Long is the road to learning by precepts, but short and successful by examples. ~ Seneca the Younger.
I am an unrepentant data geek. One facet of my geek-ness is that I am autodidactic (I actually had a Professor call me that – I had to look it up) – I seldom accept others’ conclusions; I have to see for myself (actually, the Professor said I had to learn from my own mistakes, which I sometimes do). About 10 years ago, I first stumbled over FM’s Resilience Index. Now in its 12th year, it is a composite of 18 different indicators.
I posted about it at the time; lots of graphs, but I didn’t really put them into a useful context. In this post, I want to take a look at hints that they may have for those of us trying to understand a community’s resilience, in particular factors that we should consider in the resilience indices so prevalent in the literature and in use in the US.
The variables. FM is an insurance company. So “resilience” has to do with physical phenomena – natural hazards and climate change, as examples. It bins the 18 variables included in the Index into two categories: Physical factors and Macro factors. The Physical factors, in effect risk factors, rely on FM’s experience in each country, except for the cybersecurity data. The Macro factors might be considered as those attributes related to recovery from a natural disaster, i.e., resilience factors. If you’re interested in the data sources and methodology, follow this link.
Whenever possible, the data are averaged over a five-year period. This is something that is generally not done for most (any?) of the US resilience indices. The advantage of this is that it smooths out some of the inevitable noise in the data while maintaining evidence of a significant trend.
All of the Macro factors that involve money are adjusted for Purchasing Power Parity (PPP). The intent is to remove cost-of-living differences from comparisons. For the most common resilience indicator systems in the US, this has not been done. Thus, California counties (or other units) are indicated as more resilient than they really are because important data such as median household incomes are not adjusted for the very high cost of living (CoL) in CA. Using poverty values not adjusted for CoL, the number of people living below the poverty line in CA is less than the US average. However, once the value is adjusted, California has the highest fraction of its population living in poverty of all the states. In this context, it’s not surprising that it’s taking so long to rebuild Pacific Palisades!
Physical factors
Climate risk exposure – the portion of the country’s economically productive area exposed to climatic risks today.
Climate change exposure – the portion of the country’s economically productive area exposed to climatic risks in 2050.
Climate risk quality – enforcement of building codes for wind (90% of the indicator), and mitigation of flood and wind impacts.
Seismic risk exposure – the portion of the country’s economically productive area exposed to seismic risks.
Cybersecurity – commitment as shown in action (80%), and risk reduction relative to risk.
Fire risk quality – enforcement of fire codes (80%), and risk reduction relative to risk.
I haven’t seen the proportion of economically productive area to determine exposure to hazards used before. In the US, we either don’t include exposure data in our resilience indices, or else use something like the HAZUS code to calculate hazard losses (as is done for FEMA’s Community Resilience / National Risk Index). We certainly don’t include projections of risks in the year 2050. We also don’t include fire risks to the built environment as is done here, nor effectively give credit for mitigating actions.
Macro factors:
Control of corruption – perceived amount of corruption (public resources used for private gain) as well as “capture of state by elites and private interests.”
Education – average of expected years of schooling and the mean of actual schooling.
Energy intensity – energy consumption divided by the adjusted gross domestic product.
Greenhouse gas emissions – emissions divided by the adjusted gross domestic product.
Health expenditure – mean expenditure on health per person, both public and private, adjusted for PPP.
Inflation – annual rate of inflation.
Internet usage – fraction of the population using the internet.
Logistics – how easy it is to export to a target country in terms of the quality of infrastructure, the quality and availability of logistics activities, and public sector bottlenecks; based on survey data.
Political risk – perceived likelihood that the national government will be either destabilized or overthrown, either unlawfully or by violence.
Productivity – GDP (adjusted for PPP) per capita.
Urbanization rate – on an annual basis.
Water stress – freshwater withdrawn as a fraction of available resources.
Each factor was statistically massaged so that they were on a common scale (0-100). The resilience index for each country is then the mean of the 18 values. In contrast, in FEMA ‘s resilience index, the exposure (calculated via HAZUS) is divided into the Macro factors.
I took this data and mapped each factor against the resilience index and against each other. I won’t clutter this too-long post up any further with a bunch of graphs. The results are summarized in the following table where I’ve looked at correlations among the variables. R2 is a measure of how well two variables are linearly correlated. I’ve arbitrarily chosen an R2 value of 0.5 as the threshold indicating a strong relationship. All of the strong relationships are listed in the table below. If anyone wants the complete set of correlation just let me know.
| Strong relationships R2 ≥ 0.5 | ||
| Resilience index | Control of corruption | 0.76 |
| Climate mitigation | 0.74 | |
| Productivity | 0.70 | |
| Education | 0.70 | |
| Logistics | 0.66 | |
| Fire mitigation | 0.65 | |
| Health expenditure | 0.57 | |
| Internet usage | 0.57 | |
| Productivity (GDP per capita) | Control of corruption | 0.65 |
| Logistics | 0.60 | |
| Climate risk mitigation | 0.57 | |
| Health expenditure | 0.53 | |
| Education | 0.52 | |
| Fire risk mitigation | 0.51 | |
| Internet usage | 0.50 | |
| Health expenditures | Climate mitigation | 0.56 |
| Education | Internet usage | 0.68 |
| Climate mitigation | 0.57 | |
| Fire mitigation | 0.52 | |
| Urbanization rate | 0.51 | |
| Control of corruption | 0.51 | |
| Political risk | Control of corruption | 0.55 |
| Control of corruption | Logistics | 0.66 |
| Climate mitigation | 0.54 | |
| Urbanization rate | Internet usage | 0.52 |
| Logistics | Fire mitigation | 0.63 |
| Climate mitigation | 0.57 | |
| Climate mitigation | Fire mitigation | 0.73 |
| Climate risk exposure | Climate change exposure | 0.50 |
The strongest correlation was between the resilience index and control of corruption. This factor is not considered in any of the commonly used resilience indices. In effect, we are ignoring the community’s governance/institutional capital as a factor in its resilience. The impact of official corruption on recovery from disaster is obvious. The news from Gaza bombards us daily with a reminder of how much corruption hinders recovery. And apparently misuse of $100 M in recovery funding is another factor hampering the Pacific Palisades recovery. The only index that considers this factor is Arup’s resilience index for the 100 Cities initiative. Based on its strong relationship to a country’s resilience, this factor deserves more attention. (As an aside, I compared FM’s “Control of corruption” data with the Corruption Perceptions Index from Transparency International. The two are determined rather differently; however, they are highly correlated R2 = 0.96, i.e., they apparently are reflecting the same thing!).
Logistics, internet usage and fire risk mitigation are all important factors strongly related to both resilience and productivity. None of them are currently included in common resilience indices. I have often said that resilience is a manifestation of a community’s strengths, not its vulnerabilities. Intuitively, the ability to move physical assets where they are needed is an important strength related to recovery. In a similar sense, internet usage facilitates movement of information across the community. More generally, this emphasizes the importance of dispatchable capital.
One surprise: exposure factors weren’t correlated with the corresponding “quality” factors, i.e., mitigation wasn’t related to exposure. While the two climate exposure factors were correlated, none of the exposure factors were correlated with any of the resilience factors. Similarly, greenhouse gas emissions were not correlated with any of the other variables.
This is the first time that FM has included cybersecurity. It doesn’t make any difference to the resilience index, and is not correlated with any of the other factors. It seems to be irrelevant to both resilience and natural hazards and fires.
There is a lot more that can be extracted from this data, but this post is long enough already. FM has provided a rather different window on resilience, pointing out the importance of variables not often considered when we look at our communities. I hope that those working to make their communities more resilient will include all of the community’s capital portfolio in their efforts – its logistics systems (physical capital), its information systems (social capital), and above all, how the community makes and implements decisions (governance/institutional capital).
