In the following articles, the Insurance Information Institute’s Lucian McMahon highlights a few of the many outstanding sessions offered at the CAS 2019 Spring Meeting in New Orleans.
Wildfires: Could They Be Manageable Catastrophes?
With 11 major fires resulting in billions of insured losses, 2017 and 2018 were record-setting years for wildfires in the United States. The Camp Fire alone destroyed 18,800 structures and killed 88 people and is expected to be one of the largest insured loss events in U.S. history.
The sheer magnitude and unexpectedness of these events have roiled communities and their insurers. Many are worried that catastrophic wildfires could become the new normal. As Chris Folkman, senior director with RMS, put it during a CAS Spring Meeting session on wildfires, “From the 1960s to about 1990, wildfires just weren’t that big a deal to U.S. industry, but then something changed […] Over the past decade we’ve had an incredible increase in severe events.”
As wildfire risks continue to grow across the country, actuaries will play an important role in helping insurers and vulnerable communities respond to the threat.
A perfect storm for wildfires
Folkman argued that there are at least three major reasons for the increase in catastrophic wildfires, particularly in California and other western states.
Increased exposure: For one, wildfires have gotten so much more destructive because there is so much more to burn. Folkman noted that there are roughly 40% more houses and people in high-risk, fire-prone areas in the U.S. than just three decades ago.
20th century fire suppression tactics: On a related note, Folkman pointed to the aggressive fire suppression policy across the U.S. for much of the 20th century, which helped increase the fuel load in many states. Unfortunately for Smokey Bear, per Folkman, “A more balanced fire policy of thinning programs and controlled burns is a big part of good policy.” But this new approach did not gain traction until the 21st century. “It’s going to take some time to correct a century of misguided policy,” Folkman said. “The net result of an aggressive fire policy is a very difficult fuel landscape, including dense, burnable vegetation.”
Climate change: Combined with more buildings and more vegetation, there also looms the specter of climate change. Warmer average temperatures over a long period of time can create more “ignition points” — places where a fire could ignite. Climate change can also lead to more extreme weather conditions, including longer dry seasons and shorter, more intense wet seasons. These extremes encourage rapid vegetative growth that turns into vast quantities of dry tinder over the long drought season.
Combined, these factors have created a perfect storm for wildfires to ignite and spread destruction.
Mitigation matters for individuals and communities
The good news is that individuals and communities can do a lot themselves to mitigate wildfire risks. “Fire can be a very binary peril,” Folkman said. “Mitigation matters a lot. What an individual homeowner does to protect their home and make it safer can mean the difference between 100% damage and no damage at all.”
According to the Insurance Institute for Business and Home Safety (IBHS), there are many simple ways to help protect buildings from wildfires, including creating noncombustible “defensible space” around a structure; cleaning debris from the roof and gutters; and covering vents with mesh. Costlier measures include installing noncombustible siding, a fire-rated roof, and multi-pane, tempered glass windows.
Many of these mitigation strategies are designed to minimize the impact of embers. The IBHS estimates that about 90% of damaged buildings were first ignited by embers or fires set by embers. “You’re trying to reduce the ignition probability when the building is subject to an ember hazard,” said Folkman. “If embers get on the roof or inside a vent, window or cladding, that’s when loss potential goes through the roof.” Better protection against flying embers also means fewer ignition points, which does not just benefit the homeowner, but also the wider community around them.
But mitigation does not stop with an individual homeowner. Effective wildfire mitigation requires a holistic effort, including the implementation of community and land use planning that takes into account the elevated fire risks. Regulators, legislators and community members all contribute to these efforts. And insurers can create products to protect communities and to incentivize safe behavior, such as premium discounts for home mitigation.
Insurance responses: Modeling the risk
The key for insurers to effectively address wildfire risk is improved catastrophe modeling. Prior to the 2017 wildfire season, insurers essentially treated wildfires as, in Folkman’s words, “attritional sources of loss.” No longer. Wildfires are now a legitimate catastrophe concern.
Much of the data already exists, but is only now being leveraged by insurers. Governments and firefighting officials have been collecting useful wildfire data for decades.
John Rollins, FCAS, a consulting actuary at Milliman, compared the situation out West to Florida after Hurricane Andrew in 1993. “Everyone is basically in a state of shock about what could happen,” he noted. But it was the shock from Andrew that launched the development of catastrophe modeling for hurricanes. And Folkman and Rollins both agreed that the recent catastrophic wildfires have spurred many insurers to similar action. The newest frontier for catastrophe modeling is now wildfire risk.
Complicated risk, limited data
A wildfire is a complicated risk. There are several hazards, including the flames themselves, flying embers, smoke and urban conflagrations. Folkman also noted that wildfires are very similar to floods. “Wildfires are a high-gradient peril; you can’t model them on a ZIP-code level.” How a fire burns and spreads is also highly dependent on the natural and built environment, including building construction and mitigation features.
Plus, of course, there is the data issue. “Wildfires are not events that happen frequently in history. So traditional actuarial-based loss approaches are not going to shed sufficient insight into this catastrophe peril,” said Folkman. Rather, a probabilistic model based on simulations is required. In Folkman’s work, for example, he used a 50,000-year climate simulation to yield insights into ignitions and fire spreads that consider complicated wildfire hazards and environmental features. “In the past, analytical models approximated the risks, but now, by simulating years of behavior we can actually physically simulate the risks,” Folkman said.
Folkman is confident that models will continue to improve as more data is collected. “After tragedies, insurers get more real about data collection practices,” he said. He pointed to earthquakes in the 1980s and hurricanes in the 1990s as examples of catastrophes that encouraged insurers to change how they collect data and manage risks. “That’s where we are with wildfires right now. Data collection will change.” Indeed, much of the data already exists, but is only now being leveraged by insurers. Governments and firefighting officials have been collecting useful wildfire data for decades. Folkman argued that since wildfires were not considered catastrophe risks, insurers did not need to harvest such granular data. But now, that’s changing. “Think about hail claims and damages: There is so much data, so much loss data, and sophisticated underwriting and pricing. We’re going to get there for wildfires.”
Technological improvements are a big part of this change. Folkman pointed to deploying artificial intelligence to analyze satellite imagery to understand individual building risks in wildfire-prone areas. “I would expect that this will be part and parcel of underwriting data a decade from now,” he said. “Analytically advanced carriers are already adopting it.”
National problem, state regulations
Wildfire is not just a California problem; it is a national problem. There have been major wildfire events in places such as Tennessee, South Carolina and Florida in recent years. But insurance is regulated on the state level. How different states regulate insurance — and particularly ratemaking — can impact how insurers can leverage catastrophe modeling for wildfires.
For example, ratemaking regulations in California, the epicenter of many of the worst recent fires, may complicate how insurers can react to wildfire risks. Cody Webb, FCAS, a principal with Milliman based in California, pointed out that the California code requires that catastrophe losses be based on a long-term average of catastrophe claims, and that fire losses for homeowners insurance be based on a minimum 20-year average.
Under these requirements, argued Webb, carriers that suffered wildfire losses could adjust rates accordingly, but those carriers that did not suffer loss could not — even though the risk profile may warrant a rate adjustment. “Unprecedented things can happen,” said Webb. “And if we only use historical averages, then we fail to contemplate the potential for things outside our experience to happen. If risks change, those changes won’t be captured by allowable ratemaking procedures.”
In other wildfire-prone states, this may not be as big an issue. “Most western states are not those with very strict rate regulations,” Webb added. “Insurers are more free to price how they see fit in other states besides California.”
Either way, the increase in wildfires presents an opportunity for insurers to offer new products to help protect vulnerable communities. “There could be opportunity here,” said Webb, “but big risks.” The hope is that as catastrophe models become more robust, insurers will be in a better position to underwrite, price and manage these big risks.
And data is improving all the time. “Use that data. Use analytics,” Folkman stressed. “I think good things will come out of modeling that will benefit private homeowners and communities. That is ultimately the benefit the insurance industry provides.”
Change is Coming — and Insurers Are Ready
The challenges facing the insurance industry are real and the changes they will usher in are sweeping. Speaking at a CAS Spring Meeting general session, Sean Kevelighan, CEO of the Insurance Information Institute (I.I.I.), listed several of the challenges.
For one, natural catastrophes continue to break records. Three of the 10 largest insured property loss events in U.S. history were 2017 hurricanes. The 2018 California Camp Fire is also expected to make the list after the full insured losses are assessed.
Mother Nature alone is not responsible for all major catastrophes. Cyberattacks continue to dominate headlines, and their potential for catastrophic impacts is only increasing. The World Economic Forum noted a growing trend of attacks targeting critical infrastructure, including power grids. A successful attack could cripple the national economy and disrupt global supply chains.
Growing political uncertainty is also contributing to economic uncertainty. Rising nationalism and protectionist tendencies around the globe are shaking confidence in future growth. “Growth is expected to slow,” Kevelighan said. “With the geopolitical tensions we’re seeing, companies are not comfortable investing if they don’t know where trade wars or tariffs are going.”
The insurance industry itself continues to face fundamental transformations. Commercial auto losses continue to rise, up 30% between 2007 and 2017. Personal auto is also facing an uncertain future, as loss costs keep creeping up with increases in repair costs.
Another concern is the insurance talent gap. “One of the biggest challenges we face is talent,” Kevelighan said. “Organizations will need to change to attract new types of talent. That will change the culture of our organizations. We need to be ready to embrace positive change.”
But Kevelighan believes there is abundant cause for optimism. He argued that the insurance industry is uniquely poised to be a leader in helping society overcome its most daunting hurdles. “All the disruptive forces in the world can be overwhelming,” he said. “But we as an industry have done a good job leading throughout history.” He pointed out that insurance has supported innovation since the dawn of the industrial age. “You can’t innovate or modernize unless you’re able to transfer risks.”
Insurance will not automatically keep pace with change. Kevelighan suggested that insurers need to take a more proactive approach. “We used to stand behind and support innovation. Now we have to innovate ourselves and come to terms with what we need to do as an industry.” He cited cyber insurance as a prime example of the industry quickly innovating to respond to a societal need. Total direct written U.S. cyber premiums reached $2 billion in 2018 and are expected to keep growing.
We as an industry need to talk about resilience more. We need people to understand that we are the ones who can provide solutions to a world of increasing natural disasters.
The insurance industry can also make a large impact in protecting against natural disasters, especially flood and wildfires. “We are financial first responders. We rebuild communities through our investments. That’s what we do. We mitigate risk and protect capital,” Kevelighan said.
One way to do that is to encourage resilience. “We as an industry need to talk about resilience more. We need people to understand that we are the ones who can provide solutions to a world of increasing natural disasters,” said Kevelighan. Insurers, he argued, can be at the forefront of encouraging more resilient behavior among individuals and communities. “We have an opportunity to help people learn and change their behavior to mitigate their risks. Otherwise, we’re at risk of repeating our mistakes over and over again.” Indeed, risk management services are increasingly becoming a critical value factor for customers and a way for insurers to gain a competitive advantage.
Another way that insurers can strengthen communities against disasters is by developing new insurance products, like private flood insurance. He pointed to the consistently low take-up rate in flood insurance as an area where insurers can make an immediate impact. The I.I.I. has found that only about 12% of homeowners nationwide report having flood insurance. The percentage may be overstated, as many respondents who believe they have flood insurance may not be covered for flood events under their policies. Kevelighan expressed confidence that private flood insurance offered outside the National Flood Insurance Program will be crucial to closing that gap. “Private flood is one of the fastest growing insurance lines in the U.S. right now,” he said. “It’s growing even faster than cyber.” This is good news for vulnerable communities and for the insurance industry itself.
Indeed, even with the past few years of catastrophe losses, the industry remains stable and in strong financial health. Also speaking during the general session, Benoit Carrier, FCAS and managing director at Aon, noted that average statutory surplus remains high despite recent catastrophes and deteriorating ratios.
In fact, average industry surplus continues to increase. Carrier gave several reasons for why that might be the case.
Access to alternative capital, such as catastrophe bonds, has allowed for investor diversification. Regulators play a role as well by requiring insurers to maintain certain amounts of capital so they can pay claims. Additionally, new technologies are helping reduce insurer costs (e.g., automation in claims handling and underwriting). Insurtech, which was once a fearful specter that many worried would radically upend traditional insurance, has mostly supported insurers so they can do their jobs better. As Kevelighan noted, “According to McKinsey, 60% of insurtechs are actually looking to help our value chain. Only 9% are truly looking to upend our industry.”
Advancements in predictive modeling are also helping insurers to identify better risks and price them accordingly. Catastrophe models have helped the industry prepare for large losses. “It would take a very large catastrophe to turn the market down,” Carrier said. “It’s not like in 1992 when Hurricane Andrew struck. Back then there was basically no modeling. We are in a different world now where we can understand these large catastrophes.”
Both Kevelighan and Carrier stressed that, though there may be storms on the horizon, the insurance industry is well-positioned to weather them and to help guide communities and economies through the upheaval.
“The future is bright for our industry,” Carrier concluded.
The Future Belongs to the Actuaries, If They Want It
The increasing use of technology and the rise of data science have left many actuaries asking: What will actuarial jobs look like in just a few short years? Is the profession experiencing a revolution of new capabilities — or is it declining into irrelevancy?
Panelists at a CAS Spring Meeting session agreed that actuaries could become increasingly more valuable to the insurance industry as they begin to leverage new technologies and processes. The actuaries of the future can be highly efficient, technology-enabled insurance leaders.
But the panel also warned that this outcome is not a given: Actuaries themselves have an opportunity to lead change, defining their future roles by taking initiative when it comes to adopting some of the technology tools and data science techniques that many insurers are starting to explore. Actuaries should become more vocal advocates for their profession, expounding the valuable skills and industry knowledge that they bring to the organizations that hire them, and working to establish themselves once they’re in those organizations. If actuaries don’t actively lead change and advocate for themselves, they could feel like the victims of change, being displaced as technology and data science re-shape insurance operations and transform traditional roles.
Automation: Here to help, not replace
There are many opportunities to add automation to support insurance business operations, including the actuarial function. Far from being a threat to actuaries, however, introducing automation can actually be a great boon. Technology can take on some of the more manual, repetitive tasks, like data retrieval and formatting said Joseph Milicia, FCAS, Americas P&C Business Process Excellence product leader for Willis Towers Watson. “It allows humans to focus [more time] on where they can add value and where they can make decisions that drive the value chain.”
Day Bishop, a director in Willis Towers Watson’s insurance consulting and technology business unit, cited a global survey on the future of work that his company conducted in 2018. That survey found that, even though the use of automation is steadily increasing and expected to nearly double over the next three years, that number is still relatively low at present, with 20% of the insurers surveyed expecting to introduce automation within their organizations. Automation is not expected to decimate the actuarial workforce. Rather, Bishop said that most companies see automation as a way to augment human performance and productivity. Bishop maintained that automation is not being used today to replace people — especially not by insurers; it’s being used to help people do their work.
Automation can take on independent, repetitive tasks — the plugging-and-chugging that takes up so much of an actuary’s time. As they say, 80% of analysis is just getting and cleaning up the data. With these tasks largely automated, that leaves more time to focus on valuable, stimulating tasks that require collaboration and ingenuity.
Indeed, the fear of robots replacing actuaries wholesale is unfounded. If anything, they could make an actuary’s life easier. Milicia noted that the tasks that are ripest for automation are the tasks that actuaries don’t want to do anyway.
Actuaries are more than data scientists. They are business leaders.
But even if robots won’t replace actuaries, will the data scientists? Not according to Milicia, in particular when taking account of the well-rounded knowledge that actuaries have and their exposure to all of the functional areas of an insurer’s operations. These attributes put actuaries in better positions for leadership, leveraging the contributions of data scientists, Milicia said. While data scientists may be in high demand right now, actuaries are still indispensable components of the insurance value chain for a very simple reason: Actuaries know how insurance works.
Besides, he argued, the functions of data scientists could themselves be automated away soon. “Technology is way closer to automating [model building] than it is to automating decision-making and how to implement decisions to extract value within an organization,” he said. Model fitting, a task that is done with limited collaboration, is likely to be automated soon. “I’d predict that AI will replace it within the next five years,” said Milicia.
Milicia also noted that there are many tools that already exist today that can do what a data scientist does. For example, automated machine-learning algorithms can fit models today without human intervention, but maybe not as well as a human can. As technology continues to improve, Milicia said, “We absolutely can get to where the modeling could be automated.” The last person standing, if that happens, will be the actuary, not the data scientist. Actuaries can interpret a model’s output, garnering insights for insurance applications and for extracting business value. Bishop said that because data scientists can work across a variety of different industries, they’re not necessarily going to have a deep understanding of how the insurance business works, which can make it difficult to tie what they’re doing in the models they build back to business impact.
Actuaries need to act to cement their roles and the values they bring. Leveraging automation will enable actuaries to spend more time on value-added tasks so that they can contribute the valuable insights that management needs to make strategic decisions. While a background in data science can help the actuary to fulfill this role, soft skills that are not unique to the actuary’s domain will ultimately ensure the continued relevance of the actuarial profession. “I would want actuaries to have a data science skill set, sure,” added Milicia. “But in the longer term, I think it’s the insurance knowledge, it’s the non-data science skills, it’s the soft skills that a computer can’t reproduce where actuaries add significant value.”
To survive, actuaries need to make their case
But Noelle Codispoti, CEO of Gamma Iota Sigma (GIS), argued that maintaining this strict divide between the data scientists as modelers and the actuaries as insurance experts and business leaders might be easier said than done. Codispoti offered her insights through a unique perspective — GIS has an annual membership of over 5,000 students, primarily actuarial, risk management and insurance majors, from a growing network of chapters at nearly 80 colleges and universities throughout North America, making GIS the insurance industry’s premier collegiate talent pipeline. In her role, she interacts with both students and company recruiters, with insights gleaned from an annual recruiting survey of students and a recruiter’s roundtable held at the GIS annual meeting.
Codispoti pointed out that even the recruiters within many insurance companies themselves are confused by the data scientist/actuary divide. “Most people don’t know there’s a difference between data scientists and actuaries,” Codispoti said. “Most don’t take the time to understand what those skill sets are.”
Per Codispoti, whether data scientists begin to displace actuaries in many insurance functions could very well depend on this perception. “Job openings for data scientists continue to grow in insurance. The industry is thinking that’s the direction it’s moving in. The role of data scientists is growing, but the demand for actuaries remains the same.”
The very soft skills that actuaries possess, such as critical thinking and complex problem solving, do make actuaries crucial components of the insurance value chain, but these skills are not theirs exclusively. “Those skills are for everybody,” Codispoti said. “Data scientists with those soft skills can move laterally and up — and so can anybody else.”
Codispoti argued that, even with their insurance knowledge, if actuaries do not work to establish themselves within their organizations, they could grow increasingly irrelevant — or at least perceived as such.
“I would encourage the profession to stand up for its skill set and for the value that actuaries bring,” said Codispoti. “Because right now, we’re not seeing the discussion moving in that direction. That’s why actuaries need to step up. Otherwise, we will talk ourselves out of the discussion.”
Lucian McMahon, CPCU, ARM-E, AU-M, is a senior research specialist at the Insurance Information Institute in New York City.