Organizational Rules in Insurance Fraud (6)

Consumer Insurance Fraud in the US Property

Post By: Insurance Top Stories

The process of insurance settlement is typically undertaken by a claims adjuster whose “...duty is to dispose of claims” (Hirsch, 1999, p. 5) as a matter of acceptance, rejection, or compromise (Hirsch, 1999, pp. 6-7). The claimant, having the burden of proof, is obligated to “...clearly demonstrate the actual damages claimed,” (Hirsch, 1999, p. 383) and in those cases where the claimant’s burden has been met, a good faith settlement is a regulatory (as well as contractual) requirement. Neither side possesses perfect information, and each can be expected to maximize their goals. Thus, the claimant attempts to recover all (both known and unknown) that may be obligated by the insurance contract, while the adjuster is interested in limiting financial exposure.

	How to Check Insurance Fraud?

This “give and take” process occurs in an environment in which the adjuster is confronted with a seeming unending stream of new claims, in a corporate environs concerned with the efficiency of operation; one where levels of investigation likely vary with the size, nature, and source of the claim as well as the organization’s investigatory prowess and goals[6]. The reported practices of Canadian insurers are likely analogous, in that, the settlement process is also subject to influence by underwriting goals and standards, and marketing goals, the latter subject to influence by the perceived promotional value of customer “service” as reflected in the practices of differential underwriting criteria, differential pricing, and differential claim payment procedures (Ericson et al., 2003). In sum, organizational considerations associated with market decisions are paramount in the identification, classification, evaluation and ultimate “disposal” of a claim. Firm success depends upon it. This generally results in a “negotiated” outcome – the settlement process in which in many cases the only parties directly involved may be the adjuster and the claimant.

Managing claims – particularly those illegitimately proffered – is important to firm success. What to do with a claim with suspicions as to its validity? How to ascertain validity? These questions of course, get to the heart of discriminating between those claims which are legitimate and must be paid, and those which are not[7]. Firms have evolved different and individualized responses to these questions, although they have also cooperated when in their collective interests.

Contemporary property-casualty insurers in this country have undergone a two-decade old revolution in their thinking and approaches to managing insurance fraud. By one (conservative) estimate, the total amount of funds invested by the industry in anti-fraud activities (not including the direct and indirect costs of claims adjusting and processing) was approximately $648 million in 2004[8]. An earlier and comprehensive, industry-wide, survey by the IRC of institutional anti-fraud practices was undertaken in 1992 and 1997, representing about 77 percent of the property-casualty market. Select findings included the following (IRC, 1997):

1. Self-reported industry expenditures in fraud control went from $200 million in 1992, to $650 in 1997;
2. Of the 150 companies responding, 98 percent had developed fraud control programs, with 76 percent having initiated internal special investigative units, up from about 50 percent in 1983, and 66 percent in 1992;
3. Larger insurers (those with market share 1 percent or greater) expressed greater satisfaction with effectiveness of their programs than did smaller insurers (those whose market share was 0.1 percent or smaller); and
4. Approximately, three-quarters of respondents indicated continuing increases in spending for anti-fraud activities.

A follow-up survey by the IRC in 2002 of 353 insurers revealed that most (82 percent) maintained anti-fraud programs, with nearly seven in ten indicating that they responded “thoroughly” to claims fraud (IRC, 2008a). Most were less pleased with their redress of premium, or application fraud.

A “layered” approach is typically employed by insurers as they process claims, beginning with fraud recognition training at the level of claims adjusters, including use of “fraud indicators.” The industry has promoted the use of these especially through NICB, and these have been tailored to line of insurance and type of claim. Whiddon (1998, pp. 1-7) described the status of statistical sophistication in the approaches employed in the 1990 s and compared the comparative strengths and weaknesses of the range of approaches the available. At that time there was no publicly available insights into the “stacking” practices/procedures in practice other than the general data from IRC inquiries.

Derrig’s proscription for claim processing elucidates the “sorting” required for validating and invalidating claims. Beginning from a derived pool of suspicious claims, a data mining exercise should identify those which are easily adjusted and routinely paid.

Those surviving claims should undergo additional investigation, resulting in those paid, and those surviving for additional analyses. Survivors would be referred to the SIU’s for review, resulting in civil proceedings, or referrals to external authorities for criminal proceedings. This sequential processing is intended to result in maximal validity in the identification and proper treatment of abusive or fraudulent claims (Derrig, 2002). The success of this process relies heavily upon the availability of database technology and analytical tools, investments in which the insurance community has been steadily increasing.

The Insurance Services Organization (ISO) has long been a repository of claimant information and serves as a clearinghouse for members seeking to store, manage, and otherwise analyze data on insurance markets. It has managed the referrals of suspicious claims from the bulk of industry members also utilized by NICB, and available to industry SIU personnel. In 2003 it introduced a claims scoring system to enhance insurers’ abilities to sort claims into those worthy of prompt payment, and those requiring additional investigation and possible referral to in-house investigation units. The procedure relies upon weighting of variables known to be predictive of fraud (scoring from 0 to 999). Generally, industry reliance on statistical modeling is described as an imperative by Coyne (2008).

The academic literature has recently included multivariate classification and auditing methods for fraud detection and classification, as well as subsequent internal auditing, although the number of such papers is rather small (this review identified fewer than ten). Generally speaking, these depend upon weighting a series of predictor variables (typically tied to “red flag indicators”) in statistical modeling of claims data with ultimate intention of classification as fraudulent (one form or another) or not, and warranting subsequent claims auditing by the SIU, or not. See for example the works by Belhadju and Dionne (1997) offering an expert system including software, (Dionneet al., 2003) on scoring and auditing, as well as (Tennyson and Forn, 2002, pp. 289-308; Forn and Tennyson, 2001). Viaeneet al. (2002, pp. 373-421) for a comparison of several different predictive methodologies/approaches to statistical modeling. The extent to which these approaches or their variants are in actual practice in industry is not known. Decisions about the evaluation and final disposition of a claim go beyond underwriting and marketing, however, since they are contexted by the external, legal and regulatory systems[9].