OneBeacon loss-ratio improvement aided by SAS^® Analytics

OneBeacon Insurance Group improved its loss ratio by two to four points using SAS Analytics to price insurance products accurately in several of its specialty segments. Analytic know-how gives the company the confidence to continue expanding in the sometimes difficult-to-price specialty insurance market.

OneBeacon, based in Minnetonka, MN, offers a range of specialty insurance products sold through independent agents, regional and national brokers, and wholesalers. The company's specialty businesses include insurance for professionals, collector car and boat owners, third-party general liability policies, marine and specialty property policies.

Project nets 10 times ROI
The company first used SAS to help price products in its personal and commercial lines businesses, which have since been sold. Todd Lehmann, Assistant Vice President for Corporate Research, says SAS helped him effectively price those lines and keep underwriting costs in line.

"We've been able to utilize SAS to improve pricing guidance and operational expenses and quantify the benefits," he says.

For example, the company used to order inspections for every home it insured. The inspections cost $35 to $100 each, depending on the scale and depth of the inspection. By analyzing information agents collected at the point of sale – such as the house’s age, claims data and occupant characteristics – Lehmann was able to target properties most likely to need repairs, thus reducing the number of home  inspections.

"This project netted 10 times the return on investment. It was an important early win for predictive modeling at OneBeacon. Projects like this, using SAS, typically return four to 10 times the return on investment,'' Lehmann says.

SAS^® Enterprise Miner™ speeds the process
When OneBeacon added SAS Enterprise Miner to its analytics arsenal, it immediately saw both an increase in speed and the opportunity to use multiple analytical methods simultaneously.

"It's given us richer results,'' Lehmann explains. "We can build a generalized linear model (GLM), look at it next to a decision tree or next to a neural network, and see where the best results are coming from. SAS Enterprise Miner is particularly good at comparing different types of models.''

It also shaves weeks off the time it took to build models. "We are actively working to shift model building and predictive modeling into our day-to-day work,'' says Lehmann.

The company had other choices to enhance its analytical capabilities but felt that SAS provided greater flexibility, functionality and ease of use.

Pricing specialty insurance
Model-building speed and research-staff enthusiasm are critical as OneBeacon moves into specialty lines.

Compared to single-family homes, items like yachts, collector cars and cargo ships are rare. So much less data exists on claims. And OneBeacon's subsidiaries and units capture data in unique ways using different formats.

The specialty lines also need different types of third-party data – whether it’s commercial credit scores, ZIP code information or data on where a doctor or other professional attended graduate school.  

"We get a sense from the business experts what might be useful. To supplement our data, we've used US Census data and credit agency reports from companies like Dun & Bradstreet and Experian,'' Lehmann says.

Predictive models are relatively new tools used for underwriting specialty insurance – a practice that tends to require specific expertise and knowledge about the risks as well as the constantly evolving insurance marketplace.

"Many business users have embraced the models," he says. "I hear from model users that they're almost making a leap of faith to buy in to the results. They don't fully understand the statistical methods involved. SAS can help us more quickly answer questions about a model and build confidence in that model. We don't require internal users to adopt the models built by our research team, but so far we've gone forward 100 percent of the time.''

Next up: more effort to analyze claims
When is it worth investigating a claim? Who should investigate it? These kinds of questions are difficult for insurers to answer, yet choosing which claims to fast-track and which ones to analyze further can have a big impact on the bottom line. Lehmann's team is testing models on claims processing with encouraging results.

The results illustrated in this article are specific to the particular situations, business models, data input, and computing environments described herein. Each SAS customer’s experience is unique based on business and technical variables and all statements must be considered non-typical. Actual savings, results, and performance characteristics will vary depending on individual customer configurations and conditions. SAS does not guarantee or represent that every customer will achieve similar results. The only warranties for SAS products and services are those that are set forth in the express warranty statements in the written agreement for such products and services. Nothing herein should be construed as constituting an additional warranty. Customers have shared their successes with SAS as part of an agreed-upon contractual exchange or project success summarization following a successful implementation of SAS software. Brand and product names are trademarks of their respective companies.

Copyright © SAS Institute Inc. All Rights Reserved.