Look, But Don't Touch

Automated rules for underwriting have made binding policies a seamless practice, particularly for personal auto and small commercial lines. Consistent guidelines make agents happy and free up underwriters for tougher challenges.

As long as there has been underwriting, there have been rules, and the automation of those rules goes back to the days of the mainframe. But as underwriting profit has become a mission, insurers are looking for more ways to make easy decisions. Underwriting rules are simple logic that can be implemented in any computer language, according to Jeff Chookaszian, associate principal for McKinsey & Co. “What the rules engines do essentially is separate the logic of the decision process from the data and allow you to maintain and build those rule sets in a more flexible way with more analytical power behind them,” he says. A lot of the newer policy administration systems are building that kind of rules capability as part of the system, he adds. Some policy administration systems give the user the option to integrate a rules engine or to use the engine the vendor built in.

Adjusting the Rules

Putting in rules is a fairly extensive process, and it's not just a one-time decision, notes Susan Stokes-Gibson, senior vice president, underwriting, for Anchor General Insurance in San Diego. She explains the carrier constantly is examining its automated underwriting rules. “There are regulatory concerns that need to be addressed,” she says. “You have rules that are filed for each state and must be maintained for the departments of insurance. You also have business rules you are going to apply. By having the users manage how those rules are [used], the pressure is taken off the interface with the IS department and takes us off its plate so we're not constantly asking IS [to make] changes. We can change on the fly.”

Occidental Fire & Casualty is a commercial lines insurer, and Paul Webb, vice president of IT, says the carrier is working to move underwriting decisions out of the back office and into the field–then the managing general agents can put the business rules in place. “Our underwriting guidelines call for us to gather additional information on companies that have been in business for less than three years,” says Webb. “So MGAs, who are out there working on a piece of business for us, can introduce a whole set of screens and logic into the process to force [the MGAs] to gather all that information. We do some in-line processing to make a review of that company.”

Rules constantly are being reviewed, Webb indicates. Some of Occidental's reviews are being done in conjunction with a data warehouse the carrier implemented last year. “We can spot trends that will help us modify our policies or at least better understand what we are putting on the books hopefully to mitigate what we're finding in the data warehouse,” he says.

The initial setup of the rules is fairly time-consuming, Chookaszian points out, because rules must conform to a carrier's guidelines, and knowledge and decisions must be captured that underwriters used to handle manually. “Once they are in place, there is a continual tuning of the rules to try to minimize the number of referrals and make sure the manual referral work you do is value added,” he says.

A first step for carriers is to examine the rules relative to what their underwriting guidelines are to see whether the rules properly reflect those guidelines, Chookaszian continues. “You look at changes in the flow of business to see whether the flow you are getting is fundamentally different somehow in terms of the risks you have taken,” he says. “You also look at the impact of manual underwriting activities to determine whether the work you are doing manually is resulting in different input.”

For example, he explains, if carriers were looking to reduce the number of referrals that came through, they would look at the number of referrals by reason, code, or classification and try to understand how many of them actually resulted in a change to the underwriting decision or a premium-bearing change of some sort.

How to Use the Rules

Mark Gorman, strategic research adviser for TowerGroup, sees different levels of business-rule utilization. “As we looked at this [area], we saw more robust and less robust rules environments,” he says, adding the first level of rules is used for editing. “These are rules that take a look at the data and determine whether all the data that is required is in place,” he says. “Do I have certain data that doesn't appear to be correct? And within that data, are there certain elements that tell me I can't write this risk at all or that I'm sure to write this risk and I don't need to evaluate it any further? Those are basic editing risks, and we saw that as the first utilization of rules.”

The second utilization combines workflow and business rules to create a business evaluation, taking the expert rules that are gained from an organization's underwriting expertise to the next level of automation–straight-through processing, according to Gorman. A third level involves the rules used in conjunction with predictive analytics, which further enhances an organization's ability to evaluate risk on an automated basis, he adds.

Prior to automated rules engines, carriers would review nearly all new business. Today, Webb suggests commercial lines carriers are reviewing 30 percent to 40 percent of the new business. “The overall rules haven't changed, but the opportunity to execute them in a better time frame is there,” he says. “Before, we'd be on the [bad] risk for a couple of months because of various state laws. Now, we can get off those risks. The automation has improved the timing [of the policy issuance], and it's improved the opportunity to do things we couldn't do before because we didn't have the information in time to do anything about it.”

What to Buy?

Anchor General Insurance is using a product from Decision Research to implement its business rules. “One of the reasons we selected Decision Research is its rules basically start off in an Excel spreadsheet,” explains John Amat, vice president, information systems. “What I like about it is I can turn this over to [the underwriting department] and let it manage its own rules.”

When looking at rules engines, Anchor General studied two different types of products. One was a development tool to manage the rules. The other was an Excel-file-based system. “When we looked at the development tool, Susan [Stokes-Gibson] looked at it, and nothing was catching with her,” says Amat. “Our programmers were oohing and ahhing. Then we looked at the Excel-based product. The programmers, who don't know anything about Excel, crossed their eyes, and Susan's face lit up. That's when we said we have to go down this route. She's our audience we need to please. Rather than trying to train her to learn another type of development tool to manage those rules, she's very strong in Excel, and that's why it made sense.”

Stokes-Gibson agrees. “The first meeting when we were looking at the development tool they lost me after about 15 or 20 minutes,” she reports. “I didn't understand how [the product] was going to work for me. [The vendor] was saying more or less it really was the programmers who were going to have to put the rules in. The management seemed so overwhelming.”

Anchor General then looked at the Decision Research product. “In the second meeting, they were talking about the Excel-based rules,” Stokes-Gibson continues. “I understood exactly what they were saying and how the rules needed to be in there. We definitely could get our hands around how we were going to be loading these rules and what parameters we were going to be looking at. It wasn't going to be a large learning curve on our part, and that was attractive. It is such a big change, and to have one portion of that be extremely familiar to us so we weren't going to have to struggle with it, that was very attractive to us.”

What to Aim for?

Occidental's initial goal was to have 70 percent of policies come through untouched. “We're running in the low 60s,” says Webb. “Based on the fact we're still looking at the data warehouse to find where we're going wrong, we never may get to 70, but I'm happy to be in the 60s.”

Occidental's overall staff count per policy is down, Webb claims, particularly in the policy processing area. “Our MGAs and producers seem to like [the rules-based system] just fine,” he indicates. “We haven't met with any resistance. They get an answer sooner. They can do multiple quotes and iterations and negotiate with the insured within certain parameters without us having to get involved.”

Stokes-Gibson has set a goal of between 80 percent and 85 percent of business flow-through on a hands-off basis for Anchor, a personal lines auto carrier. “Looking at the product, I don't see a problem with that occurring,” she says. “Initially we'll have more rules turned on so we can see how [policies] are flowing through, and eventually as the comfort level grows for people on the management side and the user side, we'll put more [policies] through without anyone interacting at all. We'll only connect those that are exceptions.”

Consistency was one area Anchor believed it could increase service to agents. They no longer have to deal with seven or eight underwriters, each of whom with personal quirks. “[Agents] really are after the consistencies this type of system will bring to them,” says Stokes-Gibson. “Their customers are going to be treated the same, and it will flow a lot easier out to them and the customers.”

“In personal lines, all the large carriers and most of the regional carriers have a huge portion of their applications–new business and even renewal–done automatically through the use of some sort of rules,” comments Chookaszian. In personal lines, he believes some carriers are achieving up to 98 percent in automatic underwriting, with the rate dropping when getting into small commercial lines, where the leading carriers probably are somewhere around 70 percent to 80 percent.

“We saw distinct jumps in straight-through processing and the ability to automate a risk as it is evaluated and passed through the system,” says Gorman. “What we are seeing in the marketplace is just the editing rules would allow for STP in about 20 percent of [personal lines] risks,” he says. “We clearly saw the application of business rules, in addition to the editing rules, allowed an organization to get into the 50 percent to 60 percent range. Robust business rules in conjunction with the rules required for pre-analysis allowed a company to jump into the 80 percent range, give or take five percent.”

In his recent underwriting efficiency and effectiveness study, Gorman saw movement within the industry to push rules as far forward as possible. “The big debate we saw was whether those rules get extended onto the agents' desktop. Do the rules get extended to a portal, do they get extended to a Web service, how far forward are organizations pushing those rules?” he says. “Some [carriers] are moving much more than others. The complexity of rules has both a scope and a scale element to it. There is a scale element in terms of how many rule exceptions and how many distribution channels you are dealing with. Different distribution channels may have different processing rules. Different states certainly have different processing rules. So the complexity of the environment has both a scope and a scale element to it.”

Dealing With Change

Handling the changes can be difficult for underwriters, contends Stokes-Gibson. “Underwriters are funny individuals,” she says. “They like to apply that one time they got caught when they wrote a risk and there was a huge loss on a particular piece of business. Every time a similar situation flows across their desk, they examine it extra hard–even if it totally fits the business rules.” Some 56 percent of Anchor General's policies are perfectly clean and don't need an underwriter's touch. “But [underwriters] have a stack of files on their desk, and sometimes you get the feeling they are quantifying their existence,” she adds. “They have to generate paper to make it look like they did something, whereas the risk was perfectly fine.”

Special Reports /

Look, But Don't Touch

Related Stories

Recommended For You