Tim Andrews and Tatiana Potemina analysed the keys to assessing climate change impacts for insurers at the 2021 All-Actuaries Virtual Summit General Insurance Plenary.
Bushfires – not all years are equal
We all know that the number of bushfire events varies from year to year. But it’s not so obvious what drives this variability.
The Black Summer of 2020 followed exceptionally dry six months across most of Australia. One of the contributors to the dry weather was cooler than average sea surface temperatures in the Indian Ocean to the north west of Australia (reflecting the Indian Ocean Dipole in its positive state – known as ‘pIOD’). Similarly, both the Ash Wednesday (1983) and Black Saturday (2009) bushfires were preceded by a pIOD, leading to less rainfall in the lead up to the fire season.
The phase of the Indian Ocean Dipole (IOD) is important to understanding the bushfire risk in a particular year in Australia, but it’s not only the IOD that is important. Both the Black Summer and Ash Wednesday fires occurred during El Niño episodes, which reflect warmer than average sea surface temperatures in the central and eastern tropical Pacific Ocean and contribute to dry conditions on the east coast. So pIODs and El Niño episodes acting together increase the risk of bushfires even further.
In short, the frequency of pIOD and El Niño episodes are key to interpreting past bushfire trends, and how the frequency and intensity of pIOD and El Niño episodes change in future will be key to future bushfire trends.
And it’s not just bushfires that are affected by the IOD and El Niño.
Natural climate variability
The IOD and the El Niño Southern Oscillation (ENSO) are two of the key types of natural climate variability that affect Australia, but not the only ones. Many of the variations in our climate that we notice reflect natural climate variability of different timescales varying from weekly to tens of thousands of years. For example, if you live on the east coast you would know that the 2021 summer (did we even have a summer?) was much wetter than in the previous year. The 2021 year was a La Niña year which typically brings more rainfall to Australia. Whilst not the only cause of the wetter conditions, a wetter summer was entirely consistent with what the scientists were telling us would happen before the summer started.
The frequency of droughts, bushfires, cyclones, extratropical cyclones (including east coast lows), floods and thunderstorms are all affected to some extent by natural climate variability.
The link between natural climate variability and climate change impacts for insurers
The cost of natural hazard disaster claims for insurers between 2007 and 2020 was around double the cost in the period 1991 to 2006. When seen on a chart this may look like evidence of climate change impacts and some authors/presenters have placed this very interpretation on the data. However, when normalized for the impacts of natural climate variability a quite different assessment might be made. In short, any assessment of climate change impacts from review of industry claims trends needs to take account of natural climate variability or risks getting it wrong.
Key questions for actuaries
Some of the key questions that flow from natural climate variability for actuaries are:
How much of the historical change in natural hazard costs reflects climate change, and how much is due to natural climate variability?
It is probable that most of the variation that we have recently experienced in natural hazard costs reflect natural climate variability and random variability. It is likely that there is also a signal in the data from climate change, but that this is difficult to identify as it is swamped by the variability from the other two sources.
How will climate change affect natural climate variability and what will the consequences be?
The science is telling us to expect more of the extreme natural variability episodes, which typically bring extreme weather of some sort. Plus, it seems the ENSO and the IOD will more often be in their dry phase, which may mean more droughts and more bushfires.
Can we predict the forthcoming weather seasons and use this in our budgeting, pricing and reinsurance purchase decisions?
- Currently the prediction of future episodes is difficult beyond 6-12 months lead time. This limits its application for pricing and reinsurance, but may it be useful for setting natural peril allowances in the budget process. Machine learning methods applied to predict ENSO and IOD episodes have been on the rise and appear to deliver promising results.
CPD: Actuaries Institute Members can claim two CPD points for every hour of reading articles on Actuaries Digital.