The National Insurance Crime Bureau reported in late 2013 that questionable workers compensation claims reported to it rose by 28% over the prior 12 month period, suggesting a surge in bad behavior.  The increase was equivalent to one thirtieth of one percent of annual claims filed. 

Isn’t this most likely noise?

A few weeks ago, WorkCompCentral quoted the chair of California Fraud Assessment Commission as saying that fraud is “ripping apart” the worker’s compensation system of the state. 

Is it ripping apart more than before? As a journalist in work injury risk, I am no closer to understanding fraud and abuse trends today than I was ten years ago. Few discuss risk of fraud and abuse in context. Without context, is it hard to sense if aberrant behavior is increasing or decreasing, if preventive controls work or even what these controls might be.  

A retiring head of fraud investigations for Florida wrote in 2013 that over 19 years “Together we have identified $1,306,368,823.00 in total savings, made 3,866 referrals for prosecution, and obtained 1,959 indictments and 1,928 convictions. Great work everyone!”  It would not be so great if savings could have been three times greater with better controls and better law enforcement.

It is as if retailers, to curtail shrinkage, looked at arrests and convictions without attending to risk assessment and their in-store control strategies.  Absent context, the risks themselves are obscured.  Fraud, a legal term, represents a small range within a larger continuum of aberrant behavior. The more we read about fraud alone, the less we seem to know about what is going on.

To explain what context means, let’s look at truly massive risk in workers’ compensation. This is aberrant behavior by insurers or by malefactors who profit off enabling insurers.  The damage from a single venture can exceed a hundred million dollars.  

An Ohio court approved in 2013 a $854 million judgment (under appeal) against the monopolistic state fund, the Ohio Bureau of Workers Compensation, for enabling others to game its group rating program to the detriment of many of its policyholders. 

San Allen v. Stephen Beuhrer, argued for the plaintiffs by Seaman Garson of Cleveland, Ohio, presents an example of risk that is real, usually tightly controlled, and potentially extremely costly.
  
The risk in question is that parties would create special groups within an insurer’s total portfolio of policyholders and manipulate group pricing and enrollments in order to shift losses incurred within the group onto others, such as the main body of policyholders.  A promoter such as a business association selling insurance programs as a revenue source, can make a killing. It needs an insurer to enable.

To control this risk, insurers set rules on how special groups operate.  They engage actuaries to in effect certify these rules and find it they are effectively applied. The Bureau did neither until, according to the court, $854 million in damage was done.

The story begins after the Bureau permitted “group rating programs” in 1991.

What is a group rating program? Deloitte, which the Bureau retained in 2007 to review of its practices, writes that “group rating allows employers of similar business types to be experience rated as if they were one employer.”

I asked Kory Wells, product director of data analytics as Zywave, to explain the mechanics. Consider, she says, Little Joe’s Machine Shop, with five employees, which buys a policy with very little permitted range of premium.  Even with a century of no losses, its experience modification is unlikely to go much below, say, 0.93.  It paid a policy priced almost entirely on the entire machine shop community’s loss experience, not its own.  

In actuarial reasoning, large numbers reduce variability of results over time. Year to year changes in experience should be prudently incremental, not drastic and destabilizing.  A well-managed group of similar small machine shops with a combined employment of one thousand might, per Wells, be allowed a minimum experience modification of 0.42. Its premium is driven much more by its own losses.

The Bureau’s formal policies allowed its groups to enjoy experience mods of under 0.10. The groups achieved these actuarially outlandish low mods not by dint of safety programs, but by expelling participating employers that incurred an injury.  The expelled insureds carried their losses into the Bureau’s community bucket of their respective industry codes. The groups effectively exported its loss experience.

In essence (and somewhat simplified), the Bureau enabled some of its 500 group rating programs to operate this way from about 2001 to 2007.  For these six years, the Bureau, per the plaintiffs, allowed this practice to evade and turf over to  policyholders not enrolled in these groups about $100 million a year in premium costs.

And the Bureau permitted groups to enroll employers from diverse industries.  Without true homogeneity, groups have no real actuarial credibility, so the resulting discounts made no actuarial sense from the start. 

These flaws may be recondite to most people in Ohio, but to actuaries they stand out like fish in a barrel.  Deloitte wrote, “The current pricing structure has created substantial inequity in the premiums paid by different employers in the state of Ohio. The primary driver of this inequity is the current approach to group rating. Ohio’s base rates are much higher than those of other states, largely as a result of the significant off-balance created by group rating.”

The San Allen case involves a defined risk with specific controls to contain it, but failure of controls led to abuse. 

We will understand better fraud and abuse if we applied a similar approach to discussing fraud and abuse. What is the context? What controls are supposed to be working, and why do they not work? Law enforcers should be discouraged from reporting numbers out of context. 

Peter Rousmanarie is a consultant for workers' compensation claims administrators and vendors and a veteran observer of workers' compensation industry trends.