Claims Auditing 101: Part 1

For each category management must look not only at the current year, but also for at least the past five to seven years, as many of the claims in each category may have a long tail where the final numbers will not be known for many years. Actuaries also use this type of data to project final claims cost for each policy year. Using trends, they can set premiums based on the combination of claim payments and allocated expenses plus the reserves, which provides the pure premium on which the final premium will be based. Assuming the average number of policies written for any one line remains static from year to year, and factoring in inflationary factors such as rising medical costs, management can get a good idea which line of business might be "out of line."

Suppose that management notes a severe rise in average claims costs and the number of losses in one of the areas, such as homeowners' property. Further analysis of that data is needed. One key issue is where the loss is occurring. It may be in the Midwest. When the data is further broken down by cause of loss, management may observe that loss for a specific state doubled over the past three years, with the primary causes being wind and hail. If the number of policies written in that state remained stable over the past four or five years, then management needs to determine if some other factor is coming into play.

Perhaps the insurer's new business has increased in what is sometimes called a tornado alley, while decreasing in other locations. The data may break down into sub-categories such as to which agency sold the coverage, where that agency is located, and similar data. Unless there was known to be an exceptionally high number of storms in that area during the year, there might be some other factor in play. It might be an agent who advertises locally for business or who writes applications that will appear favorable to the underwriter. There could be an underwriter for that district who is accepting risks that ought to be rejected. Or, the cause could be an overvaluation of the homes being insured, or even a change in the law, adding wind to fire in a valued policy statute. Analysis allows insurers to try to find specific causes of high loss ratios and take remedial action.

Spikes in losses may have other causes as well. Perhaps a major insured company has produced a product that is triggering claims. That, too, would call for remedial action Is the insurer still insuring that company? Is the insured still producing the product? Has his or her premium been adjusted or put on a retro-basis that incorporates a cost-plus factor? How was the product underwritten? How are claims being handled? Has every claim been litigated, or are all of the claims being admitted and settled without dispute? Each of these issues can affect the insurer's loss ratio and bottom line, perhaps even its survival.

Self-Funding Entities
Many commercial entities self-fund much of their loss through methods of high deductibles, retained and thus uninsured risk, use of a captive or risk retention group, or with pure self-insurance, which requires a large base of homogeneous exposure units. Much of what is commonly called a self-insured retention is actually just a high deductible or fronted insurance coverage where the insured funds the losses up to a specific level, at which time the insurance becomes effective.

To analyze the claim data the entity needs loss runs—the print-outs that come monthly or quarterly—that list each claim showing the names, dates, locations, causes, payments, and reserves. In retrospectively rated policies, loss runs are the basis for the annual or semi-annual premium adjustments. These loss runs, however, serve many other purposes as well. Many self-funding companies also use a risk management information system (RMIS), in which all of the individual claim data is available online and can be manipulated to produce reports on any specific issue.

Let's suppose the loss runs show an increase in workers' compensation claims in a particular branch location. If the RMIS has the capability of breaking down the data by location and cause, then the report will show all of the factors such as who, when, where, why, what, and how much. This may isolate a problem. What is causing so many injuries? If the date of loss was the same for many of them, then it could indicate some sort of accident involving employees, such as a car accident in which a number of employees were injured. Did the risk manager know about the accident? Did he or she take steps to find out what caused it, or who was at fault, or if there is a potential for subrogation? If fault was involved, were excess insurers notified? Just because the insurer's claims department or the claims administration firm knows about the claims does not mean that the risk manager has been informed.

In a self-funded claim situation, this type of data can lead to better solutions to current claims and finding future problems before they occur, if, for example, the employees were all going to lunch and the driver was drunk. Or, if the accident causing a large number of claims involved some other factor—an explosion, fire, a machine breaking down, gas release—then the risk manager must know if the situation is to be properly remedied. Without detailed claim data reports, the risk manager is helpless.

In one case, a risk manager noted on the workers' comp loss run that there were seven carpal tunnel claims in a small branch office, all within a short period of time. This was an obvious abnormality. Upon checking, it was determined that all the involved employees were using the same physician, one who had been cited in a CBS "60 Minutes" story a few months earlier.

"What was causing it?" the risk manager asked. "They all blame computer keyboards," the manager said. "Okay, [then] who made those keyboards? [The keyboards are made by] IBM. So, let it be known around the office that we will be subrogating against IBM for producing a hazardous product." Every single claim closed within a week.

In another case, occurring in California where psychological stress was becoming a common source of workers' comp claims, the monthly check register showed that one particular branch had issued numerous medical checks to a Dr. Brown. The case numbers were cross-referenced to the claims, all of which were psychological stress claims being handled by an experienced adjuster. It was assumed that Brown was the psychologist either treating these claimants or doing the independent medical exams. Just to make sure, an inquiry was made to Brown. There was, in fact, no "Dr. Brown;" as the adjuster had opened a bank account in that name and was issuing himself checks. "Dr. Brown" went into the annals of "lessons learned the hard way." The adjuster went to jail.

Data Mining
One advantage to using a RMIS or carefully reviewing loss runs is that the data can be mined for a variety of claim-related factors. It can disclose whether reserves initially set in claims in certain categories are fairly accurate when the claims are eventually settled, or if they are wildly off-base. Inaccurate reserving can be a serious problem for an insurer or self-funding entity. If the adjusters are simply slapping a standard reserve on each new claim without ever updating it to reflect what the reserve should be, all sorts of mayhem can result.

When reserves are set too high, there is a tendency to settle claims for more than their value. It's the old "get rid of it" philosophy, but that can be expensive. On the other hand, if reserves are set too low, when the facts about the claim and the demands for high amounts of settlement come in, then the reserves have to be jacked up, distorting the claim patterns. Further, it tends to lead to unnecessary litigation. The adjuster may try to low ball the settlement offers to save face for having set too low of a reserve and end up setting a more realistic claim reserve along with a legal expense reserve. If claims are being handled in that manner, then they might as well be handled by a computer, as the adjuster becomes superfluous, hence unemployed.

Loss data can reveal whether the adjusters in any particular location are adding value to the claim file or simply spinning wheels. Management, either in an insurance company or in a self-funding commercial entity, needs to mine their loss data for trends or details in the aggregate. The macro approach is how the overall picture can be analyzed for problems and possible solutions to those problems. Most claims and risk managers don't have time to look at every individual loss, although they should be looking carefully at any loss where the reserve may trigger an excess exposure.

Likewise, no individual adjuster can be aware of the aggregate loss exposure. When an insured has a primary policy with an aggregate loss limit and has multiple layers of umbrella or excess loss coverage above that, the adjuster cannot know to what extent other claims happening around the country may be eroding the aggregate, requiring notice to the excess layer insurers. Further, the adjuster may not know the terms of the agreements as to how allocated expenses (such as defense costs) will impact the retained risk or deductible amounts. Loss data reports can keep track of such details. 

Over the next several months we will take a look at factors such as the preferred claims philosophy and individual claim file auditing.