Standing Guard

The old adage claims there is an exception to every rule, but in measuring an insurance carrier's appetite for risk, too many exceptions slow the process of policy issuance and too few exceptions create risks the carrier might not be interested in carrying.

Such is the challenge for rules-based underwriting and exception-based underwriting, which Deloitte principal John Lucker describes as "intertwined." Rules-based underwriting uses a mainstream process flow and is more narrowly designed to include the basic rules a company has established in its underwriting department to induce straight-through processing. Information enters the underwriting process, and the rules-based system reviews the information and processes a logic flow.

Exception-based underwriting, explains Lucker, takes place when data proceeds through the normal flow and the system detects something different about the particular policy application. The system automatically channels that risk to either a human underwriter or a subordinate process to deal with it.

"What's particularly important about this process is, in the past, most of these types of logic flow were binary," Lucker says. "Things were done manually and somewhat subjectively."

The benefits of today's process flow are objectivity and consistency in how processing is done–traits that previously did not exist. "There are a variety of expense-based reasons why that is beneficial and a variety of underwriting benefits that can be achieved," says Lucker. The most specific underwriting benefit comes from eliminating much of the subjectivity of the underwriting; the randomness of some decision-making no longer takes place. "By randomness I mean if you don't have these exception rules in place, a number of people can look at something, and based on their own learning or training, they can come to somewhat different decisions," says Lucker. "Any one of those decisions can have shades of rightness or wrongness to it. One could be completely right or completely wrong. What you end up with is a statistical distribution of behavior, which isn't necessarily optimal."

Exception-based underwriting is an evolving area for insurers working in commercial lines, indicates Steve Sumner, a director with PricewaterhouseCoopers. In the commercial lines space, there is a combination of some sophisticated rules on the front end and the exceptions to those rules so the risk receives an appropriate level of underwriting attention. "Sometimes your underwriting rules can be so restrictive nothing actually passes through and is accepted into the binding and policy issuance process," says Sumner. "Other times, you run the risk of too much going through without that level of oversight."

By using quantitative and analytical tools to analyze how to construct these exceptions, carriers end up with an optimal process flow that is more statistically correct in how it proceeds because it is based on a consistent process, adds Lucker.

"If you don't create these rules, which the industry has worked on over a large number of years, you move too close to subjectivity and inconsistency because you don't have facts on which to base your decisions, and that tends to be suboptimal," says Lucker. "If you create a total no-touch environment around underwriting–and some insurers don't believe that is the best way to do it–you end up with potentially a highly rigorous and statistically accurate process that may not make the most business sense. Exception-based underwriting allows you to work away from that dark line of highly no-touch, highly objective, and highly consistent to a more livable process that tends to make more business sense and is more toward correct than having no rigor or analysis at all."

AUTOMATED PROCESS

ICW Group has been running an exception-based underwriting system in its workers' compensation book of business for the last six months, but as company president and CEO Kevin Prior explains, it is a culmination of an idea the carrier has been working on for the last three years. "Our idea was to take many of the manual processes out of the underwriting decisions so we could have the underwriters concentrating on agency relationship management," he says. "From there, it grew into the development and automation of underwriting criteria we felt were more germane to each individual risk."

Today, the underwriting criteria ICW has in place serve as a training tool for new employees and give the carrier consistency across the organization, according to Prior. "When we enter tough times, we at least have hard numbers and consistency around decision-making so the decisions are not being made blind," he adds.

ICW co-developed the solution with FirstBest Systems, notes Prior. The vendor has a history of being successful with its entrepreneurial ventures, so for ICW it was a good situation. "The product was designed exactly to our specifications without our having to pay the full price of customization," says Prior.

Paul Zamora, vice president, workers' compensation, for ICW Group, believes FirstBest's vision of what this product would look like, its functionality, and the level of automation matched the vision of ICW. "We were confident it had the skill set to develop the final product to our satisfaction," he says.

CREATING THE RULES

Using edits and rules to determine underwriting appetite at Penn National is nothing new, according to Helena Vendrzyk Gordon, director of projects and planning for the carrier. But prior to the installation of a rules-based system from Pegasystems, most policies ended up being touched by an underwriter.

In 2006, the carrier looked at improving its ability to use exception-based underwriting, including predictive analytics. After settling on Oracle's Skywire Software Insbridge rating engine, implementation kicked off in the first quarter of 2007.

"We designed and built an agency portal, we implemented the Pega rules engine to do a lot of our editing around what business we want to underwrite, and we are using the Skywire Insbridge rating engine that integrates into our legacy back end," says Gordon.

Carriers need to make sure the business going through the pipeline fits their underwriting appetite. "If your appetite is for a lower-hazard type of business, you want to make sure that's the type of business getting through," says Sumner. There always will be a high degree of oversight and monitoring of the business, so if there is a deterioration in loss ratios, an insurer is going to zero in on where that deterioration is coming from, he continues. "The idea is to drive toward a level of profitability that is within a company's level of acceptability," he says.

Penn National worked primarily with analysts from its underwriting division to develop its rules, relates Gordon. "We went through existing logic and underwriting guidelines and translated that into rules," she says. "We then reviewed them to see which ones we want to go forward with in the future."

In ICW's experience, writing the rules for the system was very complex, remarks Zamora. "We pooled our best underwriters and started talking about how we approach underwriting a risk," he says. Ideally, the carrier would like to ask potential customers upward of 50 questions, but in order to achieve straight-through processing, such a number is impractical, he adds, advising, "you have to be smart about the questions you ask and make sure they have an impact on the final price."

After developing the questions and guidelines, ICW's business leaders and IT staff sat down with the vendor to walk through the workflow to ensure the rules provided the automated functionality and underwriting capability to preclude those risks the carrier did not wish to write from slipping through.

Zamora believes companies should analyze their data quarterly to make sure their decision tree makes sense. The system tracks exceptions, and when the exceptions take place, the carrier then can decide on the risk. "If we have consistently said this is [a risk] we want to write, we don't want to make it an exception," he says. "Based on the results, we would work with IT to make corrections to the system."

UNDERWRITERS

The likely outcome of the changes at ICW is the carrier uses its underwriters on policies that have a higher value. "We've expressed to the staff members we are going to elevate their value to the organization by eliminating a lot of the remedial workload, and this elimination allows us to cast a broader net," says Prior.

Zamora agrees exception-based underwriting changes the role of underwriters. "They become agency relationship managers," he says. "We need underwriters to help service the business, especially larger policies. There always is going to be a need for good underwriters, but what they do is going to be far different. They won't be analyzing every aspect of every account."

ICW expects about 80 percent of its workers' comp policies will be handled through straight-through processing. "It's a large number of policies, but probably only 30 to 40 percent of our total premium," says Prior. "We wouldn't automate a $5 million to $10 million account, but we're definitely looking to automate accounts under $100,000."

The challenge for the industry, Sumner contends, isn't so much the quality of the underwriting but the fact underwriters are going to take on a different level of responsibility. "There will be a transformation from historical manual processes to a more automated environment," he says. "Technology has to be viewed as an enabler of the underwriting process. By default the industry will have to lean this way and increase a reliance on technology."

That means the underwriting responsibility will be different, particularly in the small- to middle-market commercial arena. "If you are writing a very complex risk, you may decide that's going to have to go to the underwriters," says Sumner. "Down the road, we are going to see a different approach to the process, particularly in the small to middle market."

SOFT MARKET

The market cycle also plays a factor in underwriting decisions. "In a hard market, companies can be more firm about their rules and making further exceptions," says Lucker. "In a soft market, insurers have to be more flexible and make more exceptions, but they want to work toward being right rather than not."

Depending on how those rules drive their growth or their shrinkage in the soft market, carriers need to be adaptable and change those exceptions and rules to make sure they are achieving their business objectives in a timely fashion. "The exception-based systems and processes, rules engines, and predictive models need to be flexible and rapidly changeable," says Lucker. "That's driven a whole new movement toward enterprise rules engines and flexible rules that allow these systems to be changed more rapidly and without IT involvement in some cases."

Today's insurance market has softened, so insurers, particularly in commercial lines, have been active in expanding class appetites in terms of the types of risk they are willing to consider, points out Lucker. "What they've been doing with that is not only from the perspective of leveraging predictive models but leveraging the need for new and additional data that supports the application for those expanded classes," he says.

For example, if an insurer has never insured oil delivery companies, the carrier needs to gather different information than it has ever gathered before. "You need a flexible solution to do that," he says. "Every time you expand into a new class, you can't have a massive systems undertaking to adapt. Your underlying rules-gathering and processing systems need to be flexible to allow you to expand and contract your appetite accordingly."

SPEED VS. ACCURACY

More underwriting attention needs to be applied for complex risks, according to Lucker. In particular, he cites dealing with agents a carrier has less experience with, risks where the data has been gathered in an incomplete fashion, and determining how to use more experienced vs. less experienced underwriters.

"There are a variety of different rules you can devise to carve up your book of business into risks or profiles of risk that have a higher propensity for needing attention," he says.

The time savings realized from entering the policies into Penn National's policy system have been huge, affirms Dean Kimball, project manager for the carrier. "When a piece of new business came to us prior to this project going live, we had a seven-day expected turnaround to have that business entered into our policy admin system," he says. "Now, when the entry is made and submitted into the new agency portal, it will be uploaded into our policy admin system that night and run through our policy admin system's edits. As long as it passes all edits, it will be accepted in the policy admin system that night. We went from potentially up to seven days down to the same day."

Sumner contends when independent agents get bogged down with a carrier's underwriting system, the agents turn to other carriers. "Agents will use the path of least resistance, so you have to find something that is simple enough and user-friendly so agents actually will use it," he says. "When you think of these people having to deal with a half dozen or more underwriting systems, and if they don't understand the rules or it is too complicated for them, they might have a tendency to go another direction to a system that is more user-friendly to them. This is where agency management comes into play. You have to know your good distributors from your bad distributors and what your control environment needs to be."

EXCEPTIONS

The typical process with Penn National is for exceptions to be kicked off to the underwriters if, for example, there is a class of vehicles or a type of driver characteristic the carrier would not want to underwrite. "Once underwriters see [the exception], they can work with the agent to gather more information and make a decision as to whether the customer wants to accept the policy with changes–perhaps changing the limits of the deductibles–or they reject the policy," says Gordon. "We think the rules are correct. It's a matter of the business appetite of what we want to write."

Such rules will improve underwriting in the commercial lines with the key being access to the data, states Sumner. "As technology improves and the data improves, you can begin to draw correlations with some of the risk attributes, and you will see improvement in underwriting, much as we've seen in the personal lines space," he predicts. "Over the last 15 to 20 years, it is amazing how far the personal lines companies have come. I think we're going to see that in the commercial lines space, as well."