Practical Tips · Case Study · Factor Three · July 2026
Regulators documented the largest biased-expert operation in the industry’s history — then settled it with an affidavit, two unenforceable sentences, and a mandate to hire more of the experts they had declined to regulate.
Most case studies in this series dissect a single court decision. This one dissects a regulatory settlement — because the fullest documented record of the third Demer factor in operation is not a case but an institution: UnumProvident, the largest disability insurer in the country when its claims practices came apart in public in the early 2000s, and the settlements that were announced as the fix.
The multistate market conduct examination that preceded the November 2004 regulatory settlement agreement did not overlook the expert apparatus — it led with it. The examiners’ first documented area of concern was excessive reliance on in-house medical professionals: staff physicians and nurses deployed to review files and disagree with treating physicians. And they documented the operation’s signature device — denials of benefits for lack of “objective” medical evidence, applied to conditions that are diagnosed by self-report, under policies that contained no such requirement.
Readers of the Factor 3 checklist will recognize that device immediately: it is the unsourced principle — a standard imposed by no policy term and grounded in no identified medical authority — institutionalized across a national claims operation. The largest documented biased-expert scheme in the industry’s history ran on a premise no one could have cited a source for, because there was none.
The remedy, against those findings, reads like a study in misdirection. For the in-house staff at the center of the examiners’ concern: a professional-conduct affidavit — a signed statement of good intentions. For third-party examiners: a single clause requiring selection on “objective, professional criteria” without regard to prior results — with no criteria defined, no audit, no disclosure of financial relationships or retention frequency, and no enforcement mechanism. And alongside both, the corrective plan directed increased use of independent medical examinations — expanding reliance on the very expert category the settlement had left unregulated.
The template for a real remedy existed, and it was five months old. In Hangarter v. Provident Life (9th Cir. 2004) — a UNUM-family case — the Ninth Circuit had affirmed a $7.67 million bad-faith verdict on an examiner’s perfect insurer-favorable record and a retention letter, written by an in-house consultant who had never examined the claimant, that had already asserted there were no objective findings of disability before the exam occurred. Retention frequency, outcome records, pre-commitments, the vendor channel: the elements of a working disclosure regime were in the Federal Reporter when the settlement was drafted. It adopted none of them.
The reassessment numbers. Roughly 200,000 denied claimants were noticed for possible reassessment under the settlements. Of the minority who completed the process, more than four in ten had their denials reversed in whole or in part, producing over $650 million in additional benefits — a reversal rate no sound claims operation could generate, measured by machinery the settlements themselves created. The overwhelming majority of noticed claimants never reached reassessment at all.
Follow-up examinations in 2007–2008 measured compliance with the settlements’ own procedural benchmarks, found the paperwork in order, and pronounced the reforms a success. None tested whether biased experts were still being selected, or what outcomes the newly mandated examinations produced. The books were closed on the shell’s own terms — and the federal courts then spent two decades keeping the record the follow-ups declined to. In 2008, the Second Circuit described the company’s “well-documented history of abusive tactics” as evidence bearing on its conflicted decision-making. In 2015, a federal court described its reliance on “a tainted, highly paid Unum contractor.” Decisions in 2013, 2018, and 2022 carried the same fact patterns — paper reviews, discounted treating physicians, the objective-evidence demand — into the 2020s, and a 2021 study of the company’s post-settlement ERISA claims conduct concluded the pattern had accelerated after 2004. A practitioner account published in 2016 reported deposition testimony with the detail that makes the design legible: after the examination report, reviewers were instructed to keep evaluating claims the same way — but to stop using the words “objective” and “subjective.”
Meanwhile the settlement architecture began working for the industry: procedures mandated by the settlements are now cited by insurers as evidence of reasonable, good-faith claims handling. A reform that regulates optics does not merely fail. It arms the defendant.
Ask what any “reform” did about the experts. When an insurer points to a regulatory settlement or a compliance regime, the question is whether it required anything enforceable about expert selection, disclosure, or outcomes. Usually it did not — and the UNUM record is the demonstration of why that omission is not an oversight.
The invented standard is Factor 3 in its purest form. A report that applies a requirement sourced to nothing — no policy term, no field authority — is exhibiting the same species of defect the examiners documented at industrial scale. The twenty-check screen is built to catch it on the first read.
The disclosure regime still does not exist. No regulator, state or federal, has yet required insurers or their experts to retain and disclose the metrics that would expose the mechanism — financial relationships, retention frequency, outcome distributions, pre-commitments, the vendor channel. The evidence courts now ask claimants to produce is the evidence nobody was ever required to keep. That asymmetry is the industry’s inheritance from the settlements — and it is why the discovery programs in this series exist.
Read the documented dissection → See the bias-evaluation service →
The factor this record demonstrates: The Bias You Can Read in the Report Itself. The screen it validates: the twenty-check screen. Why the bias fight is the one that matters: The Loss the Insurer Is Happy to Take.
Distilled from the project’s primary-source UNUM research corpus: the multistate regulatory settlement agreement and exhibits, the California settlement agreement, the market conduct examination report and follow-up examinations, federal court decisions 2004–2023, the U.S. Department of Labor’s 2016 rulemaking file, and contemporaneous press and practitioner accounts. Characterizations of the settlements’ design are the project’s analysis of those documents. Educational and informational only; not legal advice.