Climate is changing the insurance risk landscape. Or is it?
Article information and share options
The annual sigma report from Swiss Re Institute on the topic of Natural Catastrophes and Man-made Disasters has just been released.
This year's report delves into the drivers behind the global growth in observed nat cat losses. One of the key drivers of growing economic losses is what the authors have termed 'economic accumulation', being the combination of increasing concentration of built assets in urban environments (more than 50% of the global population now lives in urban areas) and increasing individual prosperity. Increased wealth and prosperity also translate into growth in insurance penetration, as individuals and businesses seek to protect their assets and wealth. Whilst developed market non-life insurance penetration remains relatively flat, we see strong growth in insurance penetration in emerging markets, rising from 0.3% of GDP in 1990 to 1.5% of GDP in 2018. Despite this, Swiss Re's analysis shows that the 'protection gap' (being the difference between economic and insured losses) continues to grow, as the growth in total catastrophe losses outpaces the growth of insured losses. This gap represents a significant growth opportunity for our industry as both risk and prosperity increase.
The other focus area of the report relates to climate change, and how it will intersect with economic accumulation to grow risk even further in the future.
We already have a high degree of certainty around some of the effects (and observations) relating to climate change. For example, global average temperatures have increased by 1.1 degrees since 1880 (even more in Australia), polar ice caps are melting, and sea levels are rising.
The most extreme weather phenomena are also directly or indirectly influenced by the temperature of the atmosphere and oceans. A warming climate may cause stronger cyclones; more intense rainfall and flooding (especially flash flood); more intense convective storms; longer and more severe bushfire seasons. Further, we can expect to see events occur in areas where they have previously not been experienced – cyclone and hail zones are expected to move poleward, and bushfires will occur in regions not previously vulnerable for fires.
We are however considering very complex and interconnected atmospheric systems, and so not all of these future outcomes are as certain as the observed increase in temperature, melting ice, and sea level rise mentioned previously. The sigma report goes into much more detail on this topic, discussing how our risk assessments into the future need to not only take consideration of the frequency and severity of events. Looking forward, we also need to consider the confidence/uncertainty level around the changes, and an appropriate time horizon relating to the degree of change.
I highly recommend this insightful sigma report to anyone who is required to understand natural catastrophe risk assessment, now and into the future.
Fire, ice, and floods – the Australian Summer that had it all. Was climate change to blame?
Insurers and reinsurers will no doubt be left wondering whether climate change played a role in the catastrophe losses of this summer, from which the insured losses have now exceeded AUD 4.5 bn in total.
Extreme climatic conditions through spring, then December and January, provided the perfect conditions for a devastating bushfire season – the worst Australia has experienced thus far. Millions of hectares burnt, many lives lost, more than 3000 houses destroyed, and claims now exceeding AUD 2.2 billion (ICA). The direct and indirect economic losses for this summer's fires will be significant. (For more observations refer to the special report on the Australian bushfires which is appended to this year's sigma report).
Climate change was a topic frequently raised as large swathes of the country burned for more than 200 consecutive days. Reading the summary page of the BOM Special Climate Statement 73 (covering the extreme heat and fire weather of December and January), one might think that we have reached a tipping point for the Australian climate. I summarise some of the key points here1:
- 18 December 2019 was the hottest Australia-wide day on record, peaking at 41.88 °C.
- In December 2019 there were 11 days in which the national area-averaged maximum was 40 °C or above. Prior to December 2019 there had been only 11 such days recorded since 1910, seven of which occurred in the summer of 2018–19.
- In 2019, large areas of Australia had their highest accumulated Forest Fire Danger Index (FFDI) for December. FFDI records date back to 1950.
These are indeed convincing data points which reinforce existing climate trends. However, we need to understand how much of a role climate change might have played in this year's fire season. A recent attribution study2 tries to answer this question. Firstly, the study found that 2019's observed Fire Weather Index (FWI) has a 3% probability of occurring in any given year under current climate conditions. In simple terms, this year was a relatively rare event and can be largely put down to natural variation, driven by an active Indian Ocean Dipole and Southern Annular Mode. However, the study also found that the occurrence probability of a season with a FWI as high as 2019/20 has increased by at least 30% due to climate change (relative to 1900 climate levels). The study goes on to suggest that, under a 2-degree global average temperature rise scenario, we could expect at least a four-fold increase in the probability of such a season. Studies like this highlight how insurance risk models need to take account of future projections if we are to have any hope of correctly assessing the future cost of bushfires in Australia under a changing climate. If we don't, there is a strong chance of under-pricing future bushfire risk – a situation that could quickly become unsustainable for insurers and reinsurers alike.
The extreme heat and dryness of this summer was broken by monsoon-related low-pressure troughs in mid-January, bringing moisture and atmospheric instability – perfect pre-conditions for the formation of severe convective storms. Hailstorms then impacted Melbourne, Canberra and Sydney, resulting in an insured loss which is now approaching AUD 1.2 billion incurred (ICA).
Confidence in future hail predictions varies from low to high, but the underlying projections mostly point to more large hail. Future trends aside, recent Australian experience shows an increased frequency of large hail losses (events with losses >AUD 1 billion) in the last 10 or so years. With each additional event, we observe a widening gap between hailstorm loss experience and what the risk models tell us. Earlier in the decade, as the losses occurred, I wondered whether this experience was just natural variability. Several years later it would seem we have enough evidence to suggest an increasing trend in the frequency of larger hail losses in Australia. To take account of this gap, we bolster our modelled loss view with experience-based approaches. Regardless of what drives this trend (be it climate or economic accumulation), insurers and reinsurers need to ensure that their pricing takes account of these observations.
The east coast of Australia was impacted by an intense low-pressure trough on 8-9 February, resulting in insured losses that may eventually approach AUD 1bn (currently AUD 676m incurred – ICA).
It seems a strange juxtaposition that, with the heat and drought suffered earlier in the summer, rainfall records were broken in some locations later in the same summer.3 We can expect (with medium to high confidence) that more extreme rainfall and more variability are two changes that will occur under a changing climate. Record-breaking events, like we saw in February, will occur more often, and some experts have suggested that climate change could drive more flood events like those we experienced in Townsville in 2018.4
With all of the above in mind, I leave the reader with a quote from the sigma report, which I once again recommend you read in full (it also has some tips on how to address the issue raised in this quote):
The foremost underwriting risk in the context of climate change and all other macro-risk trends, is potential underestimation of premiums by relying on historical loss data or incomplete/outdated models to assess the current risk…. If risk assessment is based on a long-term historical average of a physical hazard, a historical bias can create a gap between actual and modelled risk. This is true for both low-frequency perils than for high frequency, potentially leading to risk views that lag decades behind the current risk landscape.