In the 1960s, a political scientist observed that future measurement techniques and databases on judicial behavior could render Supreme Court justices' votes on issues predictable. In effect, he suggested that individual justices would be superfluous to judicial decision making in a Brave New World of artificial intelligence: Justice Computer would rule.
Well, it is the 21st century, and Supreme Court justices are still with us; however, when I read insurance industry commentary and the advertising promises made about the software offerings currently touted to insurers by various vendors, I reflect on the professor's almost blind faith in technology. Propelled by insurer interest in cost cutting, technological solutions are taking center stage. I wonder if the industry's "Claims Professional of the Year" award will go to some software program sometime during this decade.
It is not just the omnipresence of software solutions that arouses my concerns. Where the software comes from is also troubling because most vendors promoting outsourcing of claims have neither a current foothold in nor previous experience in the insurance industry. Consequently, promoters of claims outsourcing most likely are unaware that insurance policies are contracts of utmost good faith; that old principles of caveat emptor are inapplicable to insurance; that there are established principles of how claims are to be managed; and that insurance is a business "affected by the public interest." 1
In addition to a likelihood that the business values of outside vendors are those of their own trade or occupation, these vendors operate within a regulatory no-man's-land. For example, the National Association of Insurance Commissioners (NAIC) Market Conduct Examiners Handbook incorporates standards insurers are to follow in relationships between insurance companies and certain regulated entities (third-party administrators and managing general agents, for example). Yet, regulators are silent about how insurers should control nonlicensed entities that provide software shaping insurer determinations of how a key decision affecting policyholders' claim payments are made. This is no minor regulatory oversight. To paraphrase the late Chief Justice Rehnquist, for most policyholders, the claims process is what brings the insurance contract to life. 2
Regulatory laxity regarding software use in claims adjudication deserves greater attention because of its expanding use, because the duty of an insurer to handle claims is nondelegable, and because controversies involving software are currently monopolized by the courts. This commentary addresses some of the problems confronting insurers that rely on outside vendor software in resolving claims. It also offers suggestions for new regulatory oversight of claims software.
How Insurers Get into Trouble
Perhaps the most obvious way for insurers to face difficulties from the errant ways of selected software is the theory of agency. 3 Under agency principles, personnel and companies retained to assist in claims handling are agents of the insurance company (the principal). Since the insurance company chooses agents and defines the scope of their work, an agency relationship exists in much the same way as in the sales and marketing process, in which insurers appoint producers to sell their products. In marketing and sales, whatever representations are attributable to the agent usually become the company's representations. The insurance landscape is littered with examples of how insurance producer misconduct causes insurer problems with both regulatory agencies and plaintiffs' attorneys.
Under agency law, the same types of issues confronting insurers in marketing and sales may surface in the use of external claims aides. Just as insurance producers must follow certain standards of conduct (examples include regulations on replacements, sales and advertising, unfair trade practices, and fraud), those to whom claims duties are outsourced must also comply with principles governing the claims process. By way of example, some of these claims standards include the following.
1. Claims adjusters should be competent to perform the job they are assigned. Questions of competency may arise with computer software if an adjuster is improperly trained to use or interpret the information generated by software manipulations. This point is especially important if the adjuster is unable to explain how the software operates when challenged. In Meier v. Aetna Life & Cas. Standard Fire Ins., 500 N.E.2d 1096 (Ill. App. 1986), the plaintiff insured his 1962 vehicle for a "stated amount" of $5,000. After the car was totaled, the adjuster entered information about the vehicle into a computer program provided by Certified Collateral. When the program returned a figure of $2,000, the adjuster followed company procedure and offered to settle the claim for that amount. What the computer told him was the sole basis of his decision. At trial, the adjuster admitted that he had no idea how Certified Collateral arrived at the $2,000 figure. Although other issues were also before the court, the adjuster's inability to explain this detail about the claims procedure did not help the company's defense.
2. Fairness and impartiality are claims standards. Compliance with this standard is measurable by several indicators. For example, if analysis of the insurer-vendor relationship shows that a major incentive for vendor selection is the vendor's promise to cut claim costs, the relationship is almost assuredly suspicious in nature. No claims standard in statute, rule, or common law stipulates that claims settlement should be cheap or based on the lowest possible number. Thus, a blatant objective of saving money casts doubt on the fairness and impartiality of the investigative activities.
An example of how this kind of relationship can plague insurers appears in a line of cases stemming from the use of software packages used in valuing uninsured/underinsured motorist bodily injury claims. Perhaps the best known among the products offering assistance with bodily injury claims is Computer Science Corporation's (CSC's) software, "Colossus."
According to CSC's Web page, Colossus is a knowledge-based system for assimilating and examining facts relevant to bodily injury claims. According to CSC, it includes over 12,000 rules and more than 720 injuries to help adjusters assess the degree of pain and suffering, degree of permanent impairment, and the effect of the injury on a person's lifestyle. CSC contends that over 33,000 claims handlers use Colossus worldwide. In the United States alone, over 19,000 adjusters use Colossus in evaluating more than 50 percent of all bodily injury claims. During its 12 years of existence, Colossus has attracted proponents but also a group of critics as well. 4
Because Colossus's marketing rests on a promise to cut bodily injury costs, critics allege that insurers use Colossus and similar software programs:
- to reduce the amount of money paid on bodily injury claims and to withhold from the insureds alleged faults with the software programs that led to underpayment of bodily injury claims.
Hensley v. Computer Sciences Corporation, U.S. District Court, Western District of Arkansas, Case No. 05-CV-4034 (2005). 5
Colossus's own evaluations of its product indicate that it makes bodily injury claim estimates more consistent. One can reasonably infer from the promotional commentary that consistency is based on reduction of the variance from the average claim settlement figure. Insurers, of course, like averages. The lower, the better, and that is an opportunity offered by Colossus. Even statistical novices know, however, that averages alone offer little meaning. As one authority on the subject avers, the average American has one testicle and one ovary.
Another example of how cost-cutting imperatives can cause problems appears in McGowan v. Progressive Preferred Ins., 637 S.E.2d 27 (Ga. 2006), a recent Georgia case. In McGowan, plaintiffs alleged that insurance companies, including State Farm, Atlanta Casualty, and Progressive Preferred Insurance, cut a deal with CCC Information Services to intentionally undervalue automobile property damage claims on totaled vehicles. CCC uses a computer program as a "data base" from which it derives a vehicle's estimated value in the local market.
McGowan determined that insurers could not rely upon the appraisal clause of an insurance contract as a defense against allegations that they had used CCC to engage in a deceptive method of reducing costs. The unanimous Georgia Supreme Court ruling tracks a similar case involving CCC in Louisiana, Hayes v. Allstate Ins., 758 So.2d 900 (La. App. 2000), in which the court determined for several reasons that the method used by CCC to value vehicles was "not reasonable." Louisiana's Court of Appeal, Third Circuit, reiterated concerns about CCC in Clark v. McNabb and Shelter Mut. Ins., 878 So. 2d 677 (La. App. 2004). 6
3. Proper claims procedures require an adjuster's strict compliance with regulations governing the particular type of claim. Computers are not exempt from the requirement. In Meier, for example, the court determined that an Illinois rule in effect at the time stated that if the value of a car is not published in a recognized source, the insurer must secure at least two written dealer quotations assessing the retail value of the vehicle upon which settlement can be based. Deviation from the rule must be carefully detailed in claim records. Certified Collateral's computer program apparently did not comply with this rule, to the detriment of the insurer. Since other states may have similar rules, software programs that fail to incorporate adjustments to these requirements foster noncompliance with mandatory claims requirements.
4. Claim files should include all details of how the claim was handled, a standard I characterize as "transparency." If software programs are relied on to arrive at a loss estimate, the claimant should be informed about its use and what part it played in the adjuster's recommendations. Since software creators may prefer to maintain secrecy about their product's involvement in claim decisions, a claims file may not contain evidence of the software's use or the numbers entered into the program while the adjuster was evaluating the claim. Adherence to the outside vendor's protocols for secrecy of software usage is inconsistent with the transparency principle. Any marketer of software on which claims estimates are based should not be permitted to enforce any contract prohibiting disclosure about the product's effect on claim determinations.
Alternatively, regulators should prohibit insurers from using unverifiable estimating methods. To insist on anything less not only defies normal expectations about what should be in a claims file but also forces a claimant into a game of shadow boxing to challenge an insurer's decision. Vendors that object to transparency disclosures have an easy option: Don't do business with insurers unless you are willing to adhere to the principles applicable to the industry..
With regard to the regulatory no-man's-land, some form of regulatory oversight of outside vendors is long overdue. The authority to develop the oversight is readily available under licensure, unfair trade practices, unfair claims practices, and form and rate filing provisions common to state insurance codes.