Will climate change soon make Australia “uninsurable”?
To many investors, it’s indisputable: a warming climate has caused the number and severity of natural disasters to increase. Moreover, this supposedly rising tide of devastation threatens Australians and their insurers. These claims are being repeated increasingly frequently and fervently, yet a glaring weakness accompanies them: they almost invariably lack credible – indeed, often any – supporting evidence. In contrast, I’ve analysed this country’s most authoritative source of relevant data, the Insurance Council of Australia’s Historical Catastrophe Database, and this article summarises my results.
The mainstream’s claims are baseless: in Australia since the late-1960s, natural disasters’ frequency hasn’t increased; nor has their severity. Indeed, mega-catastrophes’ normalised costs have fallen. Bluntly, there’s no increase to cause; consequently, climate change – man-made or otherwise – can’t be causing it.
These conclusions shouldn’t surprise anyone: they parallel the Intergovernmental Panel on Climate Change’s AR5 WG1 (2014) and AR6 WG1 (2021) assessments; they also match Warren Buffett’s. Yet they’ll disconcert many and anger some investors. Once again (as during the Dot Com Bubble, etc.), they’ve allowed passion to overrule reason and drive their actions. That’s a recipe for severe disappointment (at best) and enormous misallocation, impairment and destruction of capital (at worst).
Today’s consensus
The mainstream – including insurers – increasingly regards insurance as risky. That’s because climate change is allegedly bisecting the economy into winners and losers – and insurers occupy the wrong side of a widening divide. In “Insurance Is a Climate Canary” (The Australian, 20 August 2019), for example, Alan Kohler wrote:
Last year, IAG said if global temperatures rose by four degrees, the world would become “pretty much uninsurable.” Queensland would be virtually uninsurable in a three-degree higher world ... the effects of this could be felt quite soon ... At some point soon, insurance will become expensive and hard to buy. Governments and companies need to ... through what happens if we lose the insurance industry entirely.
Despite the extreme pessimism – indeed, nihilism – of Kohler’s contention, he’s hardly a lonely voice in the wilderness. The Australian (1 December 2021) summarised the prevailing wisdom: “insurers face higher risks as climate change hits home” (see also “QBE Chief Warns of Climate Risks,” The Weekend Australian, 20-21 November 2021). On 7 October it elaborated:
risks a catastrophically expensive future, with natural disasters set to cost the country $73 billion a year by 2060. Natural disasters cost Australians $39 billion a year, but Deloitte Access Economics warned without steps to keep global warming below 3C the nation risks a massive explosion in damage.
... report warns if temperature rises are not contained Australia risks [costs of as much as $94 billion] a year from natural disasters arising from climate change ... The report warns southeast Queensland and northern NSW are the most heavily endangered parts of Australia due to a combination of worsened potential disasters including bushfires, floods and cyclones.
From these and many similar assertions, we can distil the mainstream’s two key assertions. As a result of climate change,
- The frequency of natural disasters in Australia has been rising;
- The cost of claims to insurers arising from these events has been increasing.
The ICA’s historical catastrophe database
“Factfulness,” wrote Hans Rosling in his book Factfulness: Ten Reasons We’re Wrong About the World – and Why Things Are Better Than You Think (Sceptre, 2018), “is the ... habit of only carrying opinions for which you have strong supporting facts.” Such facts emerge from dispassionate analyses of reputable data. And value investors know that whenever “experts” and the herd stampede to one side of an issue, it can be instructive – and profitable – to undertake analyses that investigate alternatives. The ubiquity and increasingly adamant tone of the mainstream’s claims, plus the paucity of hard evidence that typically accompanies them and Leithner & Company’s adherence to Rosling’s adage, prompt me to ask: is the number and cost of natural disasters in Australia really increasing?
In order to answer these questions, I analysed this country’s most authoritative source of relevant data: the Insurance Council of Australia’s Historical Catastrophe Database. According to its website, the ICA’s Data Hub “collates data from multiple government agencies … on cyclone, flood, bushfire and wind exposure to over 14 million addresses in Australia. We also collect and manage the industry claim response to Natural Hazard Catastrophe Events that have occurred in Australia 1967.”
Conceptually, the definition of catastrophe is well-known and widely accepted; its application, however, isn’t obvious. How costly in terms of lives lost, property damaged and destroyed, etc., must an event be in order to qualify as a catastrophe? According to the U.S. Insurance Information Institute (III), “a catastrophe is an unusually severe natural or man-made disaster that results in potential insurance claims in excess of $25 million.” Lacking guidance from the ICA, I adopted this definition: an NHCE is (1) one of the 707 events (as of 9 December 2021) that appear on the ICA’s Historical Catastrophe Database that also (2) generated claims to insurers of at least $1.
Are natural disasters becoming more frequent in Australia?
I grouped the 302 events that meet this definition into their financial year of occurrence and summed the number of events per FY (all subsequent references to “year” mean FY). Figure 1 plots the results. At first glance, the number of NHCEs appears to be increasing over time. The average is 5.6 per year; before the mid-1990s, the annual number was usually below-average; since then, it’s mostly been above the mean, especially in 2016 (but not, it’s worth noting, in 2020 or 2021: their numbers are the lowest in almost 20 years). As a result, the trend line slopes significantly upwards – and its rate of increase is accelerating.
Figure 1: Number of Natural Hazard Catastrophe Events per Year, 1967-2021
Detailed examination of these data, however, reveals that (1) the ICA’s definition has become more lax (i.e., its threshold lower) over time; (2) using a consistent definition, the number of NHCEs per year is stable.
Table 1 provides an example. It and Figure 2-Figure 7 use “normalised” insured loss data. In order to compare apples with apples over time, these data incorporate the effects of (a) the Consumer Price Index (2017=100), (b) the number and value of new buildings, dwellings, etc., and (c) improvements in construction standards. In other words, if a firestorm identical to the one on 7 February 1967 recurred today, then, given Tasmania’s greater population, infrastructure and number of dwellings, increased cost of the reconstruction as a result of inflation, stricter building codes, etc., its insured damage (measured in 2017 dollars) would be $2.2 billion. For details, see Benjamin McAneney, et al., “Normalised Insurance Losses from Australian Natural Disasters: 1966–2017,” Environmental Hazards, vol. 18, no. 5 (April 2019), pp. 414-433.
Table 1: Comparing the Top-Six NHCEs in Two Years
Table 1 ranks (by normalised insured damage) all of the NHCEs that the ICA recorded for 1967 and the top six of the 19 that it recorded in 2016. In the more recent years, almost all (17) of the NHCEs caused less than $25m of damage (and thus by the III’s criterion didn’t qualify as catastrophes). Indeed, 12 caused less than $10m of damage. In total, these 19 events caused $825 million of damage; on average, each event cost $43 million. Did NHCEs causing less than $25m of damage occur in 1967? It’s reasonable to suppose so, but the ICA doesn’t record them. Using the same yardstick for 1967 that prevailed in 2016, I suspect that more than six NHCEs occurred in 1967.
Yet the relatively small number of NHCEs recorded in 1967 nonetheless packed a terrible punch: in total, they wreaked $11.3 billion of damage. That’s almost 14 times the total in 2016; and on average, each NHCE caused $1.9 billion of damage (an astounding 44 times the average in 2016).
In 1967 the Gold Coast’s population (72,000) was just one-tenth of today’s (710,000) and Cyclone Dinah missed it by more than 100 kilometres. If it recurred today it would wreak almost $5 billion of damage and rank among the costliest natural disasters in Australian history (Figure 6). Together with the severe hailstorm that struck the region that year, their combined normalised cost ($6.3 billion) was double that of Brisbane’s flood in 1974.
Most tragically, the “Black Tuesday” inferno of 7 February 1967 destroyed huge swathes of Tasmania (from the Central Plateau to Hobart’s suburbs, including most of Mount Wellington, along the d’Entrecasteaux Channel and through the Derwent and Huon Valleys) – and killed 62 people, injured more than 900, destroyed an estimated 62,000 head of livestock and left more than 7,000 people homeless.
Clearly, some NHCEs are far more calamitous than others. Equally evidently, it’s easy – but flatly wrong – to assert that today’s disasters are “unprecedented.”
The key point in Table 1 generalises. Figure 2 plots the average damage per NHCE since 1967 – and corroborates my doubt that the actual (as opposed to the recorded) number of NHCEs is rising. Its mean is ca. $350 million and its trend is weakly negative. I don’t infer that NHCEs are becoming less destructive over time; rather, I infer (and will shortly confirm) that the damage they wreak is roughly constant and the ICA’s propensity to count them has risen. In other words, it didn’t record minor events in the 1960s and 1970s, but has done so since the 1990s.
Figure 2: Average Insured Damage (Normalised, Billions of $A) per Natural Hazard Catastrophe Event, 1967-2021
Figure 3, which plots the number of NHCEs per year causing less than $25 million of damage, confirms this hunch: thanks mostly to their very large number in 2016, the number of these events has increased significantly since the 1990s. McAneney et al. agree: “there has been a trend towards an increasing number of smaller events being included in the Disaster List and it might be timely that the ICA reconsiders its threshold cost for inclusion.” (Given that it included no such NHCEs in its database in 2020 and 2021, perhaps it has.)
Figure 3: Number of NHCEs, Normalised Damage < $25m, per Year, 1967-2021