How to make the most of the Actuaries Climate Index

Since 2016, the CIA and its partners have released data from the North American Actuaries Climate Index (ACI) every quarter. The most recent release continues the tradition. But what can you do with this data? What does it mean? We caught up with members of the CIA’s Climate Change and Sustainability Committee (CCSC) to learn more about the ACI and help find out how you can make the most of it.

What is the Actuaries Climate Index (ACI)?

Sponsored by the American Academy of Actuaries, the Canadian Institute of Actuaries, the Casualty Actuarial Society, and the Society of Actuaries, the ACI is a monitoring tool designed to help inform actuaries, policy-makers, and the public in general about climate trends and some potential impacts of a changing climate on the United States and Canada. The index measures the frequency of extreme weather and the extent of sea level change. The six components of the index are:

  1. high temperatures
  2. low temperatures
  3. heavy precipitation
  4. drought (consecutive dry days)
  5. high wind
  6. coastal sea level

The ACI does not provide a direct measure of average temperatures or intensity of precipitation, but instead it tracks the change in the frequency of extreme weather events (above the 90th percentile or below the 10th percentile). For example, for high temperatures, the metric used is the frequency of temperatures above the 90th percentile experienced during the reference period of 1961 to 1990. The higher the ACI, the more frequent the occurrence of extreme temperatures and the less likely this is attributable to normal statistical deviations.

What does the ACI focus on?

The taxonomy for climate risks – physical, liability, and transition – was first proposed by Mark Carney in September 2015 while he was Governor of the Bank of England and Chairman of the Financial Stability Board. Within this taxonomy, the ACI focuses on the physical risks, those that arise from the increased frequency and severity of extreme weather events that damage property and disrupt trade, but excludes both transition risks (e.g., those related to decarbonization) and liability risks (e.g., those linked to claims of failure of organizations to mitigate or adapt to climate change).

The ACI is estimated retrospectively based on data collected for the six index components. It complements other indicators that focus on the causes and thus may contribute to some prospective views. These include, for example, estimates of the gap remaining to the “+2 degrees Celsius” target of the Paris Agreement, GHG emissions reported by countries, or measurements of CO2 atmospheric concentration published by the Scripps Institution of Oceanography.

Since the ACI is a measure of probability of extreme weather events it cannot be translated into degrees of warming, but can we quantify the probabilities attached to the various ACI values reported?

Not precisely, since we do not know the probability distribution of the ACI. However, using statistical theory, one can draw on the central limit theorem to obtain an approximation by assuming that the distribution of ACI values converges towards Normal distribution (Bell curve) as the number of data points increases. Indeed, the quarterly ACI is a rolling average over 20 consecutive quarters, thus comprising 60 monthly values, each being an average of six components calculated from daily readings collected from over 300 grid points in 12 regions in the US and Canada combined.

During the three-month period ending in November 2020, the ACI for North America (US and Canada combined) has decreased from 1.24 to 1.22 while the ACI for Canada only has remained stable at 0.88. However, this does not mean the impacts of climate change for Canada are estimated to be 72% of the North American values or that temperatures have peaked and are now going down. The ACI is a rolling average of frequencies, not an average of temperatures.

My statistical theory is a bit rusty, can you explain how to estimate the ACI probabilities?

The ACI is the average of six component anomalies expressed in standard deviation units that have been normalized to an average of zero for the 1961–1990 reference period with a standard deviation of 1. Therefore, its own standard deviation (noted as σ) can be estimated to be equal to 1 divided by √6, which is 0.408.

Dividing 1.22 by 0.408 yields a value of 2.99, almost 3σ, which is very far in the right tail. A table of values for Normal distribution shows that the probability of getting a higher result than 3σ is only 0.14%. For 1.24, the probability would be reduced to 0.12%. In both cases, the ACI indicates it is very unlikely that the climate regime prevailing in the 1961–1990 period could yield such results, indirectly confirming that a change in the climate is more likely.

How different would this be for Canada only?

Following the same method, we find the probability of an ACI value exceeding 0.88 for Canada can be estimated as 1.54%, much higher but still 2.16σ, deep in the right tail of the distribution. Various factors may explain this lower ACI, one of which being that Canada experienced warming sooner than the USA and more variability produced a higher standard deviation during the reference period. As the Canadian anomalies were divided by a higher number, the normalization yielded lower index values. ACI values for different periods can be compared within a region but they cannot be compared directly between regions since the basic units calculated for the reference period are different for each region.

What conclusions can we draw from the ACI so far?

Overall, the ACI confirms that it is highly unlikely that the variations in the frequency of extreme weather events could be attributed to the natural variability of weather events as measured during the 1961–1990 reference period. The more likely explanation is that the trend in the ACI empirically confirms that climate changes are impacting climate-related risks (i.e., extreme weather).

How should actuaries use the ACI in their work?

The ACI can be used by actuaries in an educational context, for example in presentations to Boards of insurance companies to raise awareness of the reality of climate change and the need to reflect the associated risks in the company’s operations.

When the CIA decided to sponsor the ACI research project in 2016, it knew it was meant to be a first-level educational tool which would eventually lead to further research and development of more forward-looking metrics that actuaries could use for pricing and reserving work. Five years later, thanks in part to the ACI, more and more actuaries are aware of climate-related risks that may impact premiums, claims, or pension plan solvency.

The CCSC recently issued its Practice Resource Document: Climate Change Scenario and accompanying Excel tool to provide actuaries with practical considerations in building a climate scenario and for developing best practices in assessing the financial risks of climate change through climate scenario analysis. The CCSC also maintains a resources page with valuable research and information. The CCSC is dedicated to drive more research, conduct educational events, and liaise with regulatory bodies in order to assist actuaries in serving the public.