Human behavior biggest challenge in race to model pandemic risk
Capturing how individuals and governments respond to pandemics is one of the key challenges to developing risk models that could underpin future insurance solutions, according to specialists.
The race is on to develop risk models to support the development of pandemic insurance schemes, which are being discussed in several countries after COVID-19 exposed gaps in non-damage business interruption cover.
Insurers have said they cannot bear pandemic risk alone, so many of the solutions being discussed envisage public-private partnerships with governments acting as the insurers of last resort. Quantifying pandemic risk will be key to determining its insurability, and at what price, and therefore how the risk should be divided between the insurance industry and governments.
Models have already proven adept at helping the insurance industry and investors quantify life and health risks from pandemics and get comfortable with bearing the risk. The risk for insurance portion of the World Bank’s Pandemic Emergency Financing Facility, which paid out about $195.8 million for COVID-19, was modeled by AIR Worldwide, one of the big risk model vendors.
But in the COVID-19 pandemic, insured losses have largely come from the nonlife side. Robert Muir-Wood, chief research officer at risk modeling firm RMS, said “a lot of the harm is not a direct consequence of the disease. A lot of the harm is a direct consequence of the actions taken to control the outbreak.”
Factoring in government interventions, the public’s response, and how these might change in future is “very much a work in progress,” Colin Dutkiewicz, global head of life at broker Aon PLC’s reinsurance solutions division, said in an interview.
Any kind of industry standard for modeling nonlife pandemic risks has yet to emerge. “Bespoke is the best word to describe the current situation for pandemic modeling with regard to its connection to P&C risks,” Thomas Hettinger, managing director and strategic advisory leader at Marsh & McLennan Cos. Inc.-owned reinsurance broker Guy Carpenter & Co. LLC, said via email. “It is more common for individual users to use pandemic mortality, infection, and spread information from the models to drive insights around business interruption and other P&C coverages.”
The human element
The difficulty is that modeling government and individual behavior in pandemics is far from trivial. Regimes around the world have responded differently to the current pandemic, and these responses have produced a wide variation of outcomes.
Nita Madhav, CEO of epidemic risk modeling company Metabiota, said in an interview that one challenge was some decisions were based less on data and “more around political decision making and realities they may be facing.”
She said tension between epidemiological and political considerations was “something that we continue to learn about and work to incorporate into some of this modeling framework” and is “one of the most challenging aspects of what to model around epidemics.” Metabiota worked with Munich Re and Marsh on PathogenRX, a pandemic insurance solution launched in 2018.
A further complication is that models need to capture what could happen in future pandemics, as governments are likely to respond differently to the next pandemic based on what they learn from the current one. Muir-Wood said that while countries did not necessarily learn from each other’s responses to the current pandemic, they were likely to going forward.
“Even if you repeated COVID-19 in two years’ time, the responses would be quite different because a lot of this learning would be applied,” he said.
The world could be a very different place when the next outbreak hits. Doug Fullam, director, life markets at AIR Worldwide, said one of the influences on the economic impact of the current pandemic is that, thanks to technology, some industries have been able to function uninterrupted, which would not have been the case if the same event happened in 1995 or 1975.
“We need to be mindful of the fact that this event itself has its own unique characteristics and our world of technology is also creating its own world of economic loss specific to COVID-19,” he said.
Data, data everywhere
Risk models need data, and there will be no shortage of it from the current pandemic.
“We do not need years of data to make this modelling work for P&C insurance,” Hettinger said. “Data captured throughout this pandemic on spending, mobility, and human interaction is right at our fingertips in our phones.”
But analyzing all that information could be challenging.
“The data isn’t adequate,” said Dutkiewicz, explaining that despite the abundance of data, it is not possible to use artificial intelligence techniques to sift through it because it requires judgment to interpret. “You can produce data but you don’t have the context of that data, so there is a lot more work that has to go into making that data sensible.”
One positive is that the industry has foundations to work from. Muir-Wood said RMS had developed a pandemic business interruption model in 2008, which was put on hold because of a lack of demand for the associated cover.
“Because we had a basic framework for how to do this, our work has been to build on that understanding … do a lot of research, to extend and refine that capability,” adding that RMS’s capabilities were able to help inform pricing decisions “right now.” He cautioned, however, that “it may take a while” before nonlife pandemic modeling is a standard product that is licensed for use across the industry like hurricane models.
Hettinger said commercially available models could follow bespoke solutions “in a simple estimate, over the next 12 months, or faster, depending on how much pain we see from the final tally of this pandemic and how the governments respond to help in producing backstops.”
When models emerge could also be driven by how governments will require insurers to respond to future pandemics. Fulham said regulation is “going to drive a lot of changes or advancements or addenda to models,” and that it would push his and other companies “in terms of how they think about the modeling.”
Jaclyn Guerrero, Metabiota’s associate director of product, policy and partnerships said: “I do think some of the policy decisions are going to be very influential in determining next steps in the market.”