Why Time Is on Your Insurer's Side: Float, the Katrina Year, and the Economics of Waiting
The short version. An insurer collects your premium today and pays claims later, sometimes much later. In between, that money is invested, and the investment earnings belong to the insurer. Warren Buffett built Berkshire Hathaway substantially on this mechanism, which he has long described as money an insurer holds but does not own. None of this is a scandal; it is the published business model of the industry. But it has one consequence every claimant should understand: in a payment dispute, time costs you and pays them. The pressure to accept a low first offer is not a personality trait of your adjuster; it is arithmetic. This page walks through that arithmetic using the industry's own numbers from its worst-ever catastrophe year, then points at the levers that exist precisely because of it.
The model: underwriting can lose money and the business still wins
A property/casualty insurer runs two businesses at once. The first is underwriting: collect premiums, pay claims and expenses. The second is investing: hold the accumulated premium money, the "float", in bonds and stocks until claims come due, and keep what it earns.
The industry's own full-year 2005 income statement, published by ISO and the Property Casualty Insurers Association of America, shows the shape plainly. Underwriting lost $5.9 billion that year. Investments returned $59.2 billion, $49.5 billion of investment income plus $9.7 billion in realized capital gains. The industry finished with $43.0 billion in after-tax net income, at the time the largest annual dollar profit it had ever reported, on the heels of $38.5 billion in 2004.
Read that again as a claimant: the claims-paying side of the business can run at a loss, indefinitely, and the enterprise still profits; provided the money sits invested as long as possible before it goes out the door. Every day between your loss and your check is a day your settlement is still part of the float.
The Katrina test: what "unprecedented adversity" did to the balance sheet
2005 is the cleanest possible test of the model, because it was the worst year the industry had ever seen. Hurricane Katrina: plus Rita and Wilma: drove record catastrophe losses of $57.7 billion across more than 3.3 million claims (ISO's figures; both records at the time).
Here is what happened to the industry's financial position in that year, per ISO/PCI:
Policyholder surplus: the industry's claims-paying capital, rose 9.2 percent, from $391.3 billion at the end of 2004 to $427.1 billion at the end of 2005. Net income was the $43.0 billion above. The combined ratio, claims-plus-expenses against premiums, was 100.9, a year of historic catastrophe barely nudged underwriting past break-even. Within months, insurers and reinsurers raised roughly $23 billion in fresh capital, because investors wanted in.
The industry's own trade association, the Insurance Information Institute, disputes the phrase "record profits" for 2005, and its argument deserves to be stated fairly: as a rate of return, 9.8 to 10.5 percent on average surplus, 2005 was ordinary, and it trailed the Fortune 500 average of 14.9 percent, as the industry has in most years since 1987. Both things are true at once. In dollars, 2005 was the biggest profit the industry had ever booked; as a return on its capital, it was middling. What no one disputes: the costliest catastrophe year in American history left the industry's claims-paying capital larger than it started, its profit at a then-record dollar high, and capital markets eager to add more.
That resilience is, in one sense, exactly what you want from your insurer, a carrier that fails after a hurricane pays nobody, and the industry is right about that. The claimant's takeaway is narrower and it survives every rebuttal: the institution across the table from you is built to absorb the worst year in history and come out ahead. You are not. Your leverage is never capital. It is process: and process is what the rest of this page and this site's claims pages are about.
The tax chapter: what GAO found, and what changed
The float model once produced a tax result striking enough that Congress's own auditors flagged it. In 1985, the U.S. General Accounting Office (now the Government Accountability Office) reported on federal taxation of the property/casualty industry (GAO/GGD-85-10, "Congress Should Consider Changing Federal Income Taxation of the Property/Casualty Insurance Industry").
GAO's core numbers, for 1974 through 1983: the industry ran about $28 billion in underwriting losses, earned about $100 billion in investment gains, for a total gain of about $72 billion, and paid federal income tax of roughly 2 percent of that total gain. In individual years the industry's federal income tax was negative, refunds exceeded payments, while the industry as a whole was profitable. The mechanics were legal: deductions for loss reserves and policy acquisition costs, layered onto largely tax-favored investment income.
Two honest codas. First, Congress acted: the Tax Reform Act of 1986 rewrote insurer taxation, and the old arithmetic does not describe current law. Second, the industry's 2005 statement shows the after picture, $11.2 billion of tax on $54.2 billion of pre-tax income, an effective rate around 21 percent. The GAO finding is history, not a description of today. It stays in this guide because it demonstrates the same structural point as the float itself: the investment side of this business, not the claims side, is where the economics live, and for a decade that was true enough to make the claims side a tax shelter.
What this means standing at the counter
Nothing above says your insurer is cheating you. Most claims close without a fight, and this site sells data to this industry; we have every reason to state this carefully, and the numbers above are all the industry's own or the government's.
What the economics do explain:
The one incentive, stated precisely, and it is the only claim about motive on this page. Money that stays invested keeps earning; money paid out stops earning. That is arithmetic, not an accusation, and it is where the demonstrated economics end. What follows from it is an inference, and worth labelling as one: a claimant who accepts the first number closes the file at its lowest cost, and one who disputes it does not. That is an economic pressure, not a policy, and it is not the only incentive in the room: carriers also face regulatory deadlines, market-conduct examinations, reputational cost, customer retention, litigation exposure and their own claim-handling expenses, several of which push hard toward settling quickly and correctly. The float does not prove anyone is stalling. It explains why the pressure to take the first offer exists at all, and why it is structural rather than personal to your adjuster. The checks on it are external: regulators, appraisal clauses, courts.
Why delay is not neutral. For you, a month of waiting is rental cars, a loan on a car that no longer exists, a repair on hold. For the carrier, it is a month of investment income on your settlement, multiplied across every open file. Neither side is behaving irrationally. The sides are just not symmetric.
Why the burden of pushing back is on you, and why the levers exist. Claims-handling law (every state has an unfair claims settlement practices regime), appraisal clauses, and state insurance departments all exist because legislatures understood this asymmetry. The levers work, but none of them engage on their own.
The levers, in the order to reach for them
Start with the dispute itself, not the economics: identify the specific line that is wrong: the valuation, the omitted taxes and fees, the refused operation. Total loss: why the first check is low walks the valuation fight; when insurance won't pay for a proper repair walks the repair fight.
If your dispute is the number, most auto policies contain an appraisal clause: either side can demand appraisal, each hires its own appraiser, the appraisers pick an umpire, and the result usually binds both sides on value. In some states, determining auto damage values is itself a licensed activity, Massachusetts and South Carolina license motor vehicle damage appraisers, so there is a professional, regulated layer here that most claimants never learn exists. We are building the per-state appraisal register now; until it ships, the clause in your own policy is the text that governs.
If your dispute is the conduct: missed deadlines, no explanation for a denial, lowball-and-stall; that is what your state's unfair claims settlement practices rules cover, and the regulator that enforces them takes consumer complaints for free: your state's insurance department, with its complaint page and phone line. If your car was totaled, check whether your settlement had to include sales tax and fees in your state: we publish what we have verified, state by state, on your state's claims-help page.
And if you keep only one sentence from this page: the money does not move until you make it move.
Sources
Every figure above is from one of these documents, each read from the publisher's own site on 2026-08-13:
Insurance Information Institute, "2005 -- Year End Results" (ISO/PCI data): the full-year 2005 income statement ($417.7B earned premiums, -$5.9B underwriting, $49.5B investment income, $9.7B realized gains, $54.2B pre-tax, $11.2B tax, $43.0B net income, $427.1B year-end surplus, combined ratio 100.9); catastrophe losses of $57.7 billion across 3.3 million claims; surplus growth of 9.2 percent; the ~$23 billion in post-Katrina capital raising; and the industry's own rate-of-return rebuttal (9.8, 10.5 percent vs. Fortune 500 14.9 percent), which we quote rather than bury. iii.org/article/2005-year-end-results
U.S. General Accounting Office, GAO/GGD-85-10, "Congress Should Consider Changing Federal Income Taxation of the Property/Casualty Insurance Industry" (1985): the 1974: 1983 figures (~$28B underwriting losses, ~$100B investment gains, ~$72B total gains, ~2 percent federal income tax on total gains, negative-tax years). gao.gov/products/ggd-85-10
Berkshire Hathaway shareholder letters (multiple years) describe float and its economics; the characterization of float as money held but not owned is Buffett's.
How to read the claims on this page. Four kinds of statement appear above and they are not equal. DOCUMENTED FACT: every figure, each cited to the publisher and the date we read it. ECONOMIC MECHANISM: float, and the fact that invested money keeps earning until it is paid out, arithmetic, not motive. INFERENCE, labelled where it appears: that a disputed claim costs more to close than an accepted one, which follows from the mechanism but is not measured here. NOT CLAIMED AT ALL: that any insurer delays deliberately, that adjusters act in bad faith, or that the 1985 tax findings describe current law. If a sentence here ever reads like more than its category allows, that is a defect; tell us and we will correct it on the page.
General consumer information, not legal, insurance, or financial advice. The 1985 GAO findings describe pre-1986 tax law, which the Tax Reform Act of 1986 changed. State law and your policy control your claim.
General consumer information: not legal, insurance, or financial advice. Requirements, coverage, and practices vary by state, policy, and manufacturer.
Where this fits
Each link says what it is for. We add one only when a reader on this page has a real reason to need that page next.
- The Second Tier of US Auto Insurers, and Why Concentration Is the Story (who writes the rest of the market once the big four are set aside)
- The Shops Are Profitable. The Balance Sheets Are Not. (the same capital structure question asked of the repairers rather than the carriers)