The most promising yet problematic innovations are in assessing claimants.
Workers' comp claim payers check a worker's past claims history in all searchable lines of insurance. They estimate drive time from the claimant's home to place of work. They may look at opioid use, smoking and credit scores. Drawing from patient questionnaires used by doctors, they can take into account the patient's self-reported pain tolerance and anxiety. They may note key words and phrases from a claimant's Facebook page and other online sources.
Digital assessment can truly help injured workers. It can accelerate intervention by medical experts. It can correct adjuster and case manager biases, for instance, about workers who report a real injury on the first day of employment, or immigrants who doctor shop because they do not understand our healthcare system. And it advises adjusters on what claims can be safely auto-adjudicated.
But as the saying goes: "All models are wrong. Some are useful." Statistical correlation, the workhorse of underwriting, stands like a sight-impaired guest before a banquet table. Facts about the worker, such as co-morbidities, can be totally missed in both designing and using a model. The system's authors may be unable to explain what causes what.
And households express justified concern about the intrusive use of personal information. In cunning and peril, individual monitoring ever ascends in society, thanks to the torrent of data that is now accessible under the catch phrase "Big Data." For example, a large retail store reportedly guessed correctly that a teenage customer was pregnant before her parents did, by analyzing her purchases.
Insurers who sell mandatory insurance have an axiomatic obligation to act in good faith for all parties. They must, in sum, explain their models.
A recent dustup over pricing of auto insurance serves as a cautionary tale. Consider two 30-year old women who have driven for 10 years, live on the same street in the same middle-income zip code, and seek minimum liability coverage required by the state. One is a single receptionist with a high school education who rents, has been without insurance coverage for 45 days, and never had an accident or moving violation.
And the other woman is a married executive with a master's degree who owns a home, had continuous insurance coverage, and had an at-fault accident with $800 of damage within the past three years.
The Consumer Federation of America tested the online prices of five leading auto insurers in twelve markets. In 60 percent of the 60 tests, the woman with a better job and education was quoted a lower premium than the one that was accident free.
The National Association of Mutual Insurers charged that the study's methodology was seriously flawed.
Workers' compensation claims payers should conduct independent fairness tests that are based on methodology that is beyond reproach. Retain an arbitrator and a statistician, both with insurance backgrounds, to confirm the underwriting is fair--and useful--to claimants.
If we value that kind of claims management, we should explain why in public.
PETER ROUSMANIERE is an expert on the workers' compensation industry. He can be reached at firstname.lastname@example.org.
April 12, 2013
Copyright 2013© LRP Publications