Insurers as bloodsuckers. Claim denials: how the industry profits from unappealed denials, delays and underpayment

01 · The modelHow an insurer makes money: premiums upfront, payouts later and preferably smaller

A customer pays the premium at the start of the policy term and files a claim later, if something happens. Between those two moments the money belongs to customers but works for the insurer. Warren Buffett, whose Berkshire Hathaway is one of the world's largest insurance groups, explains it like this: “'Float' is money that doesn't belong to us but that we temporarily hold” 28 H. In 2025 Berkshire's float was $176bn; in 1970 it was $39m 29 H.

So an insurer earns money in two ways. The first is the underwriting result: premiums minus claims and the costs of running the business. The second is investment income on reserves for unpaid losses (customers' money) and on its own capital. The second is often more important. In 2022 and 2023 US property and motor insurers lost $22.4bn and $20.0bn on underwriting itself, but earned $71.5bn and $72.6bn in investment income and ended those years with net profits of $37.9bn and $87.3bn 8 H. In 2024 the industry's net profit reached $171.3bn, and in 2025, $146.2bn 30 H.

Simple arithmetic follows. Every dollar not paid out is either profit or one more dollar in the investment portfolio. Every day of delay is another day of income on that money. In Texas the law provides for 18% annual interest on top of a late payment, plus the customer's attorney's fees; for property-damage disputes under section 542A the rate is different: 5 percentage points above the judgment rate 31 M. In our view, such a rule should remove the benefit of delay; without it, delay could be profitable L. According to BCG, the 98 largest listed insurers created $1.48trn of value for shareholders in 2021–2025 32 H. The same report names “reducing claims payment 'leakage'” among the benefits of artificial intelligence for insurers 33 H. “Leakage” is what the industry calls payouts it considers excessive.

US health insurers: profit goes to shareholders

The five big US health insurance groups (excluding CVS/Aetna) — UnitedHealth, Elevance, Cigna, Humana and Centene — earned about $282bn of net profit in 2015–2025 and handed about $210bn to shareholders through share buybacks and dividends 35 H. Over those years UnitedHealth returned $104.9bn to shareholders, 70% of its profit 36 H. The ACA requires health insurers to spend at least 80–85% of premiums on medical care and to return the difference to customers. All such rebates in 2012–2025 come to $14.4bn 37 H, roughly what UnitedHealth paid its shareholders in the single year of 2025 ($13.5bn) 38 H.

In 2024–2025 the profits of these groups fell. In the first half of 2026 the combined profit of five groups — UnitedHealth, Elevance, Cigna, CVS and Centene (Humana excluded) — rose again, by 32%, to $26.9bn 39 H; excluding CVS, the increase is about 19% L. According to the compensation table in UnitedHealth's annual proxy statement, the total for the chief executive who returned to the post, for 2025, is $60.9m (according to press reports, mostly a one-off share award) 40 L.

The second lever is price

Between 2019 and 2025 car insurance in the US rose in price by 57%, while consumer prices overall rose by 26%. Measured from 2020, when prices were depressed by pandemic premium refunds, the figures are 64% and 24% respectively 41 M. Progressive's net profit rose from $0.72bn in 2022 ($5.7bn in 2020) to $11.3bn in 2025; Allstate's, from a loss of $1.29bn to a profit of $10.3bn 42,43 H.

Britain's regulator, the FCA, found that insurers raised prices precisely for the customers most likely to renew their policies. In 2018, 6m holders of home and car policies overpaid by £1.2bn compared with the average price for their level of risk 44 H. A driver who had been insured with the same company for more than five years paid £370 on average, while a new customer with the same risk paid £285 45 H. Since 1 January 2022 the renewal price may not exceed the price for a new customer 46 H.

02 · PayoutsHow much of each premium goes back to customers: from £17 to £68 in every £100

Britain's regulator, the FCA, publishes every year, for each general insurance product, what share of premiums insurers paid out and how many claims they accepted. It has collected these data since 2022 47 H. In 2025 the lowest payouts were on wedding and party insurance (17% of premiums), vehicle cosmetic insurance (18%) and personal accident cover as an add-on to another policy (21%). The highest were health cash plans (68%), lifetime-cover pet insurance (63%) and motor insurance (59%) 9 H.

Among the least valuable are policies sold as an “add-on” to a main purchase (a car, a loan, a phone): the same product pays back less as an add-on than as a stand-alone policy (personal accident: 21% against 34%) 9 H. In 2014 the FCA found that GAP insurance (which covers the difference between the debt on a car loan and the car's value after a crash) returned 10% of premiums to customers in 2008–2012, personal accident cover as an add-on returned less than 9%, and core home and car policies 64% 10 H. In 2023 the regulator found GAP sales channels in which only 6% of premiums went on payouts; firms that could not demonstrate the product's value agreed to suspend sales 11 H. In Australia ASIC calculated that add-on policies sold in car dealerships returned 9 cents in the dollar of premium in 2013–2015, against 85 cents in ordinary motor insurance. Dealers received A$602m in commissions, 4.2 times what customers received in payouts 48 M.

The same is true in the US. In 1997 credit insurance returned 38.7% of premiums, unemployment insurance sold with loans 12.6%, and insurance on property bought with a credit card 11.6% 49 M. Regulators considered 60% the minimum reasonable level for credit life and disability insurance 50 M. In 2015–2017 credit insurance against illness or injury returned 43.3% of premiums (NAIC data used by Indiana's regulator), and the state regulator cut the maximum rates for it by 15% 51 H.

An instructive example is British payment protection insurance (PPI). It was sold in vast numbers alongside loans, credit cards and mortgages in 1990–2010. After the scandal, financial firms paid back more than £38.3bn to customers by April 2021 52 H; the window for complaints closed on 29 August 2019 53 H.

03 · DenialsThe world map: how many claims go unpaid in Britain, Australia, South Africa, India and the US

There is no single global statistic on denials, because regulators count differently. Britain publishes the share of claims that the insurer accepted. Australia publishes the share declined among those assessed. South Africa and India publish the share of claims repudiated. The US publishes denials of claims submitted by providers. So we do not compare countries with one another. But in most countries the denial rate varies widely by type of insurance: it is highest in home insurance (Britain), in disability cover (Australia), in credit life insurance (South Africa) and in out-of-network claims (the US).

Britain. Motor insurers accepted 98.7% of claims, home insurers 62.4%. The share of home claims accepted has fallen from 67.6% in 2022 12 H. For every 100 home insurance claims there are 13 complaints to the insurer — about the decision, the amount or the timing 54 H. The FCA warns that firms report home claims inconsistently 55 H. According to Insurance DataLab's evidence to a House of Lords committee, the share of home claims accepted ranges from 45% to 85% depending on the insurer (the year is not stated) 56 M.

Australia. In the year to June 2026 life insurers declined 18.9% of assessed total and permanent disability (TPD) claims on policies bought through a financial adviser, and 28.9% on policies without an adviser (only 83 claims assessed). A further 37.5% of claims on adviser policies lodged during the year were still unassessed at the end of the period (some were lodged late in the period) 13,57 H. In 79.9% of denials on adviser policies the explanation was that the client's condition did not meet the definition in the contract 58 H. In 2013–2015, according to ASIC (15 insurers, all sales channels, a different methodology), 16% of disability claims and 14% of trauma (critical illness) claims were declined 59 H. By our calculation from APRA's totals, in 2026 across all sales channels 12.2% and 12.7% of assessed claims respectively were declined 13 L.

South Africa. In 2025 insurers repudiated 64,136 of 1,080,930 death claims — 5.9% (5.7% in 2022, 4.4% in 2024). In 2025 cell-captive business was included in credit life insurance for the first time, so the comparison with 2024 is not entirely like for like 60 H. In credit life insurance 6.9% of claims were repudiated, and 12.7% by value 61 H.

India. In fiscal year 2024–25 insurers repudiated about 8% of health insurance claims, and a further 5% were still unassessed at the end of the year 62 M. Complaints to the insurance ombudsman over full or partial repudiation of a claim rose to 35,186 in fiscal year 2025–26 from 11,942 in 2021–22 — 2.9 times as many. Decisions in the complainants' favour numbered 18,809 — 53.5% of complaints received (all types of insurance) 63 M.

The US. Insurers on the federal ACA marketplace denied 19% of claims from in-network providers and 37% of claims from out-of-network providers 4,64 H. The reason for a denial often remains opaque. KFF counts each reason separately, so these are shares of reasons, not of claims: 36% are “other or unspecified”, 25% administrative, 13% service not covered, 9% no prior authorisation or referral, and only 5% medical necessity 65 H. In KFF's 2023 survey, 18% of insured adults said that in the past year their insurance had not paid for care they believed was covered; of them only 15% formally appealed, and 69% did not know whether they had the right to 66 H.

One market, different denial rates

In the ACA marketplace in 2023 some insurers denied 1% of claims and others 54% (the extremes are insurers operating in a single state); 24 of 175 companies denied 30% or more 67 H. In 2024 the spread was from 3% to 36% 68 H. According to data that insurers published for the first time for 2025, in Medicare Advantage standard prior-authorisation requests were denied at rates from 5% (Elevance) to 17% (UnitedHealth Group) 69 H. So in one market a customer of one company is denied several times as often as a customer of another.

04 · AppealsThe appeals funnel: millions of claims are denied, a minority are appealed, and in Medicare Advantage most appealed denials are overturned

The key question is what happens after a denial. The fullest data are from the US. In 2024 insurers on the ACA marketplace denied about 85m claims from in-network providers. There were at least 262,982 internal appeals: according to KFF, that is under 1% of denials 70 H; by our calculation, about 0.3% L. In 66% of appeals the insurers upheld their denial 70 H.

In Medicare Advantage, the private plans for pensioners, insurers made 52.8m prior-authorisation decisions on treatment in 2024 and denied 4.1m requests in full or in part — 7.7% 71 H. Only 11.5% of denials were appealed, and the insurers themselves overturned 80.7% of those appealed, in full or in part 6 H. In 2019–2024, according to KFF, more than eight in ten appealed denials were overturned every year 72 M. According to insurers' own data for 2025, 67% of appealed standard denials were overturned 69 H. In a 2018 report (data for 2014–2016) the HHS Office of Inspector General (OIG) calculated that 1% of denials reach the first level of appeal, and that insurers overturn 75% of those appealed 73 H. In a 2022 report the OIG reviewed a sample of 250 prior-authorisation denials from the 15 largest insurers (denials from 1 to 7 June 2019) and found that 13% of them concerned services that should have been paid under Medicare rules 74 H.

Among types of post-acute care, the highest overturn rate is in skilled nursing facility (SNF) care after discharge from hospital. In June 2024 the largest Medicare Advantage insurers denied 12.3% of such requests. 18% of denials were appealed, and 94.6% of those appealed were overturned. At UnitedHealth Group 1,022 of 1,025 appealed denials were overturned (99.7%), but only about 19% of its 5,362 denials were appealed 14,15 H. Patients who appealed a denial typically (median) waited 6 days for a decision from the first request, and 17% waited 10 days or more. Approved requests were typically decided the same day 75 H.

Why do so few appeal? In administrative data, fewer than 1% of denials are appealed (ACA) and 11.5% (Medicare Advantage); in surveys the share is higher: among people who were denied, 15% appealed (KFF, 2023), and about half of privately insured adults did so (Commonwealth Fund, 2025) 66,79 H. In the KFF survey 69% of people who were denied did not know whether they had the right to appeal 66 H. According to the Commonwealth Fund's 2025 survey, 41% of privately insured adults who were denied prior authorisation said it delayed their treatment, and 28% said their condition worsened. Among those denied payment, 43% said it led to medical debt that they are still paying off 79,80 H. One of Cigna's internal documents, released by ProPublica, contains this estimate: only about 5% of people will appeal denials made under the PXDX process (a check that diagnosis and procedure match) 16 M. Hospitals fight denials, but it is expensive: according to a survey of 516 hospitals (2022 data) by Premier, a hospital purchasing organisation, almost 15% of their claims are denied on first submission, and 51.7% of denials are ultimately overturned. By Premier's own extrapolation, fighting denials costs hospitals about $19.7bn a year; the author of the STAT article is Premier's chief executive 81 M.

05 · AmountsLarge, medium, small: indirect evidence suggests that costlier claims are more often denied or disputed

No regulator whose data we found publishes denial rates by claim size. But indirect data from four countries — the US, Australia, South Africa and Germany — point broadly the same way: costlier claims are more often denied or more often end up in dispute. British data concern not denials but detected fraud.

The US: denials are more frequent in the costlier types of post-acute care. In June 2024 the 19 largest Medicare Advantage insurers initially denied 12.3% of requests for skilled nursing facility (SNF) care after hospital, 54.2% of requests for inpatient rehabilitation facility care and 65.3% of requests for long-term care hospital care 17,14 H. The average cost of a stay in these facilities under original Medicare in 2023 was about $16,000, $24,000 and $49,000 respectively; the pairing of prices and denials is ours 18 M. Across the whole of 2024, fewer than 8% of requests for all Medicare Advantage services were denied (CMS data; comparison with the OIG's one-month snapshot is approximate) 83 H.

The largest insurers initially deny the costliest of these types of care most often. UnitedHealth Group denied 65.8% of requests for inpatient rehabilitation facility care and 70.9% for long-term care hospital care, and CVS Health (Aetna) denied 79.6% of requests for long-term care hospital care 84 H. For-profit Medicare Advantage contracts denied more often than nonprofit ones in all three types of care; in long-term care hospitals, 67.2% against 38.7%. Nonprofit contracts are few: in long-term care hospitals, 222 requests against 3,008 85 H.

Hospitals describe a similar pattern. According to the chief executive of Premier, a hospital purchasing group, in its survey of its hospitals denials in 2022 more often concerned treatment costing more than $14,000 88 M.

Australia: for death and disability, declined claims are mostly larger than paid ones. In death cover bought through a financial adviser, the average sum insured on declined claims is A$792,000 and on paid claims A$589,000 (declined claims are 1.34 times larger). Without an adviser it is A$355,000 against A$252,000 (1.41 times). In total and permanent disability (TPD) cover without an adviser it is A$468,000 against A$302,000 (1.55 times; only 24 declined claims) 19 H.

South Africa: in three types of life insurance, the repudiated death claim is larger than the paid one. In 2025 the average repudiated claim on risk life insurance was about R164,000 and the paid claim about R115,000 (our calculation from ASISA totals): the repudiated claim is 1.42 times larger. In credit life insurance the repudiated claim was on average twice as large as the paid one (1.97 times), in funeral cover 1.36 times. In 2025 cell-captive business was included in the credit life statistics for the first time, and in 2024 the ratio there was the reverse (0.57) 20 L.

Germany: where there are more disputes over large sums, customers win less often. Germany's insurance ombudsman (Versicherungsombudsmann) hears complaints of up to €100,000. Among the types of insurance that its report singles out, the highest share of complaints over €10,000 is in four: occupational disability (Berufsunfähigkeit) (64.3%), life insurance (22.6%), accident (21.3%) and home insurance (17.1%); among all complaints such cases are only 9.1% 89 L. It is in these four types that the shares of complaints ending in the customer's favour are lowest: 29.7% in accident, 31.5% in occupational disability, 36.5% in life insurance and 37.8% in home insurance. In compulsory motor third-party liability (MTPL) insurance, where large disputes are only 1.1%, the share is 65.9% 90 H. “Success” here means the complaint being upheld in full or in part, mostly thanks to the insurer putting things right voluntarily, not the share of wrongful denials. The ombudsman's decision is binding on the insurer only up to €10,000 91 H.

Britain: fraud is found mainly in large claims. In 2024 the average detected fraudulent motor claim was about £11,100, while the average motor claim overall was £4,900: the detected claim is 2.3 times larger (our calculation from ABI data) 92 L. In 2025 the Association of British Insurers (ABI) described detected fraud in general insurance as “fewer but more costly cases” 93 M. So insurers find fraud mainly in large claims.

06 · InvestigationsWho checks claims, and how investigations end

Large insurers have special investigation units (SIUs). They check claims that look suspicious: they order inspections, demand documents, and hire detectives and “independent” medical examiners. In the US insurers refer about 2% of claims to these units as suspected fraud (routine checking by claims adjusters is not counted in these data). According to the Coalition Against Insurance Fraud (40 insurers, 2019), the figures are 1.8% of private motor claims, 1.9% of property claims and 2.3% of commercial claims 21 M. In Germany the insurers’ association GDV, relying on estimates by anti-fraud experts at the insurers themselves, considers about 10% of property and motor claims “suspicious” and in need of checking 94 M. GDV itself adds: “If a case is dubious, however, that does not automatically mean it is insurance fraud” 95 M.

No regulator whose data we found publishes what share of investigated claims ended without a payout. But regulators do publish how many insurers’ reports reached investigations and arrests. In 2024 New York insurers sent the state’s Department of Financial Services (DFS) 41,686 reports of suspected health-insurance fraud; the law obliges them to file such reports. Insurers marked 89% of them “for intelligence purposes only” or as claims they were investigating themselves. 4,585 were recommended for further review, the department opened 71 investigations, and there were 42 arrests: 0.1% of reports 22 H. Most of these are reports about medical payments under no-fault motor policies (compensation for treatment after a road accident, regardless of fault), not ordinary health insurance: 38,846 of 41,686 reports in 2024 and 36,835 of 39,676 in 2025, or 93% in each year 96 H.

The most telling part of New York’s report for 2025 (March 2026) is two sums. Insurers claimed $3–3.5bn in “savings” from the work of their fraud units in 2024, while recovering $31–40m from fraudsters: the report’s summary gives $3.5bn and $40.3m, and the main text, for the units of health insurers, HMOs and life insurers, gives $3bn and $30.6m 97 H. The report does not define what counts as “savings”: it is a figure the insurers themselves reported. For every dollar recovered from fraudsters there is $87–98 of claimed “savings” L. The sums are of different kinds, so the comparison shows only scale.

In Massachusetts in 2022 insurers passed 3,513 reports to the state’s insurance fraud bureau. The bureau accepted 1,514 for review, opened 1,245 cases and referred 160 to prosecutors: one report in 22 98 M. A ten-year study of this bureau, published by Richard Derrig in 2002 (as set out by the legal scholar Jay Feinman), found that there are about 25 reports of suspected fraud for every proven case of fraud 99 M.

SIUs take up few of the signals from automated fraud-detection systems: in 2019 only 15% of the claims the algorithms referred to them, in 2017 22% 100 M. That is not a measured error rate of the systems. Almost seven in ten insurers in a 2016 survey complained that their systems produce too many false alarms 101 M. Meanwhile SIU budgets as a share of premiums fell by 20% over 2017–2019 102 M.

Where the “10% fraud” figure comes from

The industry explains its vigilance over claims by the scale of fraud. By the estimate of the Coalition Against Insurance Fraud (2022), it costs the US $308.6bn a year 103 H. Two of the eight components, property and motor insurance ($45bn) and life insurance ($74.7bn), were obtained by applying a “10%” rule to payouts 104 H; together that is 39% of the estimate L. For property and motor insurance this 10% comes from a survey of claims adjusters that the Insurance Information Institute carried out in the late 1980s 23 H; the details of the survey, according to the legal scholar Jay Feinman, have not been made public 99 M. For life insurance the authors of the estimate cite the fact that industry organisations have widely circulated this 10% 104 H. Sharon Tennyson of Cornell University wrote as early as 2008 that this figure “appears to be more folk wisdom than fact” 105 M. The coalition itself admits: “The actual cost of insurance fraud on American citizens is impossible to measure accurately” 106 H. The large health insurers approached by the authors of the estimate did not provide their own fraud data 107 H.

Detected fraud is small. In 2024 British insurers detected 51,700 fraudulent motor claims out of 2.4m claims in total: 2.2% (our calculation) 108 L. The largest category is not invented accidents but “exaggerated loss”: 26,900 of the 93,900 cases detected in 2025 109 M. This is the grey zone between fraud and an ordinary dispute over the amount.

Surveillance and “independent” medical examiners

An investigation can be a way of not paying. The ombudsman of the Australian state of Victoria found that the private insurers acting as agents, who handle injured workers’ claims under contract with the state WorkSafe scheme, together spent almost A$9m on surveillance of them in 2017–18, and in numerous cases used it “without a shred of evidence to justify it”. The agents used the opinions of “independent” medical examiners selectively and went looking for a doctor with the desired opinion 110 H. As early as 2016 the same ombudsman found that bonuses and penalties for the agents were “driving a focus on terminating and rejecting claims to maximise profit” 111 H. At Australia’s banking royal commission the insurer TAL admitted that surveillance of a nurse who was claiming for loss of ability to work was inappropriate; her claim took more than three years to resolve 112 M. In the US, WellPoint opened more than 16,000 reviews of sick customers’ policies for possible rescission in 2008; 92% were closed without consequences 113 M.

07 · AlgorithmsFrom Colossus to nH Predict: programs that calculate the payout, and are tuned for savings

American motor insurers value compensation for road-accident injuries with a program called Colossus; according to the Consumer Federation of America (CFA) in 2010, most large insurers used it. The adjuster picks from about 600 injury codes those that match the injuries in the medical records, and the program produces a range for pain and suffering 114 M. The program’s vendor, CSC, promised in marketing materials cited by the CFA that “the program will immediately reduce the size of bodily injury claims by up to 20 percent” 115 M. In court documents a CSC executive explained that Colossus could be “tuned” to a set level of savings, such as 15% for all claims 116 M.

In the 1990s Allstate rebuilt its claims handling under a McKinsey design 117 M. When plaintiffs in Missouri demanded these documents, Allstate refused to hand them over, citing trade secrecy. A judge fined the company more than $7m, at $25,000 a day. After a ruling by the state’s Supreme Court, Allstate handed over more than 120,000 pages 118 M. One McKinsey slide was titled “Good Hands or Boxing Gloves”, a play on Allstate’s advertising slogan “You’re in good hands” 119 M. In 2010 an inspection by regulators from 45 states found that Allstate had “tuned” Colossus unevenly in different regions. The company paid $10m and undertook to tell customers that Colossus might be used to assess their claim 120 M.

In 2023 ProPublica and The Capitol Forum described the PXDX system at the health insurer Cigna. If the diagnosis on a claim did not match a list approved for the procedure, medical directors rejected such claims in batches without opening the medical records 121 M. One medical director rejected about 121,000 claims in the first two months of 2022 122 M. “We literally click and submit. It takes all of 10 seconds to do 50 at a time,” a former Cigna doctor said 123 M. As early as 2014 an internal Cigna presentation calculated that a single new item on the list would produce more than 17,800 denials and about $2.4m in savings a year 124 M. Customers have filed two class actions, which are continuing 125 H. Cigna says PXDX checks claims after treatment and does not deny care 121 M.

UnitedHealth, one of America’s largest health insurers, bought the firm naviHealth in 2020 through its subsidiary Optum 126 M. Its algorithm, nH Predict, compares a patient with a database of about 6m others and predicts how many days of rehabilitation he or she needs 127 M. According to documents obtained by STAT, naviHealth’s managers set staff a target: the length of rehabilitation should deviate from the algorithm’s forecast by no more than 1% 128 M. After UnitedHealthcare began using naviHealth’s tools, the share of initial denials in prior-authorisation requests for skilled nursing facility (SNF) care rose from 1.4% (3,016 requests) in 2019 to 12.6% (34,359) in 2022, and the number of prior-authorisation requests for home care rose from 19,283 to 356,606. This is according to the majority staff report of the Senate Permanent Subcommittee on Investigations (PSI), as set out by LeadingAge 129,130 M.

In November 2023 the families of deceased patients filed a class action. They allege that 90% of the contested denials made with the participation of nH Predict are overturned; at the same time, according to the lawsuit, only about 0.2% of policyholders appeal 131 L. In February 2025 the court allowed the breach-of-contract claims to proceed and raised the question of whether UnitedHealth had kept its promise that decisions are made by doctors and clinical staff 132 H. In March 2026 the court ordered the company to hand over documents going back to 2017, including calculations of the savings from buying naviHealth 133 M. UnitedHealth maintains that doctors, not the algorithm, made the decisions 134 H.

Cigna also owns EviCore, which reviews prior-authorisation requests for insurers that together serve about 100m people. According to ProPublica, EviCore’s algorithm can approve a request by itself but cannot deny one: it passes doubtful requests to doctors, who then deny them. According to former employees, the threshold at which a request goes to manual review could be moved 135 M. EviCore’s salespeople boasted to clients of a 15% rise in denials and promised a return of $3 for every dollar spent 136 M. In Arkansas, where EviCore publishes its data, it has denied in full or in part almost 20% of requests since 2021 137 M.

According to the PSI majority staff report, CVS said it saved more than $660m in 2018 by denying prior authorisation for hospital admissions 141 M. In early 2025 84% of the 93 large health insurers surveyed by the NAIC used AI in some capacity; the survey does not separate decisions on payments from prior authorisations 142 H. Bain consultants estimate that at full potential generative AI could cut claims-handling costs in property and casualty (P&C) insurance by 20–25% and “leakage” by 30–50%; they put the total benefit to insurers and customers worldwide at more than $100bn 143 M. Bain defines “leakage” as “the difference between what is paid vs. what is owed per the contract” 144 M. In our view, the line between overpayment and underpayment is in practice set by whoever tunes the model.

Regulation lags behind. Only in February 2024 did the federal agency CMS explain that Medicare Advantage insurers may use algorithms but must make coverage decisions on the circumstances of the individual patient, not only on large datasets 145 M. In 2026 at least seven states passed laws that restrict AI in decisions on medical coverage 146 M. In the European Union, obligations for some high-risk AI systems (Annex III of the EU AI Act) were postponed from 2 August 2026 to 2 December 2027 147 M. In insurance, such systems include risk assessment and pricing in life and health insurance; claims handling is not on that list L.

08 · DisastersAfter the hurricane: rewritten engineers’ reports and adjusters’ estimates

After Hurricane Katrina in 2005, wind damage was to be paid by private insurers and flood damage by the federal flood insurance programme. The sisters Cori and Kerri Rigsby, who adjusted claims for State Farm, filed a lawsuit alleging that the company shifted wind losses onto the government programme 148 H. In 2015 the US Fifth Circuit Court of Appeals noted that in the case “there was evidence that adjusters were effectively told to presume flood damage instead of wind damage” 149 H. A note on the report of an engineer who attributed the damage to wind read: “Put in Wind [homeowner’s policy] file—DO NOT Pay Bill DO NOT discuss” 150 H. A jury found State Farm liable on one representative claim; with trebling and a penalty the total came to $758,250 151 H. In 2016 the US Supreme Court rejected State Farm’s demand to dismiss the case over a breach of the lawsuit’s seal; the Court did not examine the allegations on the merits 152 H. Back in 2007 State Farm, under a settlement with Mississippi’s attorney general, agreed to review claims with new engineers and adjusters and set aside at least $50m for it 153 M.

After Hurricane Sandy (2012) holders of federal flood insurance policies filed 144,540 claims 155 H. In 2014 a federal judge found “potentially fraudulent” an engineer’s report on which the insurer Wright National had based its decision on a house in Long Beach: the report had been rewritten to attribute the damage to long-standing wear rather than the storm 156 M. Lawyers for the homeowners said they had found more than 500 rewritten reports 157 L. In May 2015 FEMA invited about 142,000 policyholders to have their claims reviewed. 19,464 applied; as of 1 December 2017, 14,751 claims had been closed with an additional payment, and FEMA had offered owners a further $270m 158 H.

After Hurricane Ian (2022) adjusters told Florida lawmakers that insurers had cut their loss estimates by as much as two-thirds and left their names on the amended reports 159 M. In 2024 the programme 60 Minutes reported, citing an adjuster, that the estimate for one house had been cut from $231,368 to $15,469 160 M. The chief executive of Heritage Insurance called the adjusters’ accusation that it had deceived customers “flat wrong” 161 M. On 9 October 2024 Florida’s chief financial officer used an emergency rule to ban changes to a field adjuster’s estimate without an explanation of what was changed and by whom 162 M.

In Louisiana in 2022 almost half of homeowners’ insurance claims were paid after more than 60 days, against 22% in the rest of the US 163 M. The Louisiana legislative auditor found that 67% of complaints about adjusters after Hurricanes Laura and Ida were resolved in favour of policyholders; the Department of Insurance’s own analysis gave slightly more than half 164 M. The regulator proposed fines on five insurers totalling $764,750; three of them later went bankrupt, and in all, according to an investigation by The Advocate (January 2024), 12 insurers went bankrupt in the state 165,166 M.

After the Los Angeles fires in January 2025, the California Department of Insurance examined a sample of 220 State Farm General files and counted 398 alleged violations 167 H; according to press reports, there were violations in 114 of those 220 files, or 52% of the sample 168 M. According to the examination, in six files of the sample the company charged the cost of hygienist and environmental testing for smoke damage against the policy limit, and in at least 12 files it refused to pay for such testing, citing the policy’s “Right to Inspect” provision. State Farm later acknowledged that the provision had been applied improperly 169 H. Of the smoke-remediation contractors the regulator wrote: “These vendors serve the Company’s interests in managing claims for cost control and efficiency” 170 H.

The costliest insurance event in Australia’s recorded history was the floods of February–March 2022: as of May 2023, more than 241,000 claims and A$5.87bn in insured losses 171 H. The regulator ASIC analysed 218,256 home insurance claims lodged in January–March 2022 with six insurers (63% of the market); 43% of them were claims for bad weather, mostly these floods. As of September 2022, 76% of claims had been accepted in full or in part, 8% declined, 12% withdrawn by the customers themselves and 4% not yet determined. 11% of claims led to a complaint 172 H. ASIC noted that insurers did not have enough staff to handle disputes 173 H. The chair of the parliamentary committee, Daniel Mulino, explained the title of the 2024 report, “Flood failure to future fairness”, as “the collective failure by insurers to meet their obligations to policyholders” 174 M.

A separate kind of denial is declining to renew a policy before anything has happened. In 2024 US home insurers on their own initiative did not renew 2,019,799 policies; since 2018 the rate of such non-renewals has risen by 96–216% depending on the geographical zone, and the real price of a policy by 18–43% 175 H. “Insurance companies are choosing not to renew homeowners policies more frequently,” writes the NAIC, the association of state insurance regulators 176 H. According to The Wall Street Journal, insurers use aerial photographs to pick houses whose policies they will not renew. Separately, State Farm planned not to renew policies on about 30,000 homes and 42,000 commercial properties in California, citing inflation and catastrophe risk 177 M. Such refusals do not appear in the statistics on denied claims.

09 · Motor and lifeUnderpaid total-loss cars, outdated definitions and “net termination ratios”

Total-loss cars. In motor insurance almost all claims are accepted — in Britain 98.7% 12 H. The dispute is over the amount. In 2024 the British regulator, the FCA, reviewed 12 insurers that together hold about 70% of the market. Most automatically deducted 20% from the payout if the car had previously been written off; some paid less than the fair market price and raised the first offer only after a complaint 178 H. Half of the 12 companies reviewed deducted the remaining unpaid premium instalments from the payout 179 H. “These low average settlement figures indicate that some customers' claims may have been handled unfairly,” the regulator wrote 180 H. In 2025 about 270,000 drivers are receiving roughly £200m in compensation for automatic deductions from payouts on stolen and written-off cars; by December 2025, £129m had been paid to almost 150,000 people 181 H. In the FCA's lists of fines for 2023–2026 we found no fine on an insurer for claims handling 182 M.

In the US, Progressive settled a class action in New York for $18.3m without admitting wrongdoing: the plaintiffs alleged that the company did not include sales tax in the value of total-loss cars; the class includes about 31,000 customers 183 H. In Arkansas a class action concerned “typical negotiation adjustments” with which State Farm, the plaintiffs allege, reduced its valuations of total-loss cars; in March 2026 a court gave preliminary approval to a $15.58m settlement 184 M. Progressive's net income for 2025 was $11.3bn 185 H.

Disability. In the early 2000s juries in various US states awarded customers of Unum group insurers $36.7m, $7.7m, $31.7m and $11.6m in cases over terminated disability benefits 186 M. A multistate examination by regulators, as recounted by plaintiffs' lawyer T. Coleman, found that Unum relied on its own doctors, ignored the opinions of the doctors treating the claimant and denied claims for lack of “objective” evidence where the policy did not require it 186,187 M. According to a quotation from the California settlement as set out by Coleman, Unum targeted certain types of claims for “resolution” (in his explanation, denial or termination of benefits) in the interest of improving “net termination ratios” 188 L. In 2004 Unum agreed to reassess up to 215,000 denied and closed claims and paid a $15m fine 189 H. Of the reassessed claims, the decision was reversed in full or in part in 41.7% of cases; claimants received an additional $676.2m in benefits 190 M.

Life insurance in Australia. In 2018, at Australia's banking royal commission, which investigated misconduct in the financial sector, at least 13 insurers admitted misconduct of various kinds (in sales, advice and claims) over the previous decade 191 H. CommInsure for 11 years did not update the definition of heart attack in its trauma (critical illness) policies and for at least two years knew that it did not match medical practice 192 M. ClearView admitted that it may have breached the ban on selling insurance by cold calling up to 303,000 times in just over three years 193 H. QBE estimated compensation for more than 28,000 customers at A$15.9m, including for add-on policies sold with cars that were “of little or no use” 194 H. According to ASIC's 2016 review (5,438 disputes in 2013–2015), the main causes of disputes with life insurers are evidence (25%) and delays (22%) 195 H.

10 · Courts and finesWhy denying pays: few appeal, fines are small, verdicts get cut

When a dispute reaches an independent body, the customer is often right. In 2024/25 the Financial Ombudsman Service in Britain upheld 38% of 45,606 complaints about insurers — more than the average across all financial products (34%) 198 H. In 2021/22 the share was 28% 199 H. In Germany 50.2% of the complaints about insurers (excluding life insurance) accepted for consideration in 2025 ended in full or partial success for the customer, mostly through voluntary correction by the insurer 200 H. In a third of all cases the insurer changed its own decision as soon as the ombudsman stepped in 201 H. In India the insurance ombudsman ruled in the complainant's favour in 53.5% of the complaints received about claim repudiation (18,809 of 35,186 in fiscal year 2025–26; all types of insurance) 63 M.

Big verdicts get cut. In State Farm v. Campbell a jury in Utah awarded $145m in punitive damages for refusing to settle a claim within the policy limits. The state court described evidence of “a national scheme to meet corporate fiscal goals by capping payouts on claims company wide” 202 H. The US Supreme Court in 2003 held the sum excessive 203 H. In California the insurance commissioner fined PacifiCare (UnitedHealth) about $173m for more than 900,000 violations; $82m of that was set aside by a lower court, and ≈$91m had still not been finally confirmed as of 2019 204 M. A jury in Las Vegas in 2022 awarded $200m to the family of a lung-cancer patient whom Sierra Health had refused proton therapy; that week the reviewing physician spent an average of 12 minutes per case. The Nevada Supreme Court upheld the verdict 205 M.

For most people insured through an employer under ERISA plans, such verdicts are, as a rule, not available. In denial disputes ERISA allows neither punitive damages nor a jury trial 206 M, and the Supreme Court confirmed in 1987 and 2004 that claims under state law do not apply here 207,208 H. So for an unfounded denial the insurer mostly risks only having to pay the sum that was due.

Regulators' fines are small. In August 2025 Florida fined three insurers over claims handling after Hurricanes Ian and Idalia: American Mobile ($400,000) for improper disclosure on partial payment, Monarch ($325,000) and Tower Hill ($250,000, for Ian only) for neither paying nor denying within 90 days 209 M. According to the ASIC chair, the first civil penalty against an insurer for breaching the duty of utmost good faith in claims handling — A$2m for Hollard in October 2026 (the company admitted the breach) — equals roughly 0.2% of the company's annual turnover 210,211 M. In the lists of fines by Britain's FCA for 2023–2026 (by sector tag) we found no fine on an insurer for claims handling; in 2024 there were only two fines in the sector, both on individuals 182 M. According to ASIC, as of June 2023 the largest court penalty against an insurer for breaching financial consumer-protection laws is A$40m for Insurance Australia Limited (the NRMA Insurance brand), for false assurances about discounts to more than 600,000 customers (a pricing case, not a claims case) 212 H. By ASIC's estimate, 11 Australian insurers (68% of the market) did not deliver the discounts they had promised, so more than 5.6 million customers overpaid more than A$815m (including interest, fees and taxes) 213 H.

Insurers also carry political weight. In 2025 the US insurance industry spent $172m on federal lobbying — fourth among all industries 217 M. Since 1945 insurance in the US has been exempt from federal antitrust law where the states regulate it 218 M; for health insurers the exemption was narrowed only in 2021 219 H.

Surveys show how people see this. In July 2025, 73% of American adults called denials and delays by health insurers a serious problem 220 H. In January 2026, 47% of insured adults said that in the past two years an insurer had denied, delayed or changed their treatment or medication 221 H. In Britain 78% of adults associate the insurance industry with “loopholes” and 77% with putting profit first 222 M.

11 · Ukraine37 kopiykas per hryvnia: the payout ratio, MTPL after the reform and war outside cover

In 2018–2021 the entire Ukrainian insurance market (including life insurance) paid customers 24–27% of gross premiums; on net figures after reinsurance, 32–36% 24 M. The National Bank of Ukraine (NBU), in its 2020 White Paper, described a range of 24–27% and noted that premiums were growing while the payout ratio was not 223 M. According to UAINSUR, in the first half of 2026 the payout ratio was 45.0%, against 36.6% a year earlier (37.0% for the whole of 2025) 25 M. This is a different basis from the 2018–2022 series, so the two cannot be compared with each other.

The market is small and profitable. According to Natalia Hudyma, vice-president of the League of Insurance Organisations of Ukraine, “Ukrainian insurers collected about UAH 72bn in insurance premiums in 2025 and paid out UAH 25bn” (life insurance included) 224 M. For 2025, by the estimate of VKURSI Market BI (state registers), insurers earned UAH 7.5bn in profit, against UAH 4.4bn a year earlier 225 M. According to UAINSUR, in the first half of 2026 the net financial result rose by 70%, to UAH 3.97bn 226 M. Among the 12 largest insurers the payout ratio in 2025 ranged from 31.0% (Oranta) to 64.0% (Euroins); at the two largest insurers by premiums, TAS and ARX, it was 38.2% and 35.0% (across the whole portfolio, not only MTPL) 227 M. Five of the 47 risk insurers paid out less than 0.2% of premiums. Busin, for example, collected UAH 356.2m and paid out UAH 0.22m; the NBU table gives no reasons 228 M.

MTPL after the reform

From 1 January 2025 the limits on compulsory motor third-party liability insurance (MTPL) rose from UAH 160,000 to UAH 250,000 for damage to property and from UAH 320,000 to UAH 500,000 for damage to life, per victim 27 M. In 2025 premiums written rose by 114%, to UAH 23.6bn, and compensation accrued on settled claims by 41%, to UAH 6.8bn. The ratio of compensation accrued to premiums written fell from 43.6% to 28.8% 229 H. This is partly an arithmetic effect: premiums were booked at once in the year prices jumped, while compensation lags in time; on an earned-premium basis our rough estimate is about 39% L. The average price of a policy rose by 132%, the average compensation by 33% 26 M. The number of accidents per policy rose, so compensation per policy grew by about 53% — two and a half times more slowly than the price L. In January–September 2026 compensation began to catch up with premiums: +84%, with the ratio to premiums at 46% 230 H.

Speed also depends on the procedure. In the first quarter of 2026 direct settlement through one's own insurer took 32 days on average, and the standard procedure 57 days 34 M.

War outside cover

According to the legal outlet dtkt.ua (2022), if property is not insured separately against war risks, damage from hostilities “is not recognised as an insured event” 231 M. Direct war damage at the end of 2024 was $176bn by the World Bank's estimate 232 H. The state war-risk insurance programme for 2026 has UAH 1bn; Danylo Hetmantsev, chair of the tax committee of the Verkhovna Rada, called this “a drop in the ocean” and explained that the intake of applications will stop as soon as they add up to UAH 1bn — for example, 100 applications of UAH 10m each 233 M.

Complaints and licences

In the second quarter of 2022 insurers accounted for 52% of complaints to the NBU showing signs of consumer-rights violations; one in nine written complaints concerned non-payment of insurance compensation 234 M. In 2025 the National Bank received 1,473 written complaints about insurers 235 M. The market is shrinking: the NBU's reporting covered 62 insurers on 30 June 2025, 57 on 30 June 2026 and 54 on 11 August 2026 (44 risk insurers and 10 life insurers) 236 M. From 24 July 2026 the NBU revoked the licences of ASKO DS, which did not submit a recovery plan, and of Peremoha, for risky activity that threatened the interests of policyholders 237 M.

For comparison, Poland. In 2024 Polish insurers settled 10m claims, received 270,000 complaints (not only about denials; 4–5 per 1,000 contracts), and about 248,000 cases against them were before the courts 238 H. There, the quality of claims handling makes up 25% of an insurer's supervisory score, which determines how much in dividends it may pay 239 H.

12 · TimelineEighty years of denials: from the 1945 law to the first Australian penalty of 2026

Below are 63 events drawn from studies, court rulings and regulators' reports. The filter above shows separate strands. Each event has a link to its source and a confidence label.

13 · What to doIf you were denied: six steps the data suggest

The main conclusion of this briefing is simple: every unappealed denial leaves money with the insurer, and only a minority appeal. So the most effective thing a customer can do is appeal.

1. Demand a written denial that gives a specific reason and the clause of the contract. In the US, 36% of the denial reasons recorded on the ACA marketplace in 2024 (in-network) were marked “other or unspecified” 65 H. From 2026 a federal CMS rule requires some insurers to state a specific reason for a prior-authorisation denial 240 H. Without a written reason a denial is hard to appeal.
2. Appeal. In Medicare Advantage in 2024, 80.7% of appealed prior-authorisation denials were overturned in full or in part, but only 11.5% of denials were appealed 6 H. In skilled nursing facility (SNF) care (June 2024, the 19 largest companies) 95% of appealed denials were overturned, but only 18% were appealed 14 H. In Britain the ombudsman upheld 38% of complaints about insurers, in Germany 50.2% of accepted complaints ended in full or partial success for the customer, and in India decisions in the complainant's favour made up 53.5% of the complaints received about claim repudiation 198,200,63 M.
3. Do not accept the first offer. The British regulator found that some insurers at first offered low payouts for total-loss cars and raised them only after a complaint 178 H.
4. Gather your own evidence. Your own doctor's opinion, an independent assessment of the loss, photos and correspondence. As recounted by plaintiffs' lawyer T. Coleman, regulators reproached Unum for ignoring the opinions of the doctors treating the claimant 186,187 M. Claims adjusters after Hurricane Ian testified that their estimates were rewritten 159 M.
5. Track the deadlines. A delay is also a violation. After the Los Angeles fires, the most frequent of the 398 probable violations that the California Department of Insurance found in a sample of 220 State Farm files was that the company did not tell customers in writing that it needed additional time (79 cases) 167 H.
6. Complain to the regulator. In Ukraine insurers are supervised by the National Bank; in 2025 it received 1,473 written complaints about insurance companies 235 M. A complaint to the regulator and a lawsuit are a long road, but without an appeal the denial will most likely stand.

14 · QuestionsFrequently asked questions

15 · NextRead next

16 · SourcesSources and method

The number in the text leads to the source. Confidence labels: H — primary source (regulator report, court ruling, company accounts, official statistics), M — secondary source (a publication or analytical organisation retelling the primary one) or cross-checked against several, L — an estimate, a plaintiff's allegation, a single unconfirmed source or our calculation. The data were collected and cross-checked by AI research agents with search; each strand was verified separately by a sceptic agent that tried to disprove the key figures. Assertions from lawsuits that the courts have not yet decided are labelled as plaintiffs' allegations. Amounts are in each country's currency, without conversion.

1. US House Energy and Commerce Committee — hearing on policy rescissions (16 June 2009). govinfo.gov H
2. US House hearing transcript, 16 June 2009. govinfo.gov M
3. ACA — ban on rescission of health policies. californiahealthline.org M
4. KFF — denials and appeals in ACA marketplace plans in 2024 (24 March 2026). kff.org H
5. KFF — denials and appeals in ACA marketplace plans: appeals. kff.org H
6. KFF — prior authorisation in Medicare Advantage in 2024 (28 January 2026): appeals. kff.org H
7. FCA — General insurance value measures data 2025 (21 July 2026): share of claims accepted. fca.org.uk H
8. Insurance Information Institute — Facts + Statistics: Industry overview (NAIC and S&P Global data). iii.org H
9. FCA — General insurance value measures data 2025 (21 July 2026): claims paid as a share of premiums. fca.org.uk H
10. FCA — add-on policies market study (11 March 2014). fca.org.uk H
11. FCA — letter to GAP insurance manufacturers (September 2023). fca.org.uk H
12. FCA — General insurance value measures data 2023 and 2025. fca.org.uk H
13. APRA — Life insurance claims and disputes data, June 2026 (16 October 2026). apra.gov.au H
14. HHS OIG — OEI-09-24-00331: skilled nursing facility (SNF) care in Medicare Advantage (June 2026). oig.hhs.gov H
15. HHS OIG — OEI-09-24-00331: skilled nursing facility (SNF) care in Medicare Advantage (June 2026): UnitedHealth and naviHealth. oig.hhs.gov H
16. ProPublica, The Capitol Forum — How Cigna Saves Millions by Having Its Doctors Reject Claims Without Reading Them (25 March 2023): Cigna’s internal estimate. propublica.org M
17. HHS OIG — OEI-09-24-00330: inpatient rehabilitation facilities and long-term care hospitals in Medicare Advantage (June 2026). oig.hhs.gov H
18. HHS OIG — OEI-09-24-00330: inpatient rehabilitation facilities and long-term care hospitals in Medicare Advantage (June 2026): average cost of a stay. oig.hhs.gov M
19. APRA — Life insurance claims and disputes data, June 2026 (16 October 2026): average sum insured. apra.gov.au H
20. ASISA — Annual Claims Payout Statistics 2025: average amounts. asisa.org.za H
21. Coalition Against Insurance Fraud — benchmarking study of special investigation units, as summarised by Claims Journal (2020). claimsjournal.com M
22. NY DFS — Investigating and Combating Health Insurance Fraud, 2024 report (15 March 2025). dfs.ny.gov H
23. Coalition Against Insurance Fraud — The Impact of Insurance Fraud on the U.S. Economy (26 August 2022): origin of the 10% estimate. insurancefraud.org H
24. Bulletin of Uman National University of Horticulture, vol. 103.2, art. 18 — insurance market indicators based on National Bank of Ukraine data. journal.udau.edu.ua M
25. UAINSUR — main trends in Ukraine’s insurance market in the first half of 2026. uainsur.com M
26. MTSBU — MTPL statistics, as summarised by Minfin and finance.ua: average policy price and average compensation. minfin.com.ua M
27. Minfin — compulsory motor third-party liability insurance (MTPL) limits from 1 January 2025 (19 August 2025). minfin.com.ua M
28. Berkshire Hathaway — Warren Buffett’s 2005 letter to shareholders. berkshirehathaway.com H
29. Berkshire Hathaway — letters to shareholders, 1970–2025. berkshirehathaway.com H
30. Insurance Information Institute — Facts + Statistics: Industry overview (NAIC and S&P Global data). iii.org H
31. Texas Insurance Code, § 542.060. texas.public.law M
32. BCG — The 2026 Insurance Value Creators Report (April 2026). web-assets.bcg.com H
33. BCG — The 2026 Insurance Value Creators Report: section on AI. web-assets.bcg.com H
34. MTSBU — the first quarter of 2026, as summarised by finance.ua (28 April 2026). news.finance.ua M
35. Company filings with the SEC (10-K, 10-Q), researcher’s calculation: UnitedHealth, Elevance, Cigna, Humana, Centene. data.sec.gov H
36. UnitedHealth Group — SEC filings, 2015–2025. data.sec.gov H
37. CMS and KFF — premium rebates under the minimum medical loss ratio (MLR) rule. kff.org H
38. UnitedHealth Group — 2025 financial reporting. sec.gov H
39. Company filings with the SEC (10-K, 10-Q), researcher’s calculation: first half of 2026. data.sec.gov H
40. UnitedHealth Group — proxy statement (DEF 14A), 21 April 2026. sec.gov M
41. US Bureau of Labor Statistics — price indices CUUR0000SETE and CUUR0000SA0. api.bls.gov M
42. Progressive Corp. — SEC filings (10-K). data.sec.gov H
43. Allstate Corp. — SEC filings (10-K). data.sec.gov H
44. FCA — PS21/5: pricing in home and motor insurance (May 2021). fca.org.uk H
45. FCA — MS18/1.3: final report of the pricing study (2020). fca.org.uk H
46. FCA — ban on raising the renewal price (from 1 January 2022). fca.org.uk H
47. FCA — General insurance value measures (since 2022). fca.org.uk H
48. ASIC — review of insurance sold in car dealerships (September 2016), as summarised by Consumer Action Law Centre. consumeraction.org.au M
49. Consumers Union, Center for Economic Justice — Credit Insurance: The $2 Billion A Year Rip-Off (1999). advocacy.consumerreports.org M
50. Consumers Union, Center for Economic Justice — Credit Insurance (1999): NAIC standards. advocacy.consumerreports.org M
51. Indiana Department of Insurance — Bulletin 246 (26 October 2018). secure.iot.in.gov H
52. FCA — PPI compensation. fca.org.uk H
53. FCA — PPI complaints deadline (29 August 2019). fca.org.uk H
54. FCA — General insurance value measures data 2025 (21 July 2026): complaints per 100 claims. fca.org.uk H
55. FCA — General insurance value measures data 2025 (21 July 2026): caveat on the home insurance data. fca.org.uk H
56. Insurance Business UK — Insurance DataLab analysis for the House of Lords (7 August 2026). insurancebusinessmag.com M
57. APRA — Life insurance claims and disputes data, June 2026 (16 October 2026): claims received. apra.gov.au H
58. APRA — Life insurance claims and disputes data, June 2026 (16 October 2026): reasons for denial. apra.gov.au H
59. ASIC — Report 498: Life insurance claims, an industry review (12 October 2016). asic.gov.au H
60. ASISA — Annual Claims Payout Statistics 2025. asisa.org.za H
61. ASISA — Annual Claims Payout Statistics 2025: credit life insurance. asisa.org.za H
62. IRDAI annual report for fiscal year 2024–25, as summarised by India Med Today (6 January 2026). insurancebusinessmag.com M
63. Cafemutual — complaints to India’s insurance ombudsman (government reply in the Lok Sabha, 29 August 2026). cafemutual.com M
64. KFF — denials and appeals in ACA marketplace plans: out-of-network claims. kff.org H
65. KFF — denials and appeals in ACA marketplace plans: reasons for denial. kff.org H
66. KFF — Consumer Survey Highlights Problems with Denied Health Insurance Claims (2023). kff.org H
67. KFF — denials and appeals in ACA marketplace plans in 2023. kff.org H
68. KFF — denials and appeals in ACA marketplace plans: spread across insurers, 2024. kff.org H
69. KFF — Prior Authorization Metrics Provide New Insights into Insurer Practices (13 August 2026). kff.org H
70. KFF — denials and appeals in ACA marketplace plans: outcomes of internal appeals. kff.org H
71. KFF — prior authorisation in Medicare Advantage in 2024 (28 January 2026). kff.org H
72. KFF — annual analyses of prior authorisation in Medicare Advantage, 2021–2024. kff.org M
73. HHS OIG — OEI-09-16-00410: denials and appeals in Medicare Advantage (25 September 2018). oig.hhs.gov H
74. HHS OIG — OEI-09-18-00260: Medicare Advantage denials that complied with Medicare rules (27 April 2022). oig.hhs.gov H
75. HHS OIG — OEI-09-24-00331: skilled nursing facility (SNF) care in Medicare Advantage (June 2026): waiting times. oig.hhs.gov H
76. HHS OIG — OEI-09-24-00331: skilled nursing facility (SNF) care in Medicare Advantage (June 2026): conclusion. oig.hhs.gov H
77. STAT — Denied by AI (13 March 2023). statnews.com M
78. Senate Permanent Subcommittee on Investigations (PSI) — majority staff report “Refusal of Recovery” (17 October 2024), as summarised by LeadingAge. leadingage.org M
79. Commonwealth Fund — How Health Insurance Coverage Denials Affect Americans (June 2026). commonwealthfund.org H
80. Commonwealth Fund — How Health Insurance Coverage Denials Affect Americans: claim denials. commonwealthfund.org H
81. Premier Inc. — survey of 516 hospitals, as summarised by STAT (1 May 2024). statnews.com M
82. KFF — prior authorisation in Medicare Advantage in 2024 (28 January 2026): by insurer. kff.org H
83. KFF — analysis of OIG reports on post-acute care (6 July 2026). kff.org H
84. HHS OIG — OEI-09-24-00330: inpatient rehabilitation facilities and long-term care hospitals in Medicare Advantage (June 2026) and OEI-09-24-00331: by insurer. oig.hhs.gov H
85. HHS OIG — OEI-09-24-00330: inpatient rehabilitation facilities and long-term care hospitals in Medicare Advantage (June 2026) and OEI-09-24-00331: for-profit and nonprofit plans. oig.hhs.gov H
86. HHS OIG — OEI-09-24-00331: skilled nursing facility (SNF) care in Medicare Advantage (June 2026): financial incentive. oig.hhs.gov H
87. HHS OIG — OEI-09-24-00330: inpatient rehabilitation facilities and long-term care hospitals in Medicare Advantage (June 2026) and OEI-09-24-00331: conclusion on incentives. oig.hhs.gov H
88. Premier Inc. — 2022 data on denials of expensive treatment. statnews.com M
89. Versicherungsombudsmann e.V. — annual report 2025: distribution of complaints by amount. versicherungsombudsmann.de H
90. Versicherungsombudsmann e.V. — annual report 2025: share of successful complaints. versicherungsombudsmann.de H
91. Versicherungsombudsmann e.V. — annual report 2025: the ombudsman’s powers. versicherungsombudsmann.de H
92. ABI, as summarised by Insurance Business UK and LBC; researcher’s calculation. insurancebusinessmag.com M
93. ABI — data on detected fraud for 2025, as summarised by Insurance Business UK (6 October 2026). insurancebusinessmag.com M
94. GDV, as summarised by cash-online.de (2 May 2024). cash-online.de M
95. GDV, as summarised by cash-online.de (2 May 2024). cash-online.de M
96. NY DFS — Investigating and Combating Health Insurance Fraud, 2025 report (March 2026). dfs.ny.gov H
97. NY DFS — reports for 2024 and 2025: “savings” and recoveries. dfs.ny.gov H
98. Insurance Fraud Bureau of Massachusetts, 2022 annual report, as summarised by Rutgers Center for Risk and Responsibility (November 2024). crr.rutgers.edu M
99. Richard Derrig, Journal of Risk and Insurance 69(3), 2002, as summarised by Rutgers Center for Risk and Responsibility. crr.rutgers.edu M
100. Coalition Against Insurance Fraud — benchmarking study of special investigation units, as summarised by Claims Journal (2020): automated alerts. claimsjournal.com M
101. Coalition Against Insurance Fraud, SAS — The State of Insurance Fraud Technology (2016). claimsjournal.com M
102. Coalition Against Insurance Fraud — benchmarking study of special investigation units, as summarised by Claims Journal (2020): budgets. insurancejournal.com M
103. Coalition Against Insurance Fraud — The Impact of Insurance Fraud on the U.S. Economy (26 August 2022). insurancefraud.org H
104. Coalition Against Insurance Fraud — The Impact of Insurance Fraud on the U.S. Economy (26 August 2022): methodology by line. insurancefraud.org H
105. Sharon Tennyson, Social Research 75(4), 2008, as summarised by Rutgers Center for Risk and Responsibility. crr.rutgers.edu M
106. Coalition Against Insurance Fraud — The Impact of Insurance Fraud on the U.S. Economy (26 August 2022): foreword. insurancefraud.org H
107. Coalition Against Insurance Fraud — The Impact of Insurance Fraud on the U.S. Economy (26 August 2022): health insurers’ data. insurancefraud.org H
108. ABI, as summarised by Insurance Business UK and LBC; researcher’s calculation. insurancebusinessmag.com M
109. ABI — 2025 data, as summarised by Insurance Business UK. insurancebusinessmag.com M
110. Victorian Ombudsman — WorkSafe 2: follow-up investigation (3 December 2019). ombudsman.vic.gov.au H
111. Victorian Ombudsman — the 2016 investigation, as recapped in the WorkSafe 2 report. ombudsman.vic.gov.au H
112. MinterEllison — review of the hearings of Australia’s banking royal commission, round 6 (2018). minterellison.com M
113. US House Energy and Commerce Committee — hearing materials (2009). govinfo.gov M
114. Consumer Federation of America — Consumer Alert on Colossus (December 2010). consumerfed.org M
115. Consumer Federation of America — study of Allstate and Colossus (18 July 2007), quoting CSC marketing materials. consumerfed.org M
116. Consumer Federation of America — “Low Ball” (June 2012), as summarised by InsuranceNewsNet. insurancenewsnet.com M
117. The Aldridge v. Allstate case and the McKinsey materials, as summarised by the press. amp.insurancejournal.com M
118. Associated Press — Aldridge v. Allstate case (2007–2008). amp.insurancejournal.com M
119. Associated Press — McKinsey documents for Allstate (2008). amp.insurancejournal.com M
120. Insurance Journal — Allstate’s settlement with regulators in 45 states (19 October 2010). amp.insurancejournal.com M
121. ProPublica, The Capitol Forum — How Cigna Saves Millions by Having Its Doctors Reject Claims Without Reading Them (25 March 2023). propublica.org M
122. ProPublica, The Capitol Forum — How Cigna Saves Millions by Having Its Doctors Reject Claims Without Reading Them (25 March 2023): Cigna spreadsheet. propublica.org M
123. ProPublica, The Capitol Forum — How Cigna Saves Millions by Having Its Doctors Reject Claims Without Reading Them (25 March 2023): testimony of a former doctor. propublica.org M
124. ProPublica, The Capitol Forum — How Cigna Saves Millions by Having Its Doctors Reject Claims Without Reading Them (25 March 2023): Cigna’s 2014 presentation. propublica.org M
125. Kisting-Leung v. Cigna (E.D. Cal.) and Snyder v. The Cigna Group (D. Conn.): case filings. litigationtracker.law.georgetown.edu H
126. LeadingAge — analysis of the Senate report on Medicare Advantage. leadingage.org M
127. STAT — Denied by AI (2023), as summarised by AHCJ. healthjournalism.org M
128. STAT — Denied by AI (November 2023), as summarised by AHCJ. statnews.com M
129. Senate Permanent Subcommittee on Investigations (PSI) — majority staff report “Refusal of Recovery” (17 October 2024), as summarised by LeadingAge: skilled nursing facility (SNF) care. leadingage.org M
130. Senate Permanent Subcommittee on Investigations (PSI) — majority staff report “Refusal of Recovery” (17 October 2024), as summarised by LeadingAge: home health care. leadingage.org M
131. Estate of Lokken v. UnitedHealth Group, complaint (D. Minn., 14 November 2023). statnews.com L
132. Estate of Lokken v. UnitedHealth Group — court order of 13 February 2025. litigationtracker.law.georgetown.edu H
133. Estate of Lokken v. UnitedHealth Group — discovery order (9 March 2026). afslaw.com M
134. Estate of Lokken v. UnitedHealth Group — joint report by the parties (12 September 2025). litigationtracker.law.georgetown.edu H
135. ProPublica, The Capitol Forum — investigation into EviCore (23 October 2024). propublica.org M
136. ProPublica, The Capitol Forum — investigation into EviCore (23 October 2024): sales. propublica.org M
137. ProPublica, The Capitol Forum — investigation into EviCore (23 October 2024): Arkansas data. propublica.org M
138. ProPublica, The Capitol Forum — investigation into EviCore (23 October 2024): testimony of former employees. propublica.org M
139. Senate Permanent Subcommittee on Investigations (PSI) — majority staff report “Refusal of Recovery” (17 October 2024), as summarised by LeadingAge: naviHealth instruction. leadingage.org M
140. Senate Permanent Subcommittee on Investigations (PSI) — majority staff report “Refusal of Recovery” (17 October 2024), as summarised by LeadingAge: CVS strategy. leadingage.org M
141. Senate Permanent Subcommittee on Investigations (PSI) — majority staff report “Refusal of Recovery” (17 October 2024), as summarised by LeadingAge: CVS savings. plantemoran.com M
142. NAIC — Health AI/ML Survey Report (20 May 2025). content.naic.org H
143. Bain & Company — $100 Billion Opportunity for Generative AI in P&C Claims Handling (31 October 2024). bain.com M
144. Bain & Company — definition of “leakage” in claims handling. bain.com M
145. CMS — answers to questions about algorithms in Medicare Advantage (6 February 2024). reedsmith.com M
146. Rise Health — AI health insurance laws: 7 states set new rules (12 August 2026). risehealth.org M
147. Regulation (EU) 2026/1744 (Digital Omnibus on AI), as summarised by Cloud Security Alliance. labs.cloudsecurityalliance.org M
148. US ex rel. Rigsby v. State Farm — complaint under the False Claims Act (2006). caselaw.findlaw.com H
149. US Fifth Circuit Court of Appeals, US ex rel. Rigsby v. State Farm, No. 14-60160 (13 July 2015). caselaw.findlaw.com H
150. US Fifth Circuit Court of Appeals, US ex rel. Rigsby v. State Farm (13 July 2015): background to the case. caselaw.findlaw.com H
151. US Fifth Circuit Court of Appeals, US ex rel. Rigsby v. State Farm (13 July 2015): verdict. caselaw.findlaw.com H
152. US Supreme Court, State Farm v. U.S. ex rel. Rigsby, No. 15-513 (6 December 2016). law.cornell.edu H
153. Mississippi attorney-general’s settlement with State Farm (24 January 2007), as summarised by the press. insurancejournal.com M
154. Coverager — testimony of claims adjusters in the Florida legislature (2024). coverager.com M
155. DHS OIG — OIG-18-38: review of claims after Hurricane Sandy (24 January 2018). oversight.gov H
156. Carrier Management — the Wright National case over Hurricane Sandy (December 2014). carriermanagement.com M
157. ABA Journal, based on The New York Times (17 February 2015). abajournal.com L
158. DHS OIG — OIG-18-38: results of FEMA’s review of claims. oversight.gov H
159. Claims Journal — Florida House of Representatives hearing (17 May 2023). amp.claimsjournal.com M
160. CBS 60 Minutes (29 September 2024), as summarised by Coverager. coverager.com M
161. Heritage Insurance — response to the 60 Minutes report, as summarised by Coverager. coverager.com M
162. Florida’s Chief Financial Officer — emergency rule 69BER24-4 (9 October 2024). insurancejournal.com M
163. The Advocate (Baton Rouge) — “Beaten down” (15 January 2024). insurancenewsnet.com M
164. Louisiana Legislative Auditor, as summarised by The Advocate (2024). insurancenewsnet.com M
165. Louisiana Department of Insurance — proposed fines (14 April 2022). propertyinsurancecoveragelaw.com M
166. The Advocate — insurer bankruptcies in Louisiana (2024). insurancenewsnet.com M
167. California Department of Insurance — examination report on State Farm General Insurance Company (2026). consumerwatchdog.org H
168. California Department of Insurance — examination report on State Farm General Insurance Company (2026): share of files with violations. insurancenewsnet.com M
169. California Department of Insurance — examination report on State Farm General Insurance Company (2026): summary. consumerwatchdog.org H
170. California Department of Insurance — examination report on State Farm General Insurance Company (2026): general finding 29. consumerwatchdog.org H
171. ASIC — Report 768: scale of the 2022 floods. download.asic.gov.au H
172. ASIC — Report 768: Navigating the storm (16 August 2023). download.asic.gov.au H
173. ASIC — Report 768: findings. download.asic.gov.au H
174. Australia’s House of Representatives Standing Committee on Economics — Flood failure to future fairness (18 October 2024). choice.com.au M
175. NAIC — Homeowners Property Insurance Market Dynamics (31 July 2026). content.naic.org H
176. NAIC — Homeowners Property Insurance Market Dynamics: key findings. content.naic.org H
177. The Wall Street Journal (2024), as reprinted by AI Incident Database. incidentdatabase.ai M
178. FCA — Findings of multi-firm review into insurers’ valuation of vehicles (27 March 2024). fca.org.uk H
179. FCA — review of total-loss vehicle valuations: monthly instalments. fca.org.uk H
180. FCA — review of total-loss vehicle valuations: findings. fca.org.uk H
181. FCA — Over 270,000 motorists receive motor insurance compensation (2025). fca.org.uk H
182. FCA — lists of fines for 2023–2026. fca.org.uk H
183. Progressive’s settlement in the sales-tax class action (New York, 2024). amp.insurancejournal.com H
184. Chadwick v. State Farm — preliminary approval of settlement (27 March 2026). bodyshopbusiness.com M
185. Progressive Corp. — annual report for 2025. sec.gov H
186. T. Coleman, “Unum revisited”, Plaintiff magazine (September 2016). plaintiffmagazine.com M
187. Report on the examination of Unum by multistate regulators (November 2004). investors.unum.com H
188. T. Coleman, Plaintiff magazine (September 2016): summary of the California settlement. plaintiffmagazine.com L
189. UnumProvident — Form 8-K (18 November 2004). investors.unum.com H
190. Outcome of the reassessment of Unum claims under the regulatory settlement agreement (2008). en.wikipedia.org M
191. Australia’s banking royal commission, round 6 (September 2018), as summarised by ABC. abc.net.au H
192. Australia’s banking royal commission, round 6 (September 2018): CommInsure. minterellison.com M
193. Australia’s banking royal commission, round 6 (September 2018): ClearView. abc.net.au H
194. Australia’s banking royal commission, round 6 (September 2018): QBE. abc.net.au H
195. ASIC — Report 498 (2016): disputes in life insurance. asic.gov.au H
196. FCA — review of total-loss vehicle valuations (27 March 2024). fca.org.uk H
197. ASIC — 20-302MR: the Youi case (November 2020). asic.gov.au H
198. Financial Ombudsman Service — Annual complaints data and insight 2024/25. financial-ombudsman.org.uk H
199. Financial Ombudsman Service — Annual complaints data and insight 2021/22–2024/25. financial-ombudsman.org.uk H
200. Versicherungsombudsmann e.V. — annual report 2025: outcomes of complaints. versicherungsombudsmann.de H
201. Versicherungsombudsmann e.V. — annual report 2025: insurers correcting their decisions. versicherungsombudsmann.de H
202. US Supreme Court, State Farm v. Campbell, 538 U.S. 408 (2003): account of the Utah court’s decision. law.cornell.edu H
203. US Supreme Court, State Farm v. Campbell, 538 U.S. 408 (7 April 2003). law.cornell.edu H
204. InsuranceNewsNet — California Supreme Court and the fine against UnitedHealth (January 2019). insurancenewsnet.com M
205. Eskew v. Sierra Health and Life — jury verdict (2022) and Nevada Supreme Court ruling (2024). practiceguides.chambers.com M
206. ERISA (1974): remedies in benefit disputes. en.wikipedia.org M
207. US Supreme Court, Pilot Life Insurance Co. v. Dedeaux, 481 U.S. 41 (1987). law.cornell.edu H
208. US Supreme Court, Aetna Health Inc. v. Davila (21 June 2004). law.cornell.edu H
209. WGCU — fines imposed by Florida’s Office of Insurance Regulation (2 September 2025). wgcu.org M
210. ASIC — 26-236MR: Hollard fined $2 million for serious insurance claim delay (October 2026). asic.gov.au H
211. ASIC — 26-236MR; researcher’s calculation. asic.gov.au M
212. ASIC — 23-179MR: IAL penalised $40 million over pricing discount failures (30 June 2023). asic.gov.au H
213. ASIC — Report 765: When the price is not right (2023). download.asic.gov.au H
214. Wendell Potter’s testimony to the US Senate Commerce Committee (24 June 2009), PBS transcript. pbs.org H
215. Democracy Now! (21 June 2007): Linda Peeno’s testimony to Congress on 30 May 1996. democracynow.org M
216. ASIC — 26-236MR: quotation from the court’s ruling. asic.gov.au H
217. OpenSecrets, as summarised by Repairer Driven News (9 February 2026). repairerdrivennews.com M
218. McCarran–Ferguson Act (9 March 1945). en.wikipedia.org M
219. Competitive Health Insurance Reform Act (13 January 2021). justice.gov H
220. KFF Health Tracking Poll (July 2025). kff.org H
221. KFF Health Tracking Poll (January 2026). kff.org H
222. Allianz UK — Edelman survey (22 July 2026). allianz.co.uk M
223. National Bank of Ukraine — White Paper on the development of the insurance market (2020), as summarised by the press. bank.gov.ua M
224. League of Insurance Organisations of Ukraine — All-Ukrainian National Insurance Conference (27 March 2026). interfax.com.ua M
225. VKURSI Market BI — insurers’ profit for 2025. epravda.com.ua M
226. UAINSUR — the first half of 2026. uainsur.com M
227. Interfax-Ukraine — insurers’ indicators for 2025 based on National Bank of Ukraine data (31 March 2026). interfax.com.ua M
228. Interfax-Ukraine — insurers’ indicators for 2025 based on National Bank of Ukraine data (31 March 2026): lowest payouts. interfax.com.ua M
229. MTSBU — MTPL statistics, as summarised by Minfin and finance.ua: 2025. mtsbu.ua H
230. MTSBU — MTPL statistics, as summarised by Minfin and finance.ua: January–September 2026. mtsbu.ua H
231. dtkt.ua — whether insurers must pay for property damaged during the war (2 July 2022). news.dtkt.ua M
232. World Bank — updated assessment of Ukraine’s recovery needs, RDNA4 (25 February 2025). worldbank.org H
233. GMK Center — who insures the Ukrainian economy against missiles and losses, and how (23 January 2026). gmk.center M
234. National Bank of Ukraine — review of financial-service consumers’ complaints, the second quarter of 2022. espreso.tv M
235. National Bank of Ukraine — review of financial-service consumers’ complaints for 2025. bank.gov.ua M
236. Ekonomichna Pravda, Interfax-Ukraine, UAINSUR — number of insurers according to National Bank of Ukraine data. uainsur.com M
237. Interfax-Ukraine — National Bank of Ukraine revoked the licences of ASKO DS and Peremoha (27 July 2026). interfax.com.ua M
238. Ministry of Finance of Poland — reply to parliamentary question No. 2876 (19 November 2025). api.sejm.gov.pl H
239. Ministry of Finance of Poland — reply to question No. 2876: supervisory assessment of BION. api.sejm.gov.pl H
240. CMS — Interoperability and Prior Authorization Final Rule (17 January 2024). cms.gov H

Read this briefing in Ukrainian →