Last month, the Paris climate change accord became active, and already we are starting to see some interesting research developing as a consequence of that international initiative. One of the more insightful studies I have come across is a project conducted by the Cambridge Centre for Risk Studies. The research, sponsored by the insurer Lloyd’s, is an attempt to model the “economic output at risk” for the top 300 cities around the world. The first version of this so-called “Lloyd’s City Risk Index” proposes something called the “GDP@Risk” index to analyze economic risk exposure for major population centers. Going through the team’s documentation makes for very interesting reading.
The Cambridge team describe the methodology for their analysis as follows:
We define GDP@Risk as the economic loss experienced by an economy over a ve-year period compared to the baseline economic trajectory (or “counterfactual”). It can be described mathematically as the sum of GDP over ve years in the baseline trajectory, minus the sum of GDP over ve years in the scenario where a catastrophe occurs. This is shown by the equation below where YB represents the baseline trajectory of economic output and YC represents economic output in the catastrophe scenario. These equations are represented by the two trajectories in Figure 1 where the shaded area represents the GDP@Risk.
The economic trajectory after a disaster has occurred is modelled using a number of key parameters. Two key parameters are the economic impact and the economic recovery.
The economic impact is an estimate of how much economic output decreases from the baseline trajectory before economic recovery begins. Economic recovery refers to the rate at which the economy recovers over time. Taken together, the economic impact and the economic recovery are used to estimate total economic loss or the expected loss in GDP for the given scenario.
The overall economic impact experienced by a disaster is proportional to the vulnerability of the economy to a particular disaster the size of the disaster and the exposure of the city. Economic recovery is proportional to the resilience of the economy and captures how well the economy is expected to recover from the disaster after it has occurred. Using this method it is possible to have a city that is vulnerable to a particular disaster with a corresponding large initial economic impact but with strong economic resilience, and which is therefore able to recover from a large economic impact relatively quickly. The opposite is also true: it is possible to have a city that is not vulnerable to a particular disaster but has low economic resilience so its economy will have a longer recovery period.
The final estimate of the expected GDP loss for each particular event, therefore, is a combination of assessed threat type, vulnerability and resilience. For computational reasons, the 300 cities were categorised separately based on their underlying characteristics. Each city was graded on a scale from 1 to 5, representing that city’s vulnerability to a particular disaster (e.g. 1. Very Strong, 2. Strong, 3. Moderate, 4. Weak, 5. Very Weak). This same process was also done for resilience assessment.
Each city was also given a Threat Assessment Grade (TAG) based on its level of exposure to each particular threat type. Again, for computational reasons, the cities were categorised based on their level of exposure.
For each threat type we consider three “characteristic scenarios” (Small, Medium and Large) categorised by the probability and severity of each catastrophe event. A small event has relatively high probability with corresponding low severity while a large event has low probability with corresponding high severity. Economic losses are truncated at 5 years but taken as a proportion of the total GDP loss over a 10-year horizon. Figure 2 exhibits the various components used in performing the GDP@Risk calculations.
The conclusions reached by the team at Cambridge are worth noting by anyone interested in the economics of climate change. First of all, the team calculated that “a total of $4.6trn of 301 cities’ projected GDP is at risk from all threats – out of a total projected GDP between 2015 and 2025 of $373trn.” This total represents about 1.5% of global GDP. Furthermore, the researchers note that the global risk landscape changing rapidly along three dimensions:
- Emerging economies will shoulder an increasing proportion of risk-related financial loss as a result of their accelerating economic growth – more than 70% of Total GDP@Risk is associated with emerging economies, with their cities often highly exposed to single natural catastrophes. Earthquake alone represents more than 50% of both Lima’s and Tehran’s Total GDP@Risk, for example.
- Manmade threats are becoming increasingly significant. Market crash, cyber attack, power outage and nuclear accident alone are associated with almost a third of Total GDP@Risk. Market crash puts the most GDP@Risk globally– representing nearly a quarter of all cities’ potential losses.
- New or emerging threats – cyber attack, human pandemic, plant epidemic and solar storm – are also having a growing impact. Together, they represent nearly a quarter of total GDP@Risk.
The breakdown of the specific risk exposures is below:
The city-by-city ranking is presented here in Figure 3:
Now, one may disagree with specific aspects of the Cambridge team’s approach, and they admit as much themselves; however, they also note that: “Irrespective of the precise values of GDP@Risk, the ranking of cities by GDP@Risk is quite stable. It is relatively little affected by sensitivity tests of the various threats.” This is an important quality of the work, which is why I think this research should have gotten wider coverage in the general business press. Moreover, this study is the first I have seen that attempts to lay out a global, quantitative, assessment of the economic value at risk faced by the top urban centers around the world.
Of course the Cambridge team did not just focus on climate risk, but it is notable that of the 18 threats analyzed, half of them are climate-related, so one might argue that climate risk has placed $2.3trn of global GDP at risk as of 2016. Of course, only the tiniest fraction of that amount is even close to being hedged in any way by markets, which suggests a massive global risk exposure with respect to climate change. This is a point I made in this blog back in 2014, when I wrote that in the future we may need an “IMF of Weather.” This study only reinforces that view. Indeed, a recent review of the insurance industry highlighted the gap between the economic value at risk and coverage today:
At the same time, there is a widening of the global protection gap – the divide between economic and insured losses. This is particularly the case in developing countries most vulnerable to climate impacts. The trends of climate change, the transition to a low carbon economy and a growing protection gap will require the insurance industry to evolve and broaden its role.
Should the GDP@Risk number continue to trend upward (and that is most likely) at the same time as insurance levels stay flat or increase only slightly, then sooner or later both private and public agents will reach a point at which the exposure levels are simply unacceptable. We are not far from those levels as we head into 2017, and I suspect that GDP@Risk is only the first of many such indices to come in the next ten years. As I often note to executives and students, we are currently moving from the “if and when” stage in the climate risk discussion to the “how and how much” stage. A through read of this important analysis by Cambridge only reinforces that conclusion.