Blog Post

How Much Do Weather-related Disasters Cost?

Sep 16, 2013 | Carolyn Kousky

Climate scientists predict many weather-related extreme events may be changing in frequency or intensity as the climate warms, or their location, timing, and duration may be changing.  The public also perceives a trend toward more extreme events which they pin on climate change: a 2012 poll of U.S. residents by researchers at Yale found that, by a margin of 2:1, people believe that the weather is getting worse, and a large majority believes climate change contributed to the severity of several recent natural disasters.

How costly are natural disasters?  And how worse are they getting?  It may seem like a simple task to add up the costs of a disaster event, but it is quite challenging in practice.  How the spatial and temporal boundaries are drawn can make a difference.  It is easy to double-count losses when funds are being transferred across many different groups.  And while some categories of damage are easy to observe and measure, others are difficult, such as loss of family heirlooms or the economic value of pain and suffering.  There are only a few reliable and comprehensive databases of disaster costs and they all likely underestimate the hypothetical true costs by failing to include hard-to-measure losses.

To obtain an estimate of the average worldwide costs of weather related extreme events—those most likely impacted by climate change—I collected yearly estimates of the total costs of weather-related disaster events globally between 2000 and 2012 from the four main institutions engaged in this exercise: Swiss Re, Munich Re, the Centre for Research on the Epidemiology of Disaster (EM-DAT database), and Aon Benfield.  These institutions pull together data from various sources to generate estimates of the direct costs of disasters globally.  They all have slightly different criteria for what counts as a “disaster” and use slightly different sources.  Their estimates for the global costs of weather-related natural disasters are shown in Figure 1. Across these four sources, the average annual cost worldwide for weather-related disasters over this time period ranges from over $94 billion (EM-DAT) to over $130 billion (Aon Benfield).

[caption id="attachment_6785" align="alignleft" width="465"]Disastercosts I would like to thank Steve Bowen at Aon Benfield, Bridget Carle at Swiss Re America, and Angelika Wirtz at Munich Reinsurance Company for supplying the data.[/caption]

I would like to thank Steve Bowen at Aon Benfield, Bridget Carle at Swiss Re America, and Angelika Wirtz at Munich Reinsurance Company for supplying the data.

A big question is whether disaster costs are increasing.  The short answer is yes (fatalities, on the other hand, have been trending down).  The question of why is more complicated.  Many studies have been undertaken to tease apart different drivers of the observed upward trend in inflation-adjusted disaster costs.  In these analyses, researchers standardize disaster losses by measures of wealth and population.  If the trend disappears for these standardized measures, it suggests the higher costs are due to more wealth locating in risky areas. The conclusion from all these papers is that the majority of the upward trend in losses observed to date is from more people and capital locating in hazardous locations, but even after correcting for this, there may be a small upward trend for certain regions and hazards, such that climate signals are beginning to be seen in historical loss data (for example, large thunderstorms in the Eastern US).  Of course, projections going forward are quite different than observing historical losses, which is what is done here.

What does the time trend look like for the four data sources in Figure 1? As a simple way to explore this, Figure 2 shows a fitted plot for each data series over the period 2000 to 2012.  The upward sloping lines suggest a trend toward higher losses.  Consistent with other findings that much of the trend can be explained by higher wealth, however, when these damage series are simply regressed on a time trend, it explains more of the variation of damages, as opposed to damages per unit wealth (using global GDP data from the World Bank).

[caption id="attachment_6780" align="alignright" width="457"]FittedTimeTrend Figure note: Fractional polynomial fit to each series.[/caption]

Figure note: Fractional polynomial fit to each series.

There are many opportunities for cost-effective investments in hazard mitigation, and as the climate changes the risk profile of many weather-related events, we may see more such actions being taken.  While it is easy to think of risk mitigation and adaptation to changing risks as one-off investments, such as installing hurricane shutters, or discrete policy changes, such as upgrading building codes, this fails to account for how deeply communities may be adapted to the current risks they face.  Changing institutions, such as emergency response frameworks or water rights, which may be designed for a certain climate, is long-term process.  And widespread adoption of even cost-effective, one-off measures can be a challenge, as up-front costs can be substantial and there is an on-going debate about who should bear these costs.

For more information, see:

Kousky, C. (2012). “Informing Climate Change Adaptation: A Review of the Economic Costs ofNatural Disasters, their Determinants, and Risk Reduction Options.” RFF Discussion Paper 12-28. Resources for the Future. Washington, DC, July.