It’s generally accepted that the first automobile policy in the U.S. was sold in 1897 in Dayton, Ohio. For more than 100 years, automobile insurance would be underwritten fundamentally the same way as that first policy: manually and customer-by-customer. However, the effort that began more than 10 years ago to apply business intelligence to auto-insurance underwriting has transformed what had once been an art into a standardized and automated process at many carriers, where manual processing is the exception rather than the rule.
“Personal auto is clearly the most advanced and most mature in its use of business intelligence and analytics. It really is a science,” says Deb Smallwood, founder, Strategy Meets Action (SMA). She believes that homeowners is the next line poised for a process evolution, with the only limiting factor being gaps in available data.
The adoption of advanced analytics in commercial lines has been slower. “There’s a continuum [in automation] from how much margin there is in the business and how complex it is,” says Matthew Josefowicz, managing director at Novarica.
“Low-value business owners or workers’ compensation lines are going to start to look like personal lines, with a greater emphasis on straight-through processing. Complex liability will take longer to automate, and there’s a question of the marginal value of automating it at all. But certainly there is value in using business intelligence to make those underwriters more efficient and giving them better tools to do their job,” he explains.
That’s not to say that carriers have not experienced success in leveraging intelligence in the commercial lines underwriting process. Chesapeake Employers’ Insurance Company, Maryland’s largest writer of workers’ compensation insurance, has been using a predictive analytics solution from Deloitte since 2008 to score risks and suggest the proper tier and pricing for an individual account. Now, the company is targeting other enhancements to the underwriting process.
“The high-level objectives are to make underwriting and the process of analyzing risk more efficient, to make underwriting more accurate, to increase the speed of processing and to create a better experience for our agents [when] working with underwriters,” says Steve Orr, Chesapeake’s senior vice president of Marketing and Technology.
Chesapeake is in the midst of a multi-year project to replace a 20-year-old, in-house-designed policy administration system. The first stage of that replacement, completed earlier this year, involved incorporating an Oracle Service Bus into the company’s IT architecture in order to create SOA capabilities that would allow new components to be added.
The first component will be the FirstBest UMS underwriting workstation. With final development completed in early November, the company is currently in the testing phase and plans a March 2014 rollout. A key capability of the FirstBest system will be integration with external data sources.
“The workflow in UMS will be utilized to automatically assemble a lot of the third-party information that underwriters have to go out and get today,” Orr says.
Examples include Dun & Bradstreet data, vehicle information and corporation-licensing information from the state. When an application comes to the underwriter, the system will present a task manager showing the information that has been gathered and the items that need review. The FirstBest system will also integrate with Chesapeake’s risk-scoring platform to provide straight-through processing for the majority of accounts that generate less than $10,000 in annual premium.
“We’re confident that the workflow FirstBest provides will give us the confidence to do straight-through processing that we haven’t had with our current platform,” Orr says. “The advantage to the agent will be faster—even immediate—turnaround, as well as the ability to run ‘what-if’ scenarios while they are on the line with the customer. Also, underwriters who today need to spend their time on the smaller applications will be able to apply their effort to more complex accounts, which will help us expand our marketing efforts.”
Global-commercial insurance and reinsurance firm XL Group is in the process of creating its “Global Underwriting Platform.” Traditionally, underwriting processes at XL have been paper-based or handled by different systems based on lines of business. Little information was able to be shared across different underwriting teams, and risk data was often confined to underwriting files, limiting its ability to be analyzed.
The new platform is designed to allow underwriters to collaborate with each other, access external data and integrate with internal administration systems. Kurt Schulenburg, the insurer’s vice president of IT Strategy, says that the ultimate objective is to transform the underwriting process.
“First, we wanted to make underwriting a much more efficient process through technological automation and improved underwriting workflows. Underwriters will be able to spend less time on completing tasks and more time on risk analysis and evaluation,” Schulenburg says. “Second, we want to improve analytics throughout the process. Working in our current environment with disparate systems and many manual handoffs, it’s difficult to perform analytics or even to present all the information in one spot to an underwriter.”
XL’s global platform includes several new components—FirstBest’s UMS to provide a common workbench for underwriters, Accenture Duck Creek to create a single administration system for the company’s North American ISO lines of business, and SAS Visual Analytics.
“That [platform] will give us one interface for all P&C underwriting—a common look and feel, a common way of capturing information and a common way to get from an application submission to policy issuance to endorsements and renewals,” Schulenburg says.
One of the most significant impacts of consolidating policy and underwriting data across classes of business will be improved visualization. “We will now have location management capabilities,” Schulenburg says. “By capturing and consolidating location data every time we underwrite a location, there will be a visualization that not only shows how close that risk is to earthquake and other hazard zones, but also to where XL already has exposure in that geography.”
The initial release of the system to support general liability was completed in 2012. By the end of 2014, XL plans to have about half of its P&C premium on the platform; by the end of 2015, that amount should exceed 90 percent.
Schulenburg says, “Our goal is a 40 percent premium-per-underwriter gain for those groups that have deployed the platform. Our current release for general liability shows that’s very achievable. Property and workers’ compensation are coming next, and we plan to hit that [percentage gain] there as well.”
More Than Automation
Turning underwriting from an art to a science isn’t just about automation; it’s about making more precise underwriting decisions.
“Once we generate greater efficiencies for everybody, the next question is how we can gain the benefits of all the new data we are capturing,” explains Schulenburg.
There has been an explosion in underwriting data. There are sources for geocoding, geographic information, aggregated consumer information, commercial data, tax information and more, as well as easier ways to collect that information electronically. But this data proliferation creates its own challenges.
“A human brain can only look at a few data points and correlate them,” Smallwood says. “The computer can do hundreds or thousands within predictive models. However, we’re just in the early stages of applying analytics to commercial markets.”
XL Group’s Strategic Analytics team is building predictive models based on SAS Visual Analytics. The intent is for those analyses to connect the dots within and across risks, helping underwriters make better informed decisions. But it’s no easy task.
“The ability to implement a predictive model is based on volume and consistency of risk, which is difficult to do in highly complex commercial lines,” Schulenburg says. “Within professional lines, we’ve identified some cases where there was enough volume and consistency to implement a predictive model, but we’re still researching how to make that work for property and casualty.”
In addition to modeling challenges, companies can also face internal resistance to underwriting transformation.
“There is a significant cultural issue,” Smallwood says. “Insurers need to be willing to embrace business intelligence to transform the art of underwriting into a blend of science and art. It has to start with the mandate from executive leadership, through each line-of-business head to the underwriting department. It is not about eliminating the underwriting, but augmenting the underwriting-risk analysis, decision and pricing with expanded insights and intelligence about the risks.”
Orr says, “Everyone is excited about greater integration, not having to re-key data and increased technological automation based on the objective parts of underwriting. That’s the easy part. But it’s harder to target the judgment parts of the process. Changing the way you make decisions, the ways you price; the risks you work on—that’s the next challenge.”
Showing underwriters the benefits and results of business intelligence and anlaytics is the key to overcoming change-management challenges. “As analytics is being introduced, it should be rolled out as a tool,” Smallwood explains. “Don’t force it on the underwriter, and allow the underwriter to override [the recommended action] with explanation. Then, run reports showing the difference in loss ratio between the accounts where underwriters overrode the system and where they did not.
“Overall the insights and intelligence are correct more often than not,” she adds. “Over time, the underwriters will begin to embrace analytics as a tool to augment their art.”