Looking Into the Future to Control Your Workers’ Comp Costs

Leading Self-Insured Groups (SIGs) are finally embracing “BIG DATA” trends and initiatives are underway to focus the power of machine learning on the claims and underwriting operations of SIGs. Advanced analytics and predictive modeling are helping SIGs improve loss ratios, enhance dividends and improve their market position. This was the subject of a 2015 SIIA Workers’ Compensation Executive Forum session. The speakers for this session were:

• Freda Bacon, Fund Administrator, Alabama Self Insured Workers’ Compensation Fund
• Steven J. Link, Executive Vice President, Midwest Employers Casualty Company
• Stan Smith, Predictive Analytics Consultant, Milliman, Inc.
• Stu Thompson, CEO, The Builders Group.

Underwriting

For years, underwriters tended to rely on gut and intuition versus analytics in evaluating prospects. This was referred to by a speaker as the “Magic 8-Ball” approach because there was no science behind the decisions that were made. To start using analytics in the underwriting process, you have to be committed to it and willing to determine price based on the model. One of the speakers who runs a large SIG indicated that, when they started utilizing the analytics underwriting model, they lost over 150 accounts with over $8 million in premiums. However, as a result of this action, they have significantly improved their profitability because the accounts they lost were draining resources from the profitable accounts. The analytics models for underwriting can also be extremely useful in focusing loss control efforts on the accounts in most need for such services. Another useful piece of this on the underwriting side is identifying your profitable business that is overpriced and ripe for significant competition.

Claims

On the claims side, these analytics can assist in identifying claims that have the potential for adverse development. Data can be used to determine if intervention is needed handle these claims differently so that better outcomes can be achieved. Many companies feel that they lack sufficient data for such analytics, however, studies have shown that most employers, carriers and TPAs are only using about 15% of the total data that they have access to when it comes to analyzing claims. In addition, the detailed treatment codes and the diagnosis codes in the medical bills contain a significant amount of data that is extremely useful for analysis. Analytics can also be used to adjust case loads for those handling more complex claims to better focus the efforts of nurse case managers.