While many underwriters prefer to view the decisions they make as more art than science, the underwriting of any line of insurance often can be distilled to the process of applying a set of rules to a given scenario. In fact, traditional paper-bound underwriting rule books are proof of how the underwriting judgment process is guided by a codified combination of business strategy and professional knowledge.

Any process that can be distilled to a set of if-then rules is a candidate for automation, and the realization of this led to the development and installation of expert systems at insurance carriers, which gained popularity in the early to mid-1990s. But as it turned out, the technology was not ready for prime time.
There were a lot of [expert systems] products on the market and a lot of confusion about what they could and couldnt do, says James Fridenberg, vice president of applications development at Farmers Insurance, Los Angeles, who has worked with the technology since the early 1990s. Plus, the systems werent mature enough, and there were performance issues that limited the scope of these systems. Once you got into the high-volume-transaction-activity applications, very few of the expert systems at the time could scale.

As a result, interest in expert systems declined. The thrust of insurers underwriting technology initiatives shifted to front-end systems to capture application information and workstations to provide underwriters with a portal to the various internal applications and third-party data needed to make decisions on accounts. There just hasnt been great investment in advanced underwriting decisioning technology over the past couple of years, says Kimberly Harris, research director at Gartner. The attitude seemed to be, If its not broke, dont fix it. And [other] things needed fixing a lot more, such as legacy, claims, or policy administration systems.

Problems with deploying decisioning technologies werent limited to immaturity of the technology itself. To have a system that can make a [rules-based] decision, you need to have all the pertinent data come into the system. Back in the 90s, the industry found itself looking to make automated decisions with only half the picture because it didnt have electronified data from all the different [underwriting] sources. Were just now seeing real progression to help fulfill this business requirement, Harris says.

Part of that progression has been the deployment of electronic application systems, as well as increased electronic receipt of third-party information such as motor vehicle records, Medical Information Bureau (MIB) reports, and credit information. However, some of the third-party data providers have not yet stepped up to the plate, says Harris. XML in and of itself is relatively new, and most XML implementations in the industry have been more internal than external. We havent had the data standardization across all the industries that touch the underwriting process.

Interest and Impact
However, in universal agreement among the analysts and insurers surveyed, development and purchasing activity in rules-based decision systems have seen a resurgence in recent months. There are two main underwriting activities toward which these systems have been targeted: support processes, such as automated workflow and report ordering, and the underwriting decision itself. It is this second process carriers typically think of when they look to underwriting auto-mation, hoping to achieve exception-based underwriting where only a small percentage of applications or policies receive human intervention.

The reasons for this resurgence are several. Part of it is better packaging: a shift from using the term artificial intelligence, which suggested replacement of staff, to the terms rules engines or knowledge-based systems, which suggest technology that supports the underwriter. Expert systems have been re-innovated as exception-based processing systems. Vendors are saying theyre not removing the human element but, rather, are automating the simple tasks, applying a filter, and having underwriters focus on complex cases that need more attention, says Harris.

Another reason is better technology. My first exposure to [expert systems] was 10 years ago, says George Jeffers, director of information technology at Commerce West Insurance, Pleasanton, Calif. Then, the IDE was character based; now its GUI. Then, integration was a major challenge; now the applications themselves are completely object oriented. The scalability also has improved.

Even companies that already use legacy underwriting expert systems are showing an appetite for new deployments. For example, Farmers Insurance currently has several successful automated underwriting systems based on Computer Associates AionDS that tar-get Farmers property/casualty business, which accounts for 85 percent of the 15 million policies it services. These systems include an expert system for automobile insurance deployed in 1994 and one for homeowners in 1996 that today achieve a 95 percent pass-through rate, as well as one used for automated attachment of policy forms and endorsements and distribution of notices to Farmers insureds. The insurer also deployed the Life Underwriting System (LUS), now owned by CSC, to automate report ordering and workflow of its life insurance business.

In the mid- to late 90s, we were heavily into development of expert technologies, but since then we have focused primarily on maintaining them as weve targeted other systems, Fridenberg explains. Farmers is considering an upgrade to the latest version of the Aion business rules engine and is looking for other areas to apply business rules technology. There is renewed interest here to examine all our policy processing areas to see where we can leverage business rules that are part of our code and pull them into a knowledge base, Fridenberg says. Our objective is to reduce system complexity and thus reduce the TCO [total cost of ownership]. The first likely target for that work is expanding the use of rules-based systems in automated document generation.

Underwriting technology also may be a good candidate for todays insurance IT appetite, where multiyear developments are out of favor and demand for ROI is paramount. In fact, at a time when every 10-point list of how to implement technology successfully includes the phrase intermittent deliverables, rules-based systems are a perfect fit. Insurers can establish a minimal set of rules to build user comfort with the system and then tweak these rules over time to pass more business through.

You can start with a 50 percent solution or even deploy a straw man application you let your knowledge workers tweak and tune, says Jeffers. Commerce West followed an iterative deployment method when it brought in CleverPath Aion Business Rules Expert (BRE) from Computer Associates in 2002. Installed in less than a month at, according to Computer Associates, a cost of $50,000, the system originally considered about a hundred underwriting rules for processing automobile renewal policies.

We sat down with the staff, worked through business processes with them, watched them do their jobs, and worked together to build a basic rule set based off that, Jeffers explains. Weve made changes to the rule set over time, and currently about 75 percent of renewals are going through the system automatically, amounting to about 2,000 transactions weekly.

The renewal process used to require two full-time employees. As a result of the BRE system, the process now is a part-time task for one employee. Additionally, the carrier uses BRE to automate ordering of policyholder questionnaires, sending cancellation notices to loss payees on policies, and obtaining third-party underwriting information such as motor vehicle reports. By eliminating blanket report ordering and reducing staff in the renewal process, Commerce West realized a return on investment in the first year of deployment. Additionally, one of the best things about Aion is its easy to maintain and update, Jeffers adds.

The benefits Commerce West obtained are just some of the types carriers can target. There are five main benefits we see to [rules-based underwriting] technology and the systems that support it, Harris says. There are cost savings from reducing paper and the electronification of policy information. There is improved data quality and fewer errors due to less rekeying. Third, there are shortened sales cycles. Fourth are positive impacts on customer service, speed of policy sales, and speed of producer payments. And fifth, there is quicker problem identification in the underwriting process due to auditing and workflow controls.

P&C Reward Vs. Risk
In property/casualty, the lines most appropriate for automated underwriting remain the most commoditized, including personal lines and small commercial lines business where carriers can realistically target 80 percent or higher pass-through rates. In the small market, its easier to automate the environment. In the middle markets, its difficult to guarantee youve received all the right pieces of information and they are accurate; that the information out of the broker is of high quality, or that the horizon of loss history is sufficient and accurate, says Renato Derriaik, senior associate at McKinsey & Company management consultants.

In other words, while standard, electronic reports, such as CLUE and motor vehicle reports, are sufficient in small markets, larger accounts need engineering surveys, prior carrier loss information, and other nonelectronic and nonstandard third-party data. Additionally, the variety of larger commercial lines business makes establishing underwriting rules prohibitively complex, and oversimplifying rules just to enable automation is risky.

If they arent careful, companies can more than trade off any expense savings for an increase in loss ratio, particularly because the distribution system will be very efficient at creating adverse selection, says Michael Weinstein, director of research at Conning Research & Consulting. That is, carriers that streamline the underwriting process at the expense of underwriting discipline quickly will become known by brokers or consumers as safe havens for business more judicious carriers would decline.

In fact, Conning recently released a study entitled Executing Commercial Underwriting AutomationIts Only a Tool that detailed the risks in targeting the middle and upper markets in commercial lines with decision-making technologies. What you need to do is identify and separate [types of business] that can be standardized, and where efficiency doesnt result in loss of effectiveness. How carriers then choose to deploy underwriting technologies and execute them are what will differentiate them in the marketplace, Weinstein says.

Life and Health
As with large commercial lines, individual life and health applications have characteristics that present challenges for underwriting technology. The [life] industry has been trying to automate the decisioning process for at least 15 years, but about all the best carriers have been able to do with auto approval is somewhere in the 20 to 25 percent range, reports William Pratt, principal consultant and director of underwriting services at CSC. Consider just the fact medical dictionaries have thousands of conditions that need to be evaluated in conjunction with other factors such as avocation or lifestyle.
While you could, over a long period of time, build a rule base that could evaluate a greater percentage of these cases, its currently still less expensive to let a human being do it. There are some things not worth automating.

What carriers therefore are measuring as success in automating individual life and health policies, Pratt adds, is reducing the number of touches they have to make to a case, ideally, having an underwriter review a file only once. To do this, rules-based systems are used to automate ordering of MIB reports, driver records, paramed exams, and other information, as Farmers Insurance has done with its LUS system. These systems then can present a single, consolidated application to a human underwriter.

Because group policies are underwritten differently than individual policies, they are better candidates for rules-based decision technology in life and health. For example, American Fidelity Assurance Company, the largest life and health insurance company headquartered in Oklahoma and primarily a writer of group disability and individual supplemental medical insurance, is well on the way to an 80 percent pass-through target for applications processed by its in-house-developed jet-issue system, according to Marriet Etheridge, vice president of customer service.

The jet-issue project had its roots in a 1995 in-house development of VISIONS, a laptop-based field sales support tool by which applications could be submitted electronically, and the home-office component of VISIONS would populate policy data in American Fidelitys policy administration system. However, the carrier sought gradually to reduce or eliminate manual processes in the workflow at the home office and to effect straight-through processing that included rules-based underwriting.

In early 2002, the carrier deployed automated rating and underwriting of supplemental medical policies and, in late 2002, added group disability, life, and supplemental hospital expense policies to the automated underwriting system. All systems development was done by in-house programmers using Software AGs mainframe-based Natural/ ADABAS operating on an IBM OS/390. American Fidelity established user-driven tables to control the jet-issue process. We already had built the foundation with our automated laptop system and home-office administration system and totally had revamped our infrastructure in the 1990s to allow for more distributed processing. There was no need to bring in other applications or development tools, Etheridge explains.

Batch programs, which run periodically throughout the day, verify the data on the applications, review the answers to the medical questions, validate the policy guidelines, and check all compliance regulations. Clean applications then are automatically issued, printed, assembled, and sent to the mailroom.

The impact of the total system, according to Etheridge, has been to reduce new-business processing and policy-issue time by 90 percent for policies that pass through the underwriting system untouched. She reports, over the next five years, we expect jet issue to provide a return of over 50 percent and ultimately save the company millions of dollars.

All employed field agents are provided laptops with the VISIONS application. American Fidelity plans to complete a Web-enabled version of VISIONS to facilitate sales automation through all distribution channels, which is planned for a late 2003 or early 2004 delivery.

Lasting Impact
Analysts predict the current rebirth of expert systems as rules-based underwriting technology will have a more lasting and positive impact than earlier generations of the technology. They also expect vendors to continue to push the envelope of the technology into more difficult-to-codify lines of business.

Connings Weinstein concludes rules-based systems will in fact become a must-have component of underwriters tool kits in many lines of business. It is going to become a de facto requirement, he says. Carriers will need to do this if they want to participate in the marketplace, he says. Were not heading toward an Orwellian world, but insurers will continue to need to control headcount and other expenses in the underwriting process.

Better Underwriting
Via Technology

According to Gartners Kimberly Harris, there are four key areas where insurers are improving underwriting via technology:

1. Electronification of third-party data. In the traditional underwriting world, the data underwriting needed to collect didnt come in a format that was easily stored, saved, and analyzed, she says. You had faxes, proprietary data formats, and hard-copy applications, and it took a lot of middle work to transfer, rekey, or deal with hard copy. Getting data electronically is a necessary foundation for automated reviewing.

2. Rules-based workflow. Today, were hearing companies talk extensively about workflow and business process management, Harris explains. There are multiple systems involved in the underwriting processinformation from producers, third parties, applications for fraud and risk managementsimply a lot of different pieces in the pipeline. Carriers need to focus on managing the flow of the electronic work space.

3. Rules-based decisioning. At this point, companies can start to work on exception-based underwriting based on the underwriting criteria that are codified in business rules engines.

4. Case management. Carriers are deploying workstations that provide better visibility, management, and reporting on underwriting tasks. Theyre achieving tight integration with imaging and third-party data sources and getting more real-time information into the underwriters hands when application of business rules results in the requirement of a manual case review.

The Path of Least
Resistance

Business buy-in is an often-stated requirement of a successful IT project. Because of the need for involvement of users in establishing, monitoring, and modifying rules, underwriting technology is one project where resistance of users can cause a systems premature failure.

A few years ago at one carrier we surveyed, we asked how much new business was being underwritten without human intervention as a result of its expert underwriting system, says John Johnsen, managing director of TCi Consulting & Research. When the users answered two percent, everybody looked up in surprise. Why would you ever put in an automated system with a two percent hit rate?

The reason for that particular low pass-through rate was due to what politely could be termed artificial complexity in the carriers underwriting process. You could have the best system in the world, but if your chief users are standing in the schoolhouse doors, its just not going to work, Johnsen adds.
Outside of firing the entire underwriting staff, a solution to this cultural issue may be hard to come by. Carriers that have implemented successful underwriting technology have found involving staff throughout the definition and development process can be effective.

We identified underwriters who were technically savvy and brought them into IT to teach them how to develop in Aion, says Commerce Wests George Jeffers. Its a simple, SQL database language, and its easier to take people with business expertise and teach them technical knowledge than to take programmers and try to give them business knowledge.

Also, as the Aion installation at Commerce West shows (see main article), first automating tasks that users are more than happy to be rid of can lay the groundwork for successful expansion into other areas. Says Jeffers: Underwriters immediately liked the [automated renewals] system because it freed them from doing boring knowledge-worker tasks and allowed them to do more fulfilling work.

Want to continue reading?
Become a Free PropertyCasualty360 Digital Reader

Your access to unlimited PropertyCasualty360 content isn’t changing.
Once you are an ALM digital member, you’ll receive:

  • Breaking insurance news and analysis, on-site and via our newsletters and custom alerts
  • Weekly Insurance Speak podcast featuring exclusive interviews with industry leaders
  • Educational webcasts, white papers, and ebooks from industry thought leaders
  • Critical converage of the employee benefits and financial advisory markets on our other ALM sites, BenefitsPRO and ThinkAdvisor
NOT FOR REPRINT

© 2024 ALM Global, LLC, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to [email protected]. For more information visit Asset & Logo Licensing.