Help

BI’s Article search uses Boolean search capabilities. If you are not familiar with these principles, here are some quick tips.

To search specifically for more than one word, put the search term in quotation marks. For example, “workers compensation”. This will limit your search to that combination of words.

To search for a combination of terms, use quotations and the & symbol. For example, “hurricane” & “loss”.

Login Register Subscribe

Predictive models increasingly being used as workers comp underwriting tool

Tech tool use expands to underwriting

Reprints
Predictive models increasingly being used as workers comp underwriting tool

With insurers seeing lackluster investment income on workers compensation premiums, some are using predictive models to improve the accuracy of their underwriting and pricing.

Predictive models often are used in workers comp to reduce claim costs and duration. But as the technology evolves, insurers are looking for models to boost their comp underwriting results and help select profitable accounts.

Expanding predictive analytics to workers comp underwriting is the “next logical step” in the evolution of modeling, said Michael Gavin, chief strategy officer for Duluth, Ga.-based PRIUM, a medical cost management company.

“There's data there (and) there are algorithms there that can be applied to the underwriting process,” Mr. Gavin said. “But I think we're at the very early stages of that. It's probably (only) the most innovative of claims organizations that are actually doing that.”

Models and the data they analyze are constantly changing to help improve claims management and identify areas where employers can improve worker safety, said Gary Anderberg, Philadelphia-based practice leader for analytics and outcomes with third-party administrator Broadspire Services Inc.

“I think what we're doing is a lot of experimentation, trying different techniques to see what kinds of results that we get, where we think those are going to be useful for our clients and where they will take them beyond the kind of trending analysis and fairly basic techniques that we've been using up to this point,” Mr. Anderberg said.

As companies look for ways to better use predictive models, insurers are considering whether they can use data collected by current models to reduce workers comp losses from inaccurate underwriting — particularly as insurers have seen minimal investment income in the past several years, Mr. Gavin said.

%%BREAK%%

“They're going to have to be more disciplined than they've been historically in the underwriting process because it is the only place you can make money right now,” Mr. Gavin said.

Lackluster investments have driven a recent influx of insurers considering predictive analytics for workers comp underwriting, said Dax Craig, CEO of Valen Technologies Inc., a Denver firm that develops property/casualty predictive models.

“We're seeing a big pickup in interest and ... engagement from insurers where, historically, they have been less interested or even ... unenthusiastic,” said Mr. Craig, whose company focuses primarily on workers comp models. “That has shifted tremendously over the past 18 months” in which insurers that previously “wouldn't entertain even a discussion now are racing to implement (predictive models). It's really a sea change.”

Insurers using Valen's predictive model for workers comp underwriting can see a 5.5 percentage point loss ratio improvement over time, Mr. Craig said. He said models objectively and efficiently weigh a large data set to evaluate a company's workers comp risk — something he said can be a difficult task for underwriters.

“We really believe that predictive analytics' best use is in underwriting,” said Mr. Craig, whose clients include small, midsize and large workers comp insurers. “It's really at the point of sale ... where those final decisions are made on how an insurance policy should be priced. And so our philosophy is that we help underwriters do that more effectively.”

Claude Yoder, Hartford, Conn.-based head of global analytics for Marsh Inc., noted that underwriters still play a critical role for insurers that use predictive models to price and underwrite workers comp policies.

“On the commercial line side, it's still an underwriter-centric culture and there's a lot of need for underwriting to continue,” Mr. Yoder said. “So I think the best companies are sort of marrying the science of predictive modeling with the art and judgment of underwriting.”

CNA Financial Corp. is working to evaluate potential customers in a similar way.

%%BREAK%%

The insurer has a predictive model that analyzed nearly 22,000 workers comp accounts in the past two years, said Bill Boyd, senior vice president of the risk control department in Chicago. The model evaluates safety and health programs that can affect an employer's workers comp loss experience.

As part of the analysis, CNA has determined factors that make some companies safer than others, Mr. Boyd said. For instance, companies that have certain safeguards against falls from high places have a 38% lower injury rate than companies without those protections. And employers that have specific rules against cellphone use while driving have 25% fewer workers comp claims than companies without such guidelines.

“This data is used in our underwriting process in terms of predicting accidents and, thus, it's used as we determine the quality of an account, which is part of our underwriting and pricing analysis,” Mr. Boyd said.

In addition, Mr. Boyd said the analysis has helped CNA recommend safety measures for customers that can reduce workers comp losses.

“It enables us to better help the customer understand their current exposure (and) the current cost of risk, and then (help) us prioritize where to make the investment which would have greater impact in reducing their risk,” said Mr. Boyd, speaking as a member of the American Society of Safety Engineers.

Underwriting models could play an increasing role in helping employers improve safety, Marsh's Mr. Yoder said. That's because companies could implement safety standards to improve their predictive model scores with insurers and lower their comp rates.

“As a company, you can play it back to the insurance market to say, "You should think of me differently because we've taken strides to actually improve our loss profile,'” Mr. Yoder said. “And it might be directly related to the insurance company's own underwriting model, because (the employer is) aware of the different dimensions that go into creating that model.”

Valen's Mr. Craig said he expects more insurers will look to predictive modeling to bolster their underwriting results as they search for a competitive edge in a difficult workers comp market.

“It's evolving significantly,” Mr. Craig said.