Written by Gregory Agnone, director of captive business development at Midwest Employers Casualty.
Imagine having tomorrow’s newspaper today. What would you do with that information? Think about the possibilities: prevent crime, pick stock market movers, play the lottery, bet on sports scores, prevent impending disasters. The possibilities are endless. Although not exactly the same, this is the power that we possess through predictive analytics in the age of big data.
Over the past few months, I have had the pleasure of speaking on captive conference panels from CICA in Tucson, AZ to the Connecticut Captive Annual Collaborative in Hartford, CT discussing this topic with believers, sceptics, and those just starting to grasp the concept. The premise of predictive analytics is the ability to predict future outcomes with reasonable certainty using data compiled from past experience.
The insurance industry is a data-rich environment. Most well-established companies have decades of results from which to draw relevant data. Data exists in various disciplines, from marketing through underwriting and claims.
Let’s focus on workers’ compensation claims handling as an example of the use of predictive analytics in the captive arena. One goal of every captive is to reduce overall claims. Reducing claims has a far-reaching ripple effect that in time will reduce the total cost of risk and improve the captive’s value regarding standard market options.
Predictive analytics uses variables specific to each claim to predict the possibility that a relatively small claim may become a difficult to manage migratory claim. My survey of carriers and third-party administrators using predictive analytics shows that the number of variables ranges from 20-48. Some of the variables include: body part injured, age, gender, current pharmaceuticals used, previous injuries, and PTSD along with co-morbidity factors such as obesity, hypertension, and diabetes. These attributes are fed into the model, which then scores each claim based on the variables. Scores that exceed a targeted benchmark are then “red-flagged” to trigger further review and intervention.
It is important at this stage to realize that predictive analytics is a tool and should be used as such. Without putting the new tool in the hands of a qualified professional, predictive analytics are just numbers on a report.
Further intervention often takes the form of introducing a claims specialist. The designated claims specialist handles pharmaceutical management, nurse attention, and rehabilitation. In other words, a treatment path designed to obtain an overall better claims outcome will get the injured employee back to work, away from opioid addiction, and improve the insured’s claims experience ultimately resulting in lower pricing.
Following the positive handling of claims using predictive analytics as a tool could lower future loss forecasting. Shortening the time to close claims should result in lower loss development factors and more certainty in the level of ultimate claims. As a result, lower loss funds will result in lower overall variable expenses such as commission payable, loss control, TPA fees, fronting fees, and reinsurance. This is due to these expenses often being pegged as a percentage of premium which has been reduced as the severity of losses is reduced. Taking this ripple effect a little further, the reduction of collateral needed and the release of dividends should follow as actuaries become more confident that loss development will not erode profitability.
Will this happen overnight? I’m pretty sure that the first year this happens will be thrown out as an outlier with little credibility. However, as results improve each year, it would be difficult to dismiss the impact of predictive analytics as an anomaly. Proactive captives and captive managers will be rewarded for their use of technological advances available to them with predictive analytics high on that list.
