitigation.