The risk models that helped drive one of the most lucrative bets of 2023 are increasingly being tested by smaller weather shocks fueled by climate change.
Catastrophe bonds and other insurance-linked securities, which powered last year’s highest-returning hedge fund strategy, are built on calculations that can underestimate a new breed of risk stemming from high-frequency events such as wildfires and thunderstorms, according to veteran investors.
Elementum Advisors LLC, a $3.6 billion investment manager specializing in cat bonds and other ILS products, says it’s had to devote considerable time and resources to refining the wildfire model it licensed just a few years ago.
It was “benchmarked to historical trends and not to today’s climate,” said Jake Weber, Elementum’s head of data and analytics.
After analyzing data from almost two million US wildfires, Elementum saw a “statistically significant, higher frequency of areas that were burned in northern California” than the model indicated, Weber said. In the end, the investment manager was able to come up with more accurate wildfire estimates, which helped it negotiate higher interest rates on deals.
Cat bonds, now a $47 billion market, were devised to allow insurers to transfer the financial risk from rare but highly destructive natural disasters to the capital markets. Investors get access to potentially huge returns if a pre-defined catastrophe doesn’t hit, but the bonds can be wiped out if it does.
Last year, cat bonds soared 20% in value, leading hedge funds like Fermat Capital Management to generate their best-ever results. Less specialized investors have started to buy cat bonds and European regulators are even considering allowing retail investors to hold them. Insurers, meanwhile, are issuing new cat bonds at a “record-setting pace,” according to ILS researcher Artemis.
Pricing the risk just right is what it’s all about for cat-bond investors, and some of them are brilliant at it. But the exercise is getting harder.
Instead of giant earthquakes and hurricanes, insurers are increasingly plagued by cumulative losses from smaller, more-frequent events. These so-called secondary perils, including wildfires, floods and thunderstorms, are being exacerbated by global warming.
Last year, the hottest on record, secondary perils accounted for 86% of global insurance losses, according to insurance broker Aon Plc. Mid-sized events that cause $1 billion to $5 billion in losses are now the fastest-growing type of natural disaster, says Swiss Re, the world’s second-largest reinsurer after Munich Re.
This shift is a concern for cat-bond investors. Twelve Capital, which invests in the securities, says the basic models frequently underestimate losses from secondary perils. Tenax Capital, another specialist ILS investor, warns of the growing hazards associated with secondary perils that are often bundled into cat bonds in a way that makes it hard to estimate the risk implications.
“The disconnect between modeled risk probabilities and bond spreads, especially where secondary perils are involved, indicates that models have room for improvement,” Tenax said in a statement.
About 40% of today’s cat-bond market is made up of securities that cover aggregate losses, which is where investors are most likely to feel the fallout of secondary perils, according to Artemis, which tracks unusual insurance strategies.
Moody’s Insurance Solutions, a risk modeler for the insurance industry, says buying cat bonds today without devoting serious time to understanding the dynamics of secondary perils is a bad idea.
“We encourage all our customers to have their own view of risks, using the best possible science,” said Ben Brookes, managing director for consulting services at Moody’s Insurance Solutions.
Weber says Elementum is now doing its best to make up for holes in existing models. For larger storms, for example, “we’ll do our own risk assessment, if we see the wind losses have been quantified, but not the risk of flooding,” he said.
Verisk Analytics Inc., another big catastrophe modeler, says its offering is improving. The insurance-risk specialist updated its model for severe convective storms — a form of secondary peril — in 2022.
The company used machine-learning techniques to analyze two decades’ worth of radar data, “making it one of our most intensive updates,” said Adil Imani, director of ILS at Verisk Extreme Event Solutions. He said Verisk now knows “every single point of every single event for 20 years.”
The revised model reveals a meaningfully altered picture of how thunderstorm risk is evolving. According to Verisk, severe thunderstorms in the US are now affecting areas somewhat to the south and east of what’s known as Tornado Alley, and climate change looks set to further alter those dynamics.
Roger Grenier, senior vice president of Verisk’s global resilience practice, says its data points to a rise in severe storms.
“Over the long term, we expect to see a 10% increase in conditions favorable for storms to form,” he said.
Such information still needs to be combined with data capturing population density, so that modelers can figure out the likelihood of a bad storm affecting insured buildings. But things like property values, insurance deductibles and premiums aren’t disclosed in granular detail in the offer document of a typical cat bond. That’s because insurers that issue such bonds want to shield the data from rivals.
Weber of Elementum said some commercial models “are designed more for clients to fill out forms” for regulators, but aren’t that useful for cat bond investors deciding on a daily basis whether to buy or sell.
Commercial modelers “aren’t the practitioner that’s taking on the market risk,” Weber said. “We’ve often found their models wanting.”
Karen Clark, a pioneer of catastrophe modeling and chief executive of Karen Clark & Co. in Boston, said there’s clearly a need for “more advanced physical-modeling techniques,” which need to be supported by vast amounts of data and computing power. The firm updates its models every two years on average, mainly to account for climate change.
The secondary perils that are hardest to model are winter storms, such as the five-day freeze that crippled large parts of North America in February 2021, Clark said. To calculate the risk, researchers must assess three “hazard intensity footprints,” including wind, snow and ice, she said.
“Each sub-peril has to be modeled separately for each event and then the losses combined appropriately,” Clark said.
The enduring challenge will be to assess the forward-looking impact as the planet warms.
Relying on historical patterns of loss “is like driving a car by looking at the rear mirror,” said Maurizio Savina, who manages the development of climate models at Moody’s Insurance Solutions. “The future isn’t like the past.”