Pinnacol Improves Pricing With Valen's Predictive Analytics

To stay competitive within its statutory confines, Denver-based Pinnacol Assurance ($2 billion in total assets) needed new pricing tools. "As a quasi-governmental organization operating like a domestic mutual carrier, ... our policyholders are also our shareholders, and our pricing is heavily regulated to inhibit discrimination," explains Mark Isakson, the workers' comp carrier's associate VP. "By 2004 we'd reached the policy-pricing limits of our database reporting tool -- Oracle's [Redwood Shores, Calif.] Hyperion. To grow, we needed to evolve."

Serendipitously, at that time representatives from predictive analytics firm Valen Technologies (Denver) walked in Pinnacol's door. "Even though Valen's initial [underwriting-related] idea ultimately proved inappropriate, the vendor was attractive not only for its technology but also for offering to build a proprietary system to fit our business," Isakson recalls. "Other vendors wanted to fit our business into their boxed product or hosted system."

Over the next 18 months, cross-functional Pinnacol and Valen teams researched ways to leverage the vendor's skill set. The resulting solution, now part of Valen's UnderRight predictive analytics solution, enables carriers to optimize policy-pricing tiers. "We have six pricing tiers," relates Isakson. "Historically, nearly all smaller businesses had ended up in the two highest-rated tiers because we lacked a sufficiently granular means to assess some types of risk."

In July 2006 Isakson's team received approval for a full-scale deployment. "For the next six months we met frequently with Valen to develop our model," Isakson comments. "By January 2007 internal validation began, where we used all policyholder information for the previous two years to compare predicted outcomes against our actual results. And we ran all of our new business using both the old and new models."

Systems integration, internal training and agent education occurred from August 2007 through the end of that year. "The biggest challenge was managing change with our agents and underwriters," Isakson says. "Validating the Valen model really helped us find ways to minimize policyholder impact, such as giving specific risk-reduction suggestions to affected businesses."

Impact on IT, according to Dave Sauther, Pinnacol's decision support team leader, was minimal. "Since the model is mathematical and statistical, you can use any appropriate application to integrate your systems," he states. "In our case, we wrote code in [Redmond, Wash.-based Microsoft's] SQL to communicate with our Oracle Database. On the front end, underwriters continued to interact with the database with our customized Oracle Forms user interface."

The new system went live in January 2008. "We reduced the volume of small businesses in the two highest-rated tiers from 90 percent to 20 percent," Isakson reports. "And only 5 percent of all policyholders saw premium increases. This means tens of thousands of small businesses are getting a much better price."

Isakson notes that Pinnacol will continue to provide Valen with data in order to fine-tune the model. In addition Pinnacol is utilizing the modeling knowledge to create resources for improving risk profiles at work sites of customers and prospects. "We're just beginning to tap the tool's potential for assisting with keeping employees safe and proactively managing claims," says Isakson.

Case Study Profile

company: Pinnacol Assurance (Denver; $2 billion in total assets).

lines of business: Workers' compensation.

vendor/technology: Valen Technologies' (Denver) UnderRight predictive analytics.

challenge: Leverage predictive analytics to improve policy pricing and increase market share.

Anne Rawland Gabriel is a technology writer and marketing communications consultant based in the Minneapolis/St. Paul metro area. Among other projects, she's a regular contributor to UBM Tech's Bank Systems & Technology, Insurance & Technology and Wall Street & Technology ... View Full Bio