The Big Fix-Case Studies Published in ITA Pro
The Time Was Right to Invest in Technology
Predictive analytics and predictive modeling have slowly crept into the world of workers` compensation and as its value has been demonstrated, Matt Crum, president of the family-owned Frank Winston Crum Insurance Company, felt the time was right for the workers` comp and general liability company to make the investment.
“I wouldn't say we had challenges,” says Matt Crum of the company`s decision. “We identified a problem and we needed a solution for it. We always focus on ‘if you stop getting better you stop being good.` We constantly look at ways to improve ourselves.”
Having heard plenty about the value analytics can bring, Crum Insurance went shopping to check the validity of such claims. The result was a deal with Valen Analytics, which the carrier currently is using for underwriting.
“The suite of products that Valen offers made the difference,” says Crum. “It was a lot of work to develop a model to get the data and ensure the model was accurate, but they have a great tool that our underwriters looked at and liked. It produces a quantifiable result. That's why we decided to go forward with it.”
Frank Winston Crum Insurance initially looked at internal options, but then decided to reach out to some of their existing relationships. One such relationship was with Willis Re, the carrier`s reinsurance broker.
“They bring a lot of resources to the table,” says Crum. “They`ve helped us throughout the years and recommended Valen. We did further due diligence and it took off from there. We were sold quickly once we looked at the product and talked with the people at Valen.”
Implementing predictive analytics and modeling into an underwriting system that isn't broken is no easy task, according to Crum, particularly with a system the underwriting staff is already comfortable using.
“You really have to convince the underwriters they are not being replaced by a computer program or a model,” he says. “There was some hesitation from the underwriting team when we brought up this idea, but it is a great partnership with Valen and their people. They were great at explaining things.”
A Valen actuary visited the Crum office and sat with the underwriting staff for a day, urging them to shoot holes in the Valen system.
“If they got a score they disagreed with, they were told to ask Valen for an explanation,” says Crum. “Once we were done with the training all the underwriters were on board with it, which I'm pleased with. We knew that would be the biggest hurdle. If the underwriters didn't buy in, this wasn't going to work.”
There were no such worries with the insurers IT department, which Crum explains was on board with the project right away.
“We're a pretty analytical company,” he says. “We have business analysts on staff and are analytical when it comes to decision making as an organization. This was the first time to use analytics in underwriting, but it wasn't as much of a culture shock. We got our data to Valen in a way they could develop an accurate and relevant model.”
Crum points out margins are thin in workers` compensation, so the analytics tool is just part of the puzzle. The model only knows what the staff put into it.
“You have qualitative and quantitative analysis on each risk,” he says. “Our underwriters are not completely dependent on the model. It helps them confidently price risks. On a good risk, where it looks like there is a lot of competition, they might credit the account more than they would without the model.”
The models have changed the way carrier examines risk, but Crum points out there haven`t been many process changes.
“When we get a risk we put a score on it and depending on the score we have pricing guidelines,” he says. “We have a good team and we trust their judgment. We are fine if they deviate, but they need to put it in the file and explain their recommendation for pricing.”
Implementation was completed in April, so it`s too early to tell what the overall effect of the Valen tool has been, but thus far the insurer is happy.