How Do Insurance Companies Make Money? Float and Combined Ratio

Insurance companies make money in two ways. They earn an underwriting profit when the premiums they collect exceed the claims and expenses they pay, which is measured by the combined ratio. And they earn investment income on float, the premiums they hold and invest between the day a policy is sold and the day claims are paid. The best insurers make money on both; many make a small loss on underwriting and rely on investment income to come out ahead.

Warren Buffett built Berkshire Hathaway on the second engine. In 1967 Berkshire bought two Omaha insurers for about $8.6 million, mainly for their float. By the end of 2025 that float had grown to about $176 billion, and for most of two decades it cost less than nothing, because Berkshire's insurers made underwriting profits. Berkshire was, in effect, paid to hold other people's money.

Why is insurance an inverted business?

Most companies learn their costs before they set their prices. An insurer prices first, collects first, pays later and learns its true cost last. A car policy written in January may produce a claim in December, a lawsuit years later and a final settlement a decade after that.

That inversion gives insurers their two sources of profit and one great temptation. The temptation is to accept an underwriting loss to gather more float, on the theory that investment income will cover it. When interest rates are high, that can work for a while. When rates are low, it destroys capital. The book's Chapter 8 on insurance calls this the central tension of the business.

What is the combined ratio?

The combined ratio is the main measure of underwriting:

Loss ratio = (claims paid + change in claim reserves + loss adjustment expenses) ÷ earned premium

Expense ratio = (commissions + underwriting and administrative costs) ÷ premium

Combined ratio = loss ratio + expense ratio

A combined ratio of 94% means the insurer keeps 6 cents of underwriting profit per dollar of premium before investment income. A ratio of 103% means it loses 3 cents.

The expense ratio is mostly distribution. Agent and broker commissions typically run about 10–20% of premium in personal and commercial lines. Direct writers such as GEICO and Progressive's direct channel replace commissions with advertising, which is why US car insurance is among the most heavily advertised categories in the economy.

What is float and why is it valuable?

Float is roughly loss reserves plus unearned premiums: money the insurer holds but does not own. Its cost is the underwriting loss divided by the float.

Cost of float = underwriting loss ÷ average float

If an insurer runs a 102% combined ratio on $1 billion of premium, a $20 million underwriting loss, and holds $1.3 billion of float, the float costs about 1.5% a year. If Treasury bills yield 4%, that is cheaper than borrowing. At a 94% combined ratio the cost is negative: the insurer is paid to hold the money.

How valuable float is depends on four things, according to the book's insurance industry study:

1. The length of the tail. A home insurer pays most claims within a year, so its float is small relative to premium. A long-tail casualty writer or reinsurer may hold two or three years of premium.
2. Interest rates. From about 2009 to 2021 near-zero yields made float worth little. The 2022–23 rate rise roughly doubled new-money yields.
3. What it can be invested in. Regulators and rating agencies penalize stocks and illiquid assets, so most insurers hold investment-grade bonds. Berkshire Hathaway, with very large equity capital behind its float, can buy stocks and whole businesses instead.
4. Growth. A growing insurer's float grows with premium. A shrinking one pays out more than it collects.

How do underwriting and investment income combine?

The book's stylized insurer in Appendix A earns $1,000 million of premiums a year, holds $800 million of equity and $1,500 million of float, and invests both. The scenarios are illustrative, not any real company's figures.

Two lessons follow. Float is valuable only while underwriting keeps its cost below what the money earns. And an insurer running a 105% combined ratio pays about 5% a year for its float, which is usually worse than borrowing.

Why are reserves where insurers hide mistakes?

An insurer's reported profit is premiums minus expenses minus estimated losses. The estimate is the reserve: case reserves on reported claims, plus actuarial estimates of claims incurred but not yet reported. Nobody can prove the estimate wrong for years.

Booking a 64% loss ratio instead of 69% on a $1 billion book adds $50 million to this year's profit. If the higher number was right, the bill arrives later as adverse development. Management bonuses, share prices and ratings all reward good current results, so under-reserving and underpricing tend to arrive together in soft markets. Long-tail lines are worst: US commercial auto, general liability and umbrella lines saw repeated adverse development in 2019–25 as jury awards grew, a trend often called social inflation. Reliance Insurance, placed in liquidation in 2001 after years of aggressive growth and reserve deficiencies, is a large US example, as reported.

How did AIG nearly destroy itself?

AIG shows what happens when an insurer sells protection without reserves or float. Its Financial Products unit sold credit default swaps on about $527 billion (notional, end-2007) of corporate loans and mortgage securities, including about $78 billion on super-senior tranches of mortgage CDOs. Because AIG was highly rated, it posted little collateral and held no reserves against the risk.

The contracts required AIG to post collateral if the protected securities fell in value or if AIG was downgraded. When it was downgraded on 15 September 2008, collateral calls arrived that it could not meet. The next day the Federal Reserve lent it up to $85 billion. The rescue grew to a total commitment of $182.3 billion. AIG did not run out of claims-paying money first; it ran out of collateral. The Treasury sold its last shares in December 2012, and the government reported a combined positive return of about $22.7 billion. The AIG company profile has the detail.

What happens when insurers cannot charge enough?

When regulators will not let prices rise to match risk, insurers stop writing it. State Farm, a mutual, lost about $13.2 billion on underwriting in 2022 and about $14.1 billion in 2023 before returning to profit, with 2025 net income of about $12.9 billion. It stopped accepting new property applications in California on 27 May 2023, and the state approved a 17% interim rate increase on 13 May 2025. California's FAIR Plan, the insurer of last resort, grew from about 240,000 policies in September 2021 to about 669,000 by December 2025.

Catastrophes test both pricing and capital. Insured catastrophe losses worldwide were about $107 billion in 2025, according to Swiss Re, and the Los Angeles wildfires of January 2025, at about $40 billion, were the largest insured wildfire loss on record. Insurers pass their peak risks to reinsurers and to catastrophe bonds, a market of about $58–61 billion at the end of 2025.

Did insurtech change the economics?

Not the loss side. Lemonade and Root went public in 2020 at high valuations, then spent years repairing loss ratios. Root's shares fell more than 90% after its IPO; it rebuilt its pricing, grew through partners such as Carvana, and reported net income of about $40 million in 2025, its second profitable year. Lemonade reported 2025 revenue of about $738 million and an adjusted EBITDA loss of about $118 million, narrowing. The research's verdict: technology lowers expense and acquisition costs, but it does not repeal loss ratios.

How do life insurers make money?

Life insurers earn in three ways: a mortality margin on death and disability cover, a spread between what they earn on invested assets and what they credit on annuities and savings policies, and fees on variable and unit-linked products. Their liabilities run for decades, so they are large buyers of long bonds and private credit. The spread business is closer to banking than to car insurance, and the classic failure is a guarantee that proves too generous, as several variable-annuity writers found in 2008.

The takeaway

Insurers make money from underwriting, measured by a combined ratio below 100%, and from investing float, the money they hold before claims are paid. Float is a gift only while underwriting keeps its cost low; when combined ratios rise, rates fall or reserves prove short, it becomes expensive borrowing. In money-equation terms, insurance turns on the second factor, pricing risk correctly: Berkshire's float compounded because its insurers priced well and wrote less when prices were poor, while AIG priced its protection as if losses would never come. For how insurers compare with banks and payment networks, see our guide to how financial companies make money, and for the full treatment read Chapter 8, "Insurance".