Extending Underwriters’ Reach With Intelligent Underwriting

One of the basic rules of information technology is that what you get out of it is only as smart as what you put into it, and the application of technology to underwriting isa case in point. Many insurers have already taken low-revenue risk out of the hands of relatively highly paid underwriters through automation with some sort of rules-based technology. However, simply automating does little to improve risk selection or underwriting practices—the core competency of any underwriting department.

For instance, if the logic driving underwriting rules engines is flawed from the start, automation only enables an insurer to get more unprofitable business faster, according to John Lucker, senior manager, Advanced Quantitative Services, Deloitte & Touche (Hartford). "Many insurers have simply automated what has been done before rather than leveraging the best underwriting practices or latest available tools," he says.

With the investment market flat, the competitive playing field has been leveled, according to Lucker. This means companies that can transform into the best underwriting organizations will grow and thrive. "Insurers that couple predictive models, underwriting excellence concepts and modern rules-engine technologies will be able to automate state-of-the-art underwriting tools and techniques, which can drive optimal risk selection, pricing and performance management-putting a focus on both quantity and quality."

Art and Technology

In the current business climate, the art of underwriting is more important than ever, but technology can make up for inherent human deficiencies in the way the art is actually practiced. This is especially important at a time when fewer people are actually joining the underwriting profession. "There's an over-reliance on a set of experienced practitioners who have acquired their skills and knowledge over a lifetime working in the field," asserts Matt Smyth, partner, Braxton (formerly Deloitte Consulting. New York). While valuing such skill and knowledge is understandable, there are two drawbacks, according to Smyth. "First, companies aren't necessarily able to make that individual knowledge institutional, and second, they might not have the processes and systems set up so that person can effectively step outside his or her own experience and apply other information and learning."

A few forward-looking companies have already undertaken a "next-generation" approach to rules methodology through segmentation of risk. By defining risks with greater granularity, rules can be crafted in a way to better align a company's risk appetite with policy pricing, according to Gail McGiffen, associate partner, Accenture (New York). "Rather than taking binary-type 'if-then-else' rules, for referral to an underwriter or knock-out, the next level involves taking a collection of risk attributes and grading the risk," she says. "It is as much, if not more, about the final pricing of the risk as it is about a simple decision about whether or not to accept the risk."

The greater complexity of risk segmentation methodologies involves more sophisticated technology, which can present certain challenges. "Rules engines need to support dynamic patterning of risk attributes in an iterative fashion, so that if any piece of data changes in a way that affects appetite for the quality of a risk, it will be recognized immediately," McGiffen explains. "From a technology standpoint, that requires something different from how the rules have lived within systems traditionally."

In many cases those rules are hard-coded in policy and rating systems, according to McGiffen. "There's a need to convert those rules in a way that they can be understood from a business language standpoint and then leveraged as new rules and created in a component-based architecture," she asserts. "If you have your underwriting rules coded in with your rating and policy administration system, we believe there needs to be a separation."

The cost of programming and maintaining rules on the traditional hard-coded model resulted in P&C carrier Providence Washington avoiding that stage of technology altogether for its underwriting rules. Until recently, the insurer maintained its "thousands" of underwriting rules in two three-inch binders, according to CIO Ed Leveille. To put an end to the errors inherent in manual processes, in 2000 Providence Washington (Providence, RI, premium volume $200 million) called on Mountain View, CA-based ILOG to automate the guidelines and score risks in an intelligent fashion, using a component-based architectural approach. "We felt it would give us a lot more control and consistency in underwriting and pricing policies," Leveille says.

While the implementation of the system initially encountered a degree of resistance on the part of the underwriters, the system is revolutionizing the way they do business. An underwriting desktop workstation ties the underwriters to systems, agents and external parties and gives a status-based view of policies for which they are responsible. "It organizes the information so that they know whether a policy is awaiting the results of a motor vehicle report, or whatever needs to be done," Leveille explains.

Exceptions to the Rules

While Providence Washington's middle-market focus lends itself to a fairly high degree of underwriting automation (compared to the large commercial sector), the system does allow underwriters to override a particular field or factor to intervene in a case on an ad-hoc basis. "We have built in reporting mechanisms that show management right down to an individual underwriter level how many overrides have been done, and for what reason," Leveille recounts. "While we've really tightened up underwriting discipline, we still give them the flexibility to make an exception to the rules." At the agent level, a related process allows agents to open an instant message session with an online underwriter to query a failure to pass a rule while submitting business. "At that point they can negotiate with an underwriter to get an exception or override."

Underwriters are now able to change rules, without IT intervention, in minutes. "They go in, they change the rule, submit it through an approval process, then they publish it and it's executed," Leveille says. "Whenever they want to change a rule, for whatever purpose, it takes them a matter of minutes."

In addition to saving the underwriters frustration, the solution has saved IT "countless hours of programming," according to Leveille. "We had determined that ROI was going to be really quick, and I would say it has already paid for itself." As a demonstration of the advantages ILOG brings, Leveille adds that "to put up a brand-new line of business now takes 105 days, from beginning to end," as opposed to perhaps twice that time.

Leveille is working to utilize Web services to automate many of the traditional underwriting tasks, such as ordering motor vehicle reports, loss reports and credit scores. "I'm using ACORD XML standards and the Metaserver New Haven, CT middleware product to do the trafficking of the Web services transactions," he says. For example, "when I send an XML transaction to the DMV for driving records, the DMV will send back an XML packet to update my XML database with that information, which will go through ILOG to score whether it's a good driving record or whether I have to do something else."

Work on the Web services piece is still in the beginning stages, and Leveille expects to be finished within three to four years. Part of that is internal work, and part is the ability of other parties to truly implement Web services with the ACORD XML standards, he relates. "I'm finding that many of the vendors are not as ready as they say in terms of interactive transmission and receipt of a transaction."

Concise Platform

Seeking a channel for aggregated information to support underwriting for its directors and officers (D&O) line of business, Arch Insurance (New York, $1.26 billion in premium revenue) turned to Advisen (New York), which "provides us legal, regulatory, financial, M&A activity, and background information on directors and officers, and a host of other data points, but in one concise platform where we log in and have an underwriting screen for D&O," says Craig Landi, chief underwriting officer for executive assurance and professional liability. "It's a very efficient tool for us to cull through many different databases into a usable underwriting analysis."

Having the purpose-built Advisen product amounts to a sudden leap into the information age for Arch's D&O underwriters, according to Landi. "The amount of information databases behind it was never available to us as D&O underwriters," he says. "We never had access to all the many different databases in a single consolidated place that would pre-filter, so to speak, all the information on every account. Advisen brings us that ability through their knowledge management tool."

In the past, anything approaching the amount of information served up through the Advisen solution would have required assistants to help tap and organize the sources, and leave the underwriter running triage to make the most of it, according to Bruce Brody, Arch's senior vice president of IT and operations. "That is the lack of discipline we want to avoid. Advisen lets underwriters take the flood of submission they get and almost at the push of a button get precise, in-depth information," he says. "And on the cases they want to spend more time on, they can bring the whole wired world of information to bear."

Because the application is Web-based, implementation was a "non-event" from an infrastructure point of view, Brody says. Over the last year, Arch has opened up offices in St. Paul, Chicago, Atlanta and San Francisco, and has plans to open several more in the near future, "so the ability to deploy an application simply by plugging it in is a huge efficiency for us on the IT and infrastructure side," he adds.

Managing internal information offers a different kind of opportunity. Insurance organization are overflowing with information and those companies that can tie it together in meaningful ways, coupling it with external-source information, can create predictive models that can help create and sustain competitive advantage, Deloitte & Touche's Lucker theorizes. To foster a sense of organizational sensitivity of data, companies should institute what he calls a "data czar" position or group.

The cost of locating information and establishing its validity led Chubb Group (Warren, NJ, $30.4 billion in assets) to begin a pilot in early 2002 focused on eliminating paper exchange of documents to transforming its underwriting intranet into a means of timely and vital communications between producers and consumers of the underwriting information.

Grandma's Attic

The analogy used to refer to the way information was handled previously is "Grandma's attic," according to Rick Cantor, Chubb's knowledge management unit leader. "We had been taking information on a one-way trip to the 'attic' and sticking it up there with the idea that whenever underwriters needed something they'd go up there," he recalls. A survey conducted by the carrier yielded complaints such as, "'We can't find information. And when we find it, we don't know if it's correct, because there's no author name on it, there's no date sometimes, if there is a date it's an old date, so we don't know if it's credible,'" Cantor relates.

Concentrating on six representative line-of-business sites within commercial underwriting, the pilot was built on three principles. "First, we collaborated on site redesign with home office contributors and field underwriter consumers regarding how information should be presented; second, we consolidated and consistently apply a taxonomy for the way that information is categorized," Cantor says.

"The glue that holds it all together is the implementation of operating protocols for creating information and keeping it current," which is managed by a kind of "mini-czar," a relatively senior manager in the business area concerned, designated within the pilot as a "content manager," in Chubb's terminology.

While the pilot is expected to yield substantial improvements in underwriters' effectiveness, since they will mostly be of a "soft" nature—such as the avoidance of bad risk written—internally it had to be sold for efficiency, according to Cantor. "We came up with some value statements related to the cost of not finding information and were willing to put our word on the line to improve that by 25 percent," he recounts. Given that Chubb has more than 800 underwriters worldwide, "we were able to make some pretty strong statements about yearly savings."

---------------------------------------

Measuring Risk On a Map

Natural disasters such as floods and earthquakes have long made location a factor in the underwriting of property risk. The awareness of man-made threats aroused by the 9/11 attacks have made it even more important.

"Geographical plotting of exposures can provide greater predictive power to underwriters, but there has traditionally been a separation between business applications and dedicated geographic information systems," says Henry D. Miller, an analyst with Framingham, MA-based market research firm IDC. By combining business analytic functionality and location plotting, GIS application providers "such as MapInfo {Troy, NY and Questerra Mountain View, CA are bridging that gap," Miller adds.

Questerra's Web-based intelligent underwriting application combines the factor of location with a company's internal policy data and predictive modeling calculations, says Mark Stier, vice president, product management. "Underwriting applications are then built around business rules that can be applied in real time to the decision-making process."

Anthony O'Donnell has covered technology in the insurance industry since 2000, when he joined the editorial staff of Insurance & Technology. As an editor and reporter for I&T and the InformationWeek Financial Services of TechWeb he has written on all areas of information ... View Full Bio