The Search for Clarity: Businesses Tap Cat Modeler AIR for Insight

Climate change is having a profound impact on the world, with extreme weather events such as hurricanes becoming more intense and destructive. One of the most urgent questions the insurance industry is asking relates to climate change and whether catastrophe models reflect the near-term climate? AIR is now providing detailed model documentation for their clients which discuss in depth how models are incorporating the latest science and data to reflect the current climate risk. This information is helping insurance clients not only better assess their own view of risk but also respond to the increasing volume of questions from other stakeholders, including rating agencies, regulators, businesses and investors. However, most importantly the (re)insurance industry in concerned with quantifying how climate change risks could impact losses to their books of business.

Climate change was “arguably the most critical issue in the industry before the pandemic” and still presents a serious risk for the (re)insurance industry, said AIR president Bill Churney.

AIR’s 2020 Report on Global Modeled Catastrophe Losses estimated the average annual loss from catastrophes worldwide at almost $100 billion. Meanwhile, insurable losses are twice as high as this figure, and economic losses are estimated at more than $400 billion annually.

With the ongoing effects of climate change and continued development in disaster-prone areas exacerbating the problem, losses from such catastrophes will only increase.

“While climate is a key focus for the industry, we still need to recognize, as we see with disasters like Haiti, the major protection gaps that exist and we should not lose sight of addressing near-term resilience to such events,” Churney said.

“The effects of climate change still have so many unknowns, and there remains a considerable amount of uncertainty in how extreme event risk may evolve in a warmer climate,” he said. “Our research into hurricane risks has found that the growth in the number of stronger storms, and landfalling storms overall, can increase losses by as much as 20%, with greater increases expected for those areas already experiencing a large number of such storms.”

AIR is helping the industry understand the possible impacts of climate change on U.S. hurricane activity by participating in research and collaborative efforts aimed at adapting its models to reflect the changing climate. The results of this growing body of research are helping clients achieve a better understanding of the sensitivity of risk to climate change; make more informed business decisions; and better meet regulatory requirements. The results of these efforts are published articles and white papers to guide clients. A small sample from among the many publications includes:

• UK Windstorm and Climate Change report developed with the UK Met Office and the Association of British Insurers;

• Climate Change Impacts on Extreme Weather AIR’s comprehensive survey of research on the effects of climate change on extreme events;

• Weathering Hurricane Variability in a Warm Atlantic covering the issue of climate variability in understanding climate change;

• Quantifying the Impacts of Climate Change on U.S. Corn Yields novel research on past and projected future climate change on corn (maize) yields; and

• Quantifying the Impact from Climate Change on U.S. Hurricane Risk estimates of loss impacts conducted in partnership with the Brookings Institution.

AIR has also been engaged with the Bank of England Prudential Regulatory Authority (PRA) working groups on climate change and has prepared a document detailing guidance for clients computing the impact for three region-perils, for three time horizons and two climate scenarios. This work follows on previous guidance for PRA requirements in 2019.

Additionally, AIR is currently working on a new framework for climate risk modeling that will blend its existing physical and statistical approach with machine learning to better predict the likelihood and impact of such extreme weather events, while also incorporating the forecasted impact of climate change on various perils.

“The result will capture both large- and small-scale atmospheric features that give rise to extreme events, as well as both local and distant correlations in weather phenomena," Churney said.

AIR has long been an advocate of open data standards to help combat the issues that stem from increasingly frequent weather events, and in 1993 developed UNICEDE, a globally accepted industry data format to standardize insurance data exchange.

And as recently as March 2021 the modelling firm announced that it was making its Catastrophe Exposure Data Exchange (CEDE) database schema available as an open standard.

This will allow companies to make better risk management decisions and help combat the growing risk presented by natural disasters, Churney said.

“AIR’s exposure data schema is part of a broader industry call for open and transparent data formats, and we are proud to be a part of the solution,” he said. “Having an open data standard will allow the industry to adopt this common data format and more easily transfer exposure data across the insurance value chain.”

Indeed, this open and collaborative approach to modelling has led to AIR becoming the modeler of choice for the so-called ‘Class of 2020’, a group of (re)insurance and specialty companies formed last year that includes Conduit Re, Core Specialty, Inigo Ltd, Integral ILS Ltd, and Vantage Risk.

“As a new insurer focused on the Lloyd’s market, we determined very quickly that AIR’s models and software were able to provide solutions for all of our key specialty lines, including reinsurance and property insurance, and also provide in short order a valuable portfolio roll-up capability,” said Russell Merrett, CUO of Inigo Ltd.

And with the COVID-19 pandemic bringing the interconnected nature of risks to the fore, with the virus leading to increases in claims for business interruption losses, supply chain disruptions, and increased cyber attacks among others, AIR’s collaborative approach to modelling is bearing fruit and helping the business to push new frontiers.

“We are seeing a clear need to explicitly model these interconnections, which is why we are working with our colleagues at Verisk Analytics to develop databases of the networks that connect businesses to better understand the risks that they are exposed to,” Churney said.

One example of this is AIR is working on is developing robust tools for modeling how liability risk and cyber disruptions can propagate throughout the global supply chain.

Understanding the patterns and correlations across many data sets provides a more holistic viewpoint of the complex interdependencies that exists in today’s interconnected world, Churney said. "With better analytics to identify and measure these risks, we can find ways to close the protection gap and build resiliency against future risks," he said.

This focus on resilience is an important strategy for AIR, and the wider (re)insurance industry can learn from this approach, particularly those carriers who see an opportunity in the widening protection gap, Churney said.

For regions and perils covered by catastrophe models, the protection gap represents potential business growth opportunities for the insurance industry to offer essential protection to vulnerable home- and business-owners, Churney said. "But companies need a comprehensive view of catastrophe risk on a global scale in order to be able to properly model their exposures and ensure solvency in high-loss years.

“Businesses may even start using these modeling tools themselves, instead of solely relying on their insurers or brokers. The new insight this provides can then spur them to seek additional insurance protection, presenting additional growth opportunities for insurers who are prepared to serve this segment,” he said.

So while the climate change crisis will continue to present greater risks for the planet and those (re)insurers charged with protecting it, there are still positives to be found for those willing to rise to the challenge.

“A growing protection gap forces the burden of disaster recovery onto taxpayers, and we have seen firsthand in recent years the enormous toll that catastrophes can take on underinsured economies such as Haiti and Nepal," Churney said.

“But by working with industry in an open and collaborative way, we are looking to find the opportunities out there that will enable the (re)insurance industry to not only grow, but also make the world’s underserved communities a more resilient place in which to live and do business in.”

AIR is leveraging technology and analytics to develop more sophisticated models by incorporating the power of AI and machine learning.

“AIR has already been investigating and leveraging machine learning techniques for a few years, and we’ve identified a few new areas of application,” Churney said. One is for enhancing AIR's exposure databases, which the company maintains for more than 100 countries around the world.

For example, AIR is training algorithms to identify high-rise buildings from the patterns of shadows in satellite imagery. “In regions where this information is not typically collected, the availability of an alternate method to accurately identify these high-value buildings during disasters is crucial for reliable modeling results.”

The re/insurance industry is also looking toward technology to improve workflow efficiencies and to automate processes to make better decisions more quickly. Companies like AIR and Verisk are leading the way by integrating their suite of applications to make things more seamless for the client as one end-to-end solution. For example, they have integrated AIR loss and hazard analytics into Sequel Impact and data can be seamlessly passed between the two. AIR can also pass output directly into Analyze Re and they are leveraging the high performance Analyze Re engine to power portfolio rollups in Touchstone Re. These product integrations are supplemented by AIR’s Data Services practice, which can format data not only for Touchstone, but also Sequel Impact and AIR’s suite of APIs.

“Our vision is to bring an all lines big data warehouse with the ability to ingest various data sets, perform lightning fast analytics on complex terms, along with dynamic and powerful visualization capabilities,” Churney said.

Future State of Models

One of the key developments over the past few years is the growing desire for non-modelers to create and/or modify models.

In the past, the industry recognized that the modeling firms held a certain amount of expertise that gave them both the experience and the skill to develop credible models. To be sure, model development is no easy task – the modeling firms employ large teams of scientists, statisticians, and engineers with hundreds of combined years of education and training to study all the available research, synthesize it and create complex and realistic models that the industry can rely on to effectively understand and manage their risk.

Now, this has begun to change. “We fully understand the desire that some organizations that keep their own very detailed claims histories, might want to make modifications to model output to help ensure that the model output aligns with their own experiences,” added Churney. To support this, AIR developed tools that enable companies to modify the losses produced by the models based on a number of factors including event intensity, location, and building characteristics.

While developing one’s own model may sound appealing, most organizations that have tried it have quickly discovered just how complicated it is to develop a credible model. To try to make this process a little less daunting, AIR developed a series of tools to facilitate the development of models that use AIR’s basic framework, that can, in turn, be run in their software platform and can use their financial model, so that at these new modelers can focus on the core hazard and vulnerability components without having to also develop a framework for interpreting complex policy conditions.

This year’s catastrophes, set against the backdrop of an unprecedented pandemic that is still unfolding, are an important reminder of the persistent possibility of high loss years. The insured proportion of natural peril risk throughout the world is only a portion of the total economic loss. The difference between the two, known as the protection gap, means that much of the cost of large disasters is ultimately borne by the public, placing a tremendous strain on the financial well-being of governments and communities.

“Our mission as a catastrophe modeler is to always learn from recent events to help design innovative solutions that help to close the protection gap and improve the ability of companies, governments, and society to manage future extreme events,” Churney said.