Roger Pielke Jr. recently published an article in Nate Silver's new 538 website titled Disasters Cost More Than Ever But Not Because of Climate Change. You should click the link and read his entire piece but, his thesis is essentially the following:
- The cost of disasters tied to extreme events is indeed rising rapidly - but once you normalize this to the increase in the wealth of the world as represented by global GDP (we own a lot more "stuff" now than before) - the trend basically vanishes
- Disasters lead to high loss of life where property damage is the lowest and highest property damage in places with lowest loss of life. As examples he cites the following: "U.S. hurricanes, for example, are responsible for 58 percent of the increase in the property losses in the Munich Re global dataset" and "Consider that since 1940 in the United States 3,322 people have died in 118 hurricanes that made landfall. Last year in a poor region of the Philippines, a single storm, Typhoon Hayain, killed twice as many people."
- Richer countries are in a better position to protect people's lives in times of disasters at the expense of higher property losses. He says: "As countries become richer, they are better able to deal with disasters — meaning more people are protected and fewer lose their lives. Increased property losses, it turns out, are a price worth paying"
His article was roundly criticized by some well known climate scientists, as summarized by Emily Atkin at Think Progress (additional discussion in this post by Laurence Lewis at Daily Kos). The crux of the criticism as I see it:
- Improvements to deal with disasters (e.g., newer building codes and improved construction, better forecasting, better preparedness and responses) over the years reduces disaster damages/costs (lives and property) and this is not properly reflected in Pielke's assertions - this has been pointed out to him before, yet he continues to publish his misleading claim
- Unlike Pielke's claims that somehow climate change is not causing an increasing frequency of extreme events, there are several situations where this is in fact the case
Pielke responded to these critiques in an email to Think Progress (same link as above):
Pielke particularly took issue with Mann’s claim that Pielke “completely ignores technological innovations (sturdier buildings, hurricane-resistant structures, better weather forecasting, etc.)” when analyzing disaster damage and its cause. Pielke said he has considered mitigation data in previous work, but has found through four of his own research papers not mentioned in the FiveThirtyEight article that there were no strong effects on the data.
In one paper, for example, Pielke says that even though stronger building codes have been shown to be able to reduce losses from hurricanes by more than 40 percent, those types of codes “have only been implemented in recent years and in some cases vary significantly on a county-by-county basis.” This means that stronger building codes are unlikely to change the results on overall loses, he says.
His studies found that data on mitigation efforts had an effect on losses from U.S. earthquakes and Australian tropical cyclones, but not on U.S. hurricanes, tornadoes, or global floods, he said.
“Seriously, there are lots of important issues to debate about climate change,” he said. “You guys really want to make a big deal out of this?”
While I agree that some degree of appropriate normalization would provide us a better understanding of the factors behind the sharply growing disaster related costs worldwide, there are several problems with Pielke's broad inferences.
First, as Brad Johnson stated in his comments:
It's rather confusing that Pielke Jr. is citing Neumayer-Barthel 2011 but excludes Neumayer-Barthel 2012. N-B 2011 is an important work, as it offers a much-needed correction to Pielke Jr's normalization methodology. Pielke Jr would have done well to include this N-B 2011 caution: "one cannot infer from our analysis that there have not been more frequent and/or more intensive weather-related natural disasters."
Pielke Jr's failure to cite N-B 2012 is notable primarily because that paper's findings provide evidence to contradict his bold thesis that "future changes in extremes that cause the most damage won’t be detectable in the statistics of weather (or damage) for many decades."
From the N-B 2012 abstract: "We find no significant trends at the global level, but we detect statistically significant upward trends in normalized insured losses from all non-geophysical disasters as well as from certain specific disaster types in the United States and West Germany."
Second, global GDP is a blunt instrument to use in normalization. GDP has many components - and some of them are not expected to play a big role in increased disaster costs. For instance, services, as opposed to goods, would be less subject to destruction from something like a hurricane - e.g., healthcare and financial services which form a meaningful portion of GDP. Transportation is a key part of GDP (e.g., automobiles, aircraft) - and with advance warning systems that allow for people to leave high risk zones earlier (in vehicles, planes etc.), it allows for increasingly lower levels of disaster damage to this sector of the economy. Electronics have become pervasive and ubiquitous among the "stuff" people own and are part of GDP growth - yet, increasingly these are smaller, more compact and easy to carry with you when you evacuate a disaster-threatened area. So, to truly understand what is the best method to normalize costs, one needs to find a better factor than just global GDP.
Third, residential and non-residential construction and structures are usually a significant fraction of GDP - especially in the last 10+ years across many economies where there has been rapid population and economic growth. Thus, a big chunk of global GDP growth is based on a substantial increase in newer buildings that are better equipped to deal with extreme events like hurricanes and torrential rain. These are precisely the types of buildings that would have benefited from better building codes that make them more resilient to disasters. However, Pielke divides increasing disaster costs by the increasing level of global GDP but simultaneously argues that improvements in many of the buildings underlying that growing GDP can be ignored for the purposes of his analysis. This is not an appropriate way to approach normalization.
Fourth, it's not clear that Pielke's broad inferences about property versus human loss are generally true - and more so, in the context of understanding disaster costs in relation to GDP. Let's take the examples he mentioned:
- "U.S. hurricanes...are responsible for 58 percent of the increase in the property losses" and "since 1940 in the United States 3,322 people have died in 118 hurricanes that made landfall"
- "Last year in a poor region of the Philippines, a single storm, Typhoon Hayain, killed twice as many people."
In the case of the US, the fact that property losses have ballooned despite the fact that the US has been at the forefront of driving disaster mitigation approaches - especially in building codes - should have been another cautionary point that Pielke's analysis did not consider seriously. In the case of Typhoon Haiyan, massive loss of human lives was also accompanied by massive devastation of property and land. In the fast growing and urbanized and increasingly economically strong city of Tacloban - which bore the brunt of Haiyan - the devastation was vast ("apocalyptic"). Thus, despite rapid urbanization and population growth, both property damage and the human death toll was extreme. Additionally, the vulnerability of the city to storm surges - which are expected to be more intense due to rising sea levels from climate change - has been well known. Pielke's normalization fails to account for this - one might conclude mistakenly from his inferences that climate change had no impact in causing a disaster of this scale. Even more importantly, the Philippine government estimated that the widespread damage to farms, buildings and infrastructure would result in roughly 0.5 percentage points lower GDP at the country level. Indeed, given the city was wrecked, the local GDP (if one were to measure and report it) would have more than collapsed. Thus, using a rising global GDP to offset global disaster costs to show a flat or declining ratio of the two, when at a local level the combination of a massive disaster cost divided by a much lower GDP would have resulted in a spike in the ratio, leads us to an inference that could be far from reality.
In summary, Pielke's analysis, if one might call it that, is too superficial and fraught with many issues - both in terms of incorrect data interpretation and also misalignment with other published research. It does not in any way allow us to conclude that rising disaster costs are not tied to climate change. Analyzing these costs requires a deeper understanding at the local scale.
Comments