The right analytics approach could yield substantial savings
The “80/20” rule is commonly applied to workers’ compensation claims: namely, that 80 percent of the cost comes from 20 percent of the claims. What if you could consistently predict whether a claim would fall in that costliest 20 percent so you could handle it in a way that mitigates the risk?
Workers’ compensation insurance is a mandate and virtually all employers carry this expense, whether they self-insure or insure through an insurance carrier or a state fund. While the costs vary by state, industry, employer size, and type of worker, it’s safe to say that the potential spend can be significant; loss ratios of 70 percent are not uncommon (for example, a large insurance company may have $1 billion in worker’s comp premium and $700 million in associated claim losses).
Tracing a claim—where is the potential opportunity?
A typical workers’ comp claims process begins when an insurance carrier (or a company’s internal claims department or its third-party claims administrator) receives a First Report of Injury (FROI) detailing an incident. This is the critical decision point because limited resources must be allocated to a subset of claims. Considering the 80/20 rule of thumb, where 80 percent of the cost comes from 20 percent of the claims, you will want to apply your limited resources to the costliest, most risky 20 percent.
For example, say the FROI documents a slip-and-fall injury. Depending on the nature and severity of the injury, a claims department assigns a junior, mid-level, or senior adjustor to handle the claim. Perhaps a nurse consultant is also assigned, an independent medical exam is required, or another type of intervention is called for.
This quickly becomes an issue of resource allocation optimization—determining how to most effectively handle that claim in a way that yields a favorable outcome for the patient while managing costs and resources. If the claims department had a way to know up front if that claim was in the riskiest 20 percent, it could handle it accordingly. However, identifying those risky claims has not traditionally been easy.
At Deloitte, we’ve done studies with insurance companies looking retrospectively at their closed claims that fell in that worst 20 percent. A panel of senior claims adjustors was asked to review those claims (blindly) to see if they could identify, based on the information known on day one (FROI) when the claim was initiated, which of those claims fell in the 20 percent category. In every study, senior adjustors could only correctly identify about 35 percent of those claims as being high cost/high risk.
This percentage is particularly significant considering that companies don’t have a panel of senior adjustors reviewing every newly arising claim. Claims come in, are quickly reviewed, and then acted upon. It’s not hard to imagine an even lower percentage of correctly identified high-cost/high-risk claims in actual day-to-day practice.
This is where the potential for significant savings opportunities lies. If claim complexity and severity is only identifiable up front about one-third of the time, the challenge—and opportunity—lies in figuring out how to identify the remaining two-thirds of risky claims. This is where analytics comes in.
Applying analytics to improve claims risk management
Analytics involves data science, data mining, and looking at information about claims that adjustors would not typically take into account, combined with mathematical analysis of the various risk factors to estimate the potential severity of a claim and whether it falls in the costliest 20 percent.
Currently, claims analytics solutions exist from various vendors, but vary in their ability to correctly identify risky claims. The more sophisticated approaches integrate with the organization’s risk management information system (RMIS) and provide not only higher accuracy but also higher levels of specificity. So for example, they not only identify a claim as potentially high risk in general (e.g., a lower-back sprain vs. a cut finger) but also score its level of risk within this type of injury (e.g., whether it is likely to be in the costliest 20 percent of lower-back sprains) and provide rationale for why that score was assigned.
Having this analytical assessment up front aids in assigning the right level of resources to effectively manage individual claims to mitigate cost and risk and achieve more favorable outcomes for the injured employee. Without this assessment, organizations risk devoting more resources than necessary to some claims, while not devoting enough resources to others.
Significant cost savings are possible
Deloitte has conducted multiple long-term (five-year) actuarial studies with organizations using our analytics solution design. Comparing the “before state” to the five years after implementing our solution, the organizations’ average claim cost had been reduced 12–15 percent each year. This is particularly noteworthy because worker’s compensation claims costs typically mirror health care costs and rise every year.
For a company that spends $700 million annually on workers’ compensation losses, a savings of 12–15 percent means $84 million¬ to $105 million in savings every year.
Evaluating vendors and solutions
While these types of analytics solutions have become more prevalent over the last 10 years or so, their results vary greatly. Insurers and self-insured companies would be wise to consider potential tools with a critical eye, especially looking at whether a potential solution demonstrates results over time. Because workers’ comp claims have what actuaries call a “long tail,” meaning they can take years to close, shorter-term results, say of only a year or so, cannot be actuarially accepted. Workers’ compensation measurements can take up to 5 years to actuarially evaluate and book to your financials, and results can only be booked if the actuaries accept the results.
Therefore, longitudinal case studies of 5 years (i.e., 60 months) are the benchmark to look for when evaluating vendors and solutions. These studies should include 2 accident years of workers’ comp data before implementation of the solution to provide a baseline, and then 60 months of solution usage after implementation—all to measure the Accident Year 1 impact.
If this seems like a considerable amount of effort, consider the potential payoff: Double-digit savings on workers’ comp claim costs every year make the due diligence effort well worth it.
Also consider that insurance companies are increasingly using these advanced predictive analytics models and it may be in your business interests to do the same. A 2017 study by Rising Medical Solutions1 found that 24 percent of insurers surveyed were using advanced analytics in 2014; by 2016 that number had risen to 32 percent.
In other words, roughly one-third of insurers (your competitors perhaps?) could already be realizing analytics-driven savings on workers’ claims costs that they can potentially pass on to customers in the form of lower premiums. Delaying your own use of claims analytics could soon put you at a disadvantage in today’s highly competitive marketplace.
1 Denise Zoe Algier, 2017 Workers’ Compensation Benchmarking Study, Claims Management Operational Study: Quantifying 3-Year Progress, Expanding Claims Differentiators, Rising Medical Solutions.