Insureds, when they submit a claim, have fundamental reasonable expectations. They expect the insurer will perform a thorough and objective investigation, a coverage decision will be based on an honest assessment of the facts, any expert the insurer retains will be qualified and impartial, and the claim will not be unfairly denied or underpaid. These expectations are not just reasonable; they are the cornerstone of every insurance policy. However, court cases and national media attention reflect insurers’ widespread use of biased experts to minimize claim payments. This practice, which has evolved into a form of institutionalized bad faith, not only impairs the insured’s right to benefits under the policy but also breaches the insurer’s implied covenant and duty of good faith and fair dealing. In essence, using biased experts violates contract law’s core principles and is illegal.
The practice is supported by structural deficiencies in the law and lax governmental oversight. In the early 1990s, the California judiciary joined a national insurance-reform movement. It recognized the “genuine dispute” doctrine for first-party claims. This safe harbor shields insurers from bad-faith liability if the law or facts supporting a claim are reasonably debatable. An insurer’s retention of an expert to opine on coverage issues, such as the existence or interpretation of facts related to causation, scope, or amount of damages, is generally sufficient under the doctrine to create a genuine dispute and avoid bad-faith liability. While retaining a biased expert undermines the genuine-dispute defense and evinces bad-faith conduct, the courts have eschewed issuing guidance to assess expert bias, allowing the practice to expand unchecked. Similarly, despite significant catastrophes and insurance scandals revealing the predatory practice, regulators have ignored consumers’ pleas for reform.
The Ninth Circuit is the lone exception. In a series of cases spanning nearly two decades, the court incrementally addressed the key issues, finally offering in Demer v. IBM Corp. LTD Plan (9th Cir. 2016) 835 F.3d 893, a well-reasoned paradigm of standards, factors, and presumptions for assessing and eliminating expert bias. Yet, except for a few federal cases, the Ninth Circuit’s direction has largely gone unnoticed, likely because the last two cases involved ERISA-based policies issued under federal statutory authority, exempt from bad faith liability, and evaluated under trust principles rather than contract law.
Despite the differences, the underlying obligations, incentives, and foundational legal principles are the same in both groups of claims. The cases and statutory frameworks overwhelmingly suggest that the Demer paradigm applies to both ERISA and non-ERISA claims, and the recent case of Bagramyan v. Gov’t Employees Ins. Co. (Cal. Ct. App., July 20, 2023, No. B315018) 2023 WL 4636118, without explicitly stating, firmly endorsed the Demer paradigm.
Demer’s Paradigm
Beginning with Guebara v. Allstate Ins. Co. (9th Cir. 2001) 237 F.3d 987, 996, the Ninth Circuit first recognized the problems associated with insurers’ use of biased experts in the claims process, identifying a handful of circumstances where an expert’s opinion would not immunize an insurer’s conduct under the genuine dispute doctrine, such as where the experts were unreasonable or the insurer deceived the insured, dishonestly selected its experts, or failed to investigate the claim thoroughly. The circumstances were not a test, nor were they even factors evidencing bias, but rather examples of unreasonable conduct that violated the duty of good faith and fair dealing.
The Ninth Circuit next addressed several examples of evidence that reflect bias in coverage decisions. In Hangarter v. Provident Life and Acc. Ins. Co. (9th Cir. 2004) 373 F.3d 998, 1010-1011, the court firmly acknowledged the substantial nexus between the insurer’s use of a biased expert and the genuine dispute rule, citing the Gueberra circumstances in holding that an insurer’s biased investigation of a disability claim “may preclude a finding that the insurer was engaged in a genuine dispute, even if the insurer advances expert opinions concerning its conduct.”
Applying the circumstances, the court provided an example of conduct from which the insurer’s selection bias could be inferred, holding that the insurer exhibited bias in retaining an expert who rejected the insured’s claims of total disability in thirteen of thirteen comparable cases. Similarly, in Nolan v. Heald Coll. (9th Cir. 2009) 551 F.3d 1148, 1152-1155, the court opined on the financial conflict of interest—compensation bias—that arises with experts in coverage matters and held in a disability case that an inference of bias is permitted where an independent medical review company and its physicians receive substantial work and monies from the insurer.
The Ninth Circuit’s analysis of biased experts culminated in Demer v. IBM Corp. LTD Plan, supra, 835 F.3d at 901-903, where the court offered a comprehensive framework to assess both the insurer’s selection bias and the expert’s compensation bias in claims investigations. After first identifying inference of bias as the standard for evaluating expert bias, the court held that the insured initially bears the burden of offering evidence of possible bias, which the insured satisfied in Demer with two simple metrics: the amount of compensation received by the experts and the frequency of claims investigated. The experts received between $125,000 – $175,000 per year from MetLife and worked on between 250 – 300 claims per year over the prior two years. The magnitude of these numbers alone was sufficient to “raise a fair inference that there is a financial conflict which influenced [the experts’] assessments.” However, the court noted the inference would have been even higher had the plaintiff also provided evidence showing the expert’s “parsimonious pattern of assessments disfavorable to claimants” or direct financial outcome in the claim.
The court added that once the insured met its initial burden, a rebuttable presumption of bias arises, shifting the burden to the insurer to show the expert’s impartiality. The court then distinguished between the structural conflict of interest arising from the dual role as both insurer and claim evaluator and the financial conflict that often occurs with experts, stressing the insurer’s reasonable measures taken to avoid the former (e.g., walling off the claims department from the profit center) differs from the measures taken to assure accurate claims assessment (e.g., providing an analysis of the experts’ opinions in other insureds’ claim files to show neutrality in practice).
The court further emphasized the experts’ lack of thoroughness and failure to use sound principles and methodologies in evaluating the claim, noting the experts performed only “paper reviews” of the insured’s medical condition, failed to explain why they rejected the credibility of the insured and offered erroneous opinions that conflicted with other medical reviewers. In holding that the insurer abused its discretion in denying the claim, the court looked at the “totality of the circumstances,” including the insurer’s use of biased experts and the expert’s lack of reliability.
In all, the Demer paradigm introduced the “inference of bias” standard, a rebuttable presumption of bias, and four non-exclusive factors for evaluating inferential bias: the expert’s past and expectant benefits for providing opinions; the expert’s patterns and practices; the expert’s failure to use reliable principles and methodologies; and the insurer’s reasonable measures to safeguard expert impartiality and reliability.
Demer’s Paradigm and Non-ERISA Policies
While significant differences exist between ERISA and non-ERISA insurance policies (e.g., auto, homeowner, and commercial general liability policies), the Demer paradigm applies equally in both groups of claims. Both groups of policies involve the same discretionary power conferred upon an insurer to evaluate claims and determine benefits, the insurer’s attendant obligation to perform a full and fair claim investigation, and the insurer’s abuse of power by retaining a biased expert to offer opinions that support full or partial denial or underpayment of claims.
Similarly, identical structural and financial conflicts of interest exist in both claim groups, where the insurer performs the same role as the claim evaluator and benefits payer, and the expert has the same financial expectant interest based on prior business dealings. Demer addressed the power, duty, and abuse by applying trust law to the ERISA statutory scheme. The non-ERISA policies address identical issues applying contract law to the implied duty of good faith and fair dealing. (See, e.g., Wilson v. 21st Century Ins. Co. (2007)42 Cal.4th 713, 720-723; Carma Developers (Cal.), Inc. v. Marathon Development California, Inc. (1992) 2 Cal.4th 342, 371-373; Egan v. Mutual of Omaha Ins. Co. (1979) 24 Cal.3d 809, 818-819; Cal. Admin. Code, tit. 10, § 2695.7(d).)
The principal inquiry into the expert’s leanings and flaws is identical in ERISA and non-ERISA claims. The analysis of expertise has evolved over the past two decades from evaluating an expert’s qualifications to closely scrutinizing the reliability of the expert’s testimony, with a critical examination of the expert’s underlying principles, theories, and methodologies and the expert’s interpretation and application of facts. (See generally, Sargon Enterprises, Inc. v. University of Southern California (2012) 55 Cal.4th 747, 769-772; Daubert v. Merrell Dow Pharmaceuticals, Inc., (1993) 509 U.S. 579.) A medical examiner’s diagnosis and reliability in a health, disability, or personal injury claim is unaffected by whether the claim arose under an ERISA or non-ERISA policy.
Likewise, in both groups of claims, the insurer is held to a high standard of care in evaluating claims: in the former arising from its statutory designation as a fiduciary, and in the latter from the characterization of the insurer-insured’s relationship as “special” or “quasi-fiduciary.” The same nine non-exclusive factors cited in Egan, supra, or its progeny are present in both contexts and support the characterization and higher standard of care. While each factor is relevant to the benefit plan or insurer’s misuse of biased experts, three hold greater significance in requiring a fair benefits determination: the adhesive nature of the plan or policy, the plan and insurer’s unfettered discretion, and the participant and insured’s vulnerability and blind trust.
California courts have only touched upon the core bias issues in other contexts, but they are remarkably consistent with Demer. For example, outside the ERISA context, courts typically evaluate bias using the probability-based “inference of bias” test—a standard adopted by the U.S. Supreme Court for judges and arbitrators. California followed the Supreme Court’s guidance, using a myriad of terms to describe the inferable bias standard in various contexts, such as the appearance of bias, the impression of possible bias, and intolerable risk of bias (see, e.g., Haworth v. Superior Court (2010) 50 Cal.4th 372; Natarajan v. Dignity Health (2021) 11 Cal.5th 1095), often interchangeably in the same case. This differing terminology suggests that the standard may exist across a spectrum of relationships ranging from slight (appearance of bias) to more significant (intolerable risk of bias) and that a higher evidentiary standard may be required under certain circumstances.
California courts have also briefly addressed several factors identified in the Demer paradigm. In Haas v. County of San Bernardino (2002) 27 Cal.4th 1017, 1025, the Court reflected on the bias that infers when a relationship contains a financial element, opining, “[o]f all the types of bias that can affect adjudication, pecuniary interest has long received the most unequivocal condemnation and the least forgiving scrutiny.” And in Michael v. Aetna Life & Casualty Ins. Co. (2001) 88 Cal.App.4th 925, 938-940, the court applied the inference-of-bias standard in concluding that an arbitrator need not disclose a relationship that is based on a social acquaintance, joint membership in a professional organization, or involving insubstantial business dealings but must disclose a substantial current, prior or continuing business relationship that involves financial consideration.
California law also suggests that the burden lies with the insurer in demonstrating expert neutrality. While the courts have again remained conspicuously silent on the issue, an insurer that interposes the genuine-dispute defense based on an expert’s fair and thorough investigation should be required under the Code to show impartiality and reliability, at the very least where the insured raises a weak inference of bias. (See, e.g., Cal. Evid. Code § 500). The most recent amendments to Rule 702 of the Federal Rules of Evidence similarly reflect the movement to place the burden on the party offering expert opinions to demonstrate by a preponderance of the evidence that the views are reliable.
Finally, as with Demer for ERISA claims, California courts apply a “reasonableness” measurement and a “totality of the circumstances” standard in evaluating an insurer’s abuse of discretion in unreasonably denying or delaying payment of a non-ERISA claim. (See, e.g., Wilson, supra, at 723.)
While critical differences exist between ERISA and non-ERISA policies, such as the deference on review, the measure of damages, and the degree of consideration the insurer must give to the insured’s interests, these differences are irrelevant to the core bias issues and the standards, factors, and presumptions for evaluating expert reliability. One federal court has already cited Demer in a non-ERISA claim, granting discovery of relational metrics (e.g., compensation and assignments) to show an inference of bias. (See, e.g., Leung v. UNUM Life Ins. Co. of America (S.D. Cal. June 15, 2023, case no. 22-cv-00767), 2023 WL 4056041 *6.)
Application to Bad Faith and Genuine Disputes
In non-ERISA claims, the biased-expert inquiry extends beyond benefits eligibility and coverage. An insurer’s use of biased experts weighs in the calculus of whether the insurer breached the duty of good faith and fair dealing by unreasonably delaying or failing to pay a claim, which includes a duty to perform a thorough, fair, and objective investigation.
The systemic use of biased experts may also constitute unfair business practices under the Unfair Competition Law. The biased expert inquiry also factors into insurers’ most potent defense in summary judgment proceedings to first-party bad faith claims—the genuine-dispute doctrine. In response to the defense, insureds invariably focus on the thoroughness of the investigation and the expert’s conclusions without considering the expert’s objectivity. And in the absence of evidence to the contrary, courts presume the expert is unbiased and the dispute is “genuine.” Hence, mere reliance on an expert by an insurer is generally sufficient to raise a dispute and defeat a bad faith claim.
The recent case of Bagramyan v. Gov’t Employees Ins. Co., supra, typifies the bias and genuine-dispute issues insureds face in summary judgment proceedings. It is a perfect example of an underdeveloped and likely erroneous decision resulting from the California judiciary’s failure to recognize the Demer paradigm or provide guidance on the key issues. Bagramyan may be most notable as the first state appellate decision nationwide to identify “inference of bias” as the reference standard applicable to experts—albeit only indirectly in an unpublished decision. Bagramyan is also the first case to recognize several Demer factors as essential in the bias calculus.
After correctly identifying the standard and factors approach to evaluating bias, the remainder of the opinion lacked more thoughtful reasoning or legal analysis. The factors in Bagramyan created, at the very least, a weak inference of bias, far more than necessary to survive a summary judgment motion. The insurer’s accident-reconstruction expert offered only experientially based conclusions, needing more independent verification for the principles and methodologies employed and precisely the type of opinions expected from a biased expert. The insured’s expert easily disputed the views. On the issue of metrics, the record reflected that the insurer “does not track how many times it has hired [the expert]” and “does not know how often [the expert] makes findings to support a denial of coverage.” These statements stretched the limits of credulity, and while no mention was made of the expert’s compensation in either party’s papers, those two statements alone were tantamount to an admission that the insurer failed to take reasonable measures to ensure expert neutrality or reliability.
Sufficient evidence of selection and compensation bias was present in Bagramyan to shift the burden, which the insurer did not and could not meet. Yet, in granting the insurer’s motion based on the genuine-dispute doctrine, the court stressed the insured’s failure to produce metrics and practices information—evidence exclusively in the insurer’s possession and control, which it withheld from production.
Had the court carefully considered the Demer factors or rebuttable presumption, the outcome would likely have differed. While the case appears highly flawed, it at least finally recognizes the key bias issues and decisively supports the Demer paradigm for evaluating expert bias in non-ERISA disputes.
The California courts’ full recognition of the Demer paradigm is long overdue. Each Demer factor is vital in genuine dispute and summary judgment analysis. Without guidance, insurers have every incentive to suppress the discovery of expert bias and shun reasonable measures to ensure expert impartiality and reliability, which is precisely what they’ve done for several decades.
Practical Considerations
An insurer’s duty to fully and fairly investigate claims implicates two pillars of inquiry (thoroughness and fairness). Yet, practitioners and courts mistakenly focus solely on whether the investigation was complete, not whether the investigation was performed objectively. It is a fatal strategy for most insureds, as retention of a biased expert, despite lacking objectivity, generally satisfies the full investigation pillar.
Demer levels the playing field by offering a simple roadmap for expert objectivity and reliability, forcing insurers to take reasonable measures to ensure fairness. Demer’s paradigm is simple to understand but challenging to implement since each factor used to evaluate inferential bias is subject to insurer manipulation.
The first of the four Demer factors examines the direct and indirect prior substantial business dealings between the expert and the insurer (including its representatives, such as attorneys, vendors, and outsourcers). The inquiry’s general thrust is on the total amount of compensation and number (frequency) of assignments that create a sufficient temptation for the expert to tilt the principles or facts underlying opinions in exchange for future business, either with this specific insurer or, in some cases, the industry.
It is the most critical factor for evaluating the expert’s bias and one for which courts generally grant discovery. In unique cases, particularly those involving professional trial experts, the inquiry may extend to the expert’s dealings with other insurers and the insurance industry. This factor is susceptible to insurer manipulation using a proxy or intermediary, as insurers then suggest a lack of selection bias because they did not directly retain or pay the expert.
Experts likewise attempt to neutralize this factor by suggesting the insured retained or paid them merely because the insured signed an authorization to permit the expert’s investigation and agreed to be liable for any amounts not covered by insurance. Both suggestions are misguided attempts to conceal the selection and compensation bias inherent in the practice: the insurer is still indirectly selecting a biased expert, and the expert’s expectant benefits are still conditioned on providing favorable opinions to the insurer.
The second factor detects patterns and practices evidencing the expert’s leanings in other insureds’ claim files (OICFs). As Bagramyan accurately notes, the focus is on whether the expert supports coverage denial or underpayment. Since this factor is also metrics-based, segmentation of the claims into categories using any number of considerations (e.g., the cause of damage, amount in dispute, retention pre- or post-denial, and retention for trial versus coverage) permits a deeper analysis and reveals superior insights into bias.
The insurer manipulates this factor by reframing the inquiry to focus on whether the insurer paid some amount on the claim, which is relevant only if the insurer paid the entire amount the insured sought under the claim. Although the information to determine full payment is often not included in the claim files, the information is likely present in cost-to-repair cases. It may be essential, as courts have universally held that “[w]here the parties rely on expert opinions, even a substantial disparity in estimates for the scope and cost of repairs does not, by itself, suggest the insurer acted in bad faith.” (See, e.g., Chateau Chamberay Homeowners Assn. v. Associated Internat. Ins. Co. (2001) 90 Cal.App.4th 334, 348.)
While metrics and the lack of reasonable measures to ensure impartiality may be sufficient to show unfair bias in cost-to-repair cases, a pattern and practice of substantially underestimating scope and costs relative to other insureds’ estimates may prove invaluable and is relatively simple to analyze.
The third factor scrutinizes the expert’s reliability, including whether the underlying principles are sound, whether proper methodologies are followed, and whether all facts were considered and correctly applied to the principles to arrive at reliable opinions. Key determinants include whether the expert’s views are disputed (or disputable), independently verifiable through objective testing, based on experiential or empirical analysis, or involve subjective interpretation of facts. The OICFs are again highly relevant to an expert’s reliability, as bias may appear in the inconsistencies between how the expert applied the principles, methodologies, and facts in the present claim versus how they were used across a spectrum of OICFs. Variances may identify precisely how the expert’s leanings are tailored to meet specific policy exclusions.
The fourth factor examines the reasonable measures taken by the insurer to ensure expert impartiality and reliability. While Demer focused on the insurer presenting metrics demonstrating the expert’s impartiality, better evidence is likely found in the insurer’s selection, approval, and performance monitoring practices, particularly where the expert is identified on a preferred or approved vendor list or performs substantial work for the insurer.
The insurer’s files should include disclosures made by the expert to obtain the assignment (or preferred vendor status) and periodic updates, including a description of any facts that may suggest bias, such as the percentage of work performed for insurers and the financial consideration received for performing such work. The insurer’s files should also include a review of all complaints made against the expert by other insureds or their experts. Of most interest are the identification and reconciliation of any specific reliability issues and the authoritative resources the expert relied upon. If properly documented and maintained, the insurer’s records should contain in one place the necessary facts to fully evaluate bias, and the absence of such records may factor into the bias calculus and intent for bad faith and punitive damages and alone provide sufficient grounds for more invasive discovery.
Finally, while state courts outside California are often hostile to discovering OICFs and communications with regulators, which can be costly and time-consuming, California courts generally grant discovery, although sequential discovery may be necessary. Metrics under the first factors are most accessible to obtain and typically will provide sufficient evidence to proceed with further discovery.
While insurers may argue that tax information is privileged, accounting invoice and payee information is not, nor is the tax privilege absolute when the accounting information is unavailable. The metrics also help define the OICF universe, with the added benefit of addressing a common insurer ploy to inflate the universe, increase the estimated costs, and avoid discovery. Insurers delaying attempts or failure to provide sufficient compensation and assignment metrics, as in Bagramyan, may also provide grounds to obtain the OICFs immediately. However, courts may first require some minimal showing that the expert’s conclusions are either disputable, experientially based, involve subjective determinations, suffer procedural irregularities, or reflect other indicia of flawed principles, methodologies, or application of facts.
Finally, contact details for other insureds, while also generally discoverable over the insurer’s privacy objections, subject to a protective order, are generally unnecessary to show bias, bad faith, and intent. The mere request often elicits judicial skepticism and creates unnecessary hurdles (e.g., objections based on probative value and “mini-trials”) that can defeat the entire discovery request. For contact details, focused sequential discovery (e.g., specific OICFs) or sampling may be appropriate.
Conclusion
Elementary requirements of fairness and impartiality are vital to every proceeding affecting a party’s rights. Comprehensive statutory frameworks and case law exist to ensure neutrality for a myriad of ultimate decision-makers, typically focusing on self-assessment, disclosure, and disqualification.
Though imperfect, the laws guard against egregious forms of inferential bias. Yet, no framework exists for experts providing opinions for coverage in a quasi-adjudicatory, non-tribunal setting, where the insurer maintains unfettered discretion and the insured needs due process protections. The U.S. Department of Labor is the only agency that has taken meaningful steps to remediate the expert bias problem—over the objections of the insurance industry. Shortly after Demer, the DOL amended the ERISA regulations to require unbiased claim investigations and evaluations by plan fiduciaries. Yet these amendments failed to offer apposite guidance on the necessary expert disclosures or the factors for assessing experts, except for a relatively minor reference to reputational considerations (e.g., pattern metrics).
Considering the ubiquitous presence of experts in the legal and dispute resolution realm and even more significant presence in the insurance claims arena, the conspicuous silence by regulators and courts to eliminate “hired guns” and “junk science” is disconcerting. While legislative action is likely necessary to end this systemic problem, and clarification work remains for the Demer paradigm, at the very least, Demer provides a well-developed roadmap for courts and practitioners to eliminate expert bias.
Note: This article was accepted for publication in a forthcoming issue of Advocate (www.advocatemagazine.com).