Commercial Insurers Slowly Warm Up to Predictive Analytics

by   |   March 18, 2013 6:27 pm   |   0 Comments

Insurance has always been a business of information, where knowledge about policyholders’ potential for risk can mean billions of dollars lost or gained. But many commercial insurers still underwrite in ways that seem surprisingly blunt, considering the amount of available data sources that could give a more nuanced view.

Take new businesses, for example, says John Lucker, global advanced analytics leader with Deloitte Consulting. Long experience has taught commercial insurers that new businesses tend to be more prone to accidents or failures. So they avoid covering businesses younger than three or four years old – but once the business gets past those first few years, it sloughs off that bothersome “new business” stamp, and enjoys policies comparable to much older, more established businesses. However, it really shouldn’t.

“There’s nothing magical about a business turning four, except that insurers are suddenly willing to talk to them,” Lucker said. In reality, insurers should divide these businesses into more complex categories; businesses started more than four years ago, but less than eight or 10 years old are still riskier than older businesses, Lucker said, and their insurance should be priced accordingly. But insurers still often underprice the risk, and pay the consequences when claims come in.

Related Stories

Earnix develops predictive analytics and optimized pricing for insurers.

Read more»

Banks face cultural, technical hurdles to create analytics services for commercial customers.

Read more»

Lex Machina arms corporate leaders and patent attorneys with predictive analytics.

Read more»

More coverage of predictive analytics.

Read more»

To be sure, one half of the property and casualty industry has long been underwriting with sophisticated analytics that draw data from a wide variety of external datasets: Personal lines insurers, such as auto insurers, have been using extremely fine-grained data analysis to price their risks for a decade or better, Lucker and other experts say.

But when it comes to insuring businesses, progress has been slower.  Businesses require a web of different insurance products: workers’ comp, auto, property, and a host of specialty liability products, and as the companies get bigger, the risks get more complex. What’s more, businesses are a smaller group, yielding less data than the masses of individual consumers who buy auto or homeowners insurance each year.

Those challenges meant companies were slower to adopt predictive analytics, but that is starting to change. Commercial insurers are increasingly beginning to see the promise of adopting a new model for pricing risk – one that plumbs a far larger set of data sources, and uses software analysis to catch subtle relationships that indicate a higher risk profile. If that sharper edge catches even a small percentage of underwritten risks, analysts say, it can yield big payoffs to the insurers.

XL Group, a Bermuda-based insurer, started a company-wide initiative toward more enhanced data analysis in underwriting in early 2011, when it hired Kimberly Holmes as senior vice president of strategic analytics.

Under the initiative, XL is taking a newly data-driven look into which factors tend to trigger claims, Holmes said, checking into hundreds of thousands of external database sources to draw in new information. Some, like economic data, can be applied across any industry; other datasets are specific to certain industries. But there are a lot of them, Holmes notes – for example, the U.S. government alone has 400,000 datasets through various departments and offices. By using them, XL can more precisely price its policies, and pinpoint areas that may hold unforeseen risks.

“Ultimately, we’re trying to have an impact on profitability, and offer the best deals for our insureds,” she said. “When there are inefficiencies in pricing the underwriting, some insureds end up subsidizing other insureds. We can refine that approach – we can create a better solution for our insureds – but we can also improve the loss ratio.”

Holmes declined to give further specifics on the data sources, saying it would give away XL’s competitive advantage to reveal its methods. In commercial insurance, these types of predictive analytics are still far from industry standard, and insurers are reluctant to reveal much about their approach.

Lucker reckoned that about half of commercial-lines insurers employ this type of predictive analytics in their policy underwriting. Holmes said XL Group, with its 2011 initiative, was the first to apply these methods to medium- and large-sized commercial lines, and that while other insurers do piecemeal analytics for their underwriting, few other insurers have yet adopted an enterprise-wide effort to underwrite with these methods.

Among the rest, some are merely figuring out how go about this type of data change, Lucker said; others are still skeptical that a major change in underwriting practices is even worth the trouble.

“Compared with a lot of other industries, [insurers] are laggards in this stuff,” said Stuart Rose, global insurance marketing manager for software vendor SAS. In their defense, he added, the famously conservative industry is dotted with companies that have survived decades or centuries precisely because they are risk-averse, prone to carefully weighing options before leaping into a new innovation.

But, Rose said, it’s not surprising that personal insurers are years ahead of commercial lines in terms of predictive analytics in underwriting. Personal-lines insurers, after all, have a massive amount of data to draw from. Consider how many millions of people drive and own homes; each is a data point that insurers can use in their analysis. Commercial policies, by contrast, are a much smaller group, and tend to price risks by relying heavily on internal information that they’ve gathered themselves over years of underwriting.

But that’s starting to change, in large measure, because of the explosion of reliable data sources now available to them, while software vendors, meanwhile, are stepping up to offer the technology to analyze this new wealth of information.

Previously, commercial insurers were using maybe a dozen variables in their underwriting analysis, Rose said; insurers who use predictive analytics are now using hundreds of variables to make those decisions.

XL’s Holmes said many commercial insurers had the mindset that commercial lines’ relative smallness meant they couldn’t gain the same kinds of benefits.

But commercial companies don’t even need to be as precise in their underwriting as an auto insurer, she said, because these commercial policy limits are each so much larger. When the scale is that much greater, even avoiding a few losses or charging more per policy for a handful of poorer risks can have an impact.

“When you offer $25 million policy limits, even if you can identify the worst 5 percent and avoid one loss every two years, that’s a huge amount of savings,” she said.

Laura Schreier, a freelance writer based in Boston, can be reached at lauraschreier@gmail.com.

Home page image of Republic Fire Insurance Co. of New York from 1860 from Library of Congress via Wikipedia.

Tags: ,

Post a Comment

Your email is never published nor shared. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>