Insurance Float

Concept Analysis

Definition & Origins

Insurance float is the pool of money an insurer holds between collecting premiums and paying claims. Because premiums arrive before losses are settled — an interval that sometimes extends over many years, and in extreme workers' compensation cases over many decades — the insurer gets to invest this money for its own benefit in the meantime. In a well-run insurance operation, float is essentially free — or even negatively-cost — financing, and it became the structural engine behind Berkshire's compounding.

The concept was not new when Buffett bought National Indemnity in 1967 for $8.6 million. His insight was to see that float generated under disciplined underwriting is not a transient liability but a permanent, growing source of capital. Accounting treats float as a liability in full, as if it had to be paid out tomorrow; Buffett treats it as a revolving fund — old claims are paid daily, new business replaces them daily — whose true economic cost is measured by the underwriting result, not by its balance-sheet size. Berkshire's float illustrates the scale this insight reached: $39 million in 1970, $1.6 billion in 1990, $27.9 billion in 2000, and $83.9 billion by yearend 2014.

Core Ideas

Float is only attractive at low cost. The key metric is not the size of the float but the cost of acquiring it — the underwriting loss, expressed as a percentage of float. If an insurer runs a 5% underwriting loss, its float costs 5% a year, which may still beat market rates for money but is increasingly marginal. If premiums exceed expenses and eventual losses, the float is free or negative-cost: the insurer is paid to hold other people's money. Buffett's test is stark: an insurance business has value only if its cost of float over time is below what the funds would otherwise cost; above market rates, the business is a lemon.

Scale creates a compounding float advantage. During the twelve years through 2014, Berkshire's float — money that does not belong to it but that it can invest — grew from $41 billion to $84 billion, while the operation earned $24 billion of pre-tax underwriting profit. Each new policyholder adds to the pool; the investment returns on the pool compound further. Float at this scale, held by an insurer that can never face immediate demands large relative to its cash resources, becomes a pillar of the economic fortress that competitors cannot replicate.

Float duration matters as much as size. Long-tail lines — workers' compensation, environmental liability, medical malpractice — produce float that stays for years or decades, allowing investment in longer-term, higher-yielding assets. Short-tail lines such as auto pay claims quickly. Berkshire also writes retroactive reinsurance, taking on other insurers' old losses: an upfront charge, then no further costs and the use of the float for decades.

Catastrophe years test the float model. What matters is the combined ratio across a full cycle, not in any single year. Catastrophe years will push the ratio above 100%; quiet years pull it far below. Buffett accepts that volatility because the long-term average determines whether float is genuinely free — and because Berkshire's balance sheet lets it write catastrophe covers at volumes no competitor can match.

Float has a stated cost that can be negative. Buffett reports the cost of float in the annual letters. Over the years it has run very close to zero, with underwriting profits in most years offsetting terrible ones such as 1984, when the cost was a staggering 19%. A negative cost — an underwriting profit — means Berkshire is paid to hold the float pool, a result no other large insurer has sustained over an extended period.

Practical Application

National Indemnity (acquired 1967): The original float model. A specialty property/casualty insurer willing to write unusual risks other insurers refused — at adequate premiums. The discipline of charging properly for odd risk, and walking away when competitors underpriced, built the underwriting culture on which the whole float structure rests.

GEICO (fully acquired 1995–96): The defining float machine. Direct distribution eliminates agent commissions, so GEICO can price below competitors and still underwrite profitably. The 2000 letter makes the contrast concrete: in a difficult year, GEICO's underwriting loss of 4% of premiums produced a float cost of 6.1%, while State Farm — the industry's largest player — tolerated a float cost of about 23%. Buffett's expectation for GEICO is that float, over time, will be free.

General Re (acquired 1998): The cautionary case. General Re brought a large reinsurance float but had underpriced severely, and fixing it took years of repricing and dropped business under new management. In 2000 the cleanup contributed to a $1.6 billion Berkshire underwriting loss — a 6% cost of float. The lesson: float quality depends entirely on the underwriting culture generating it, and culture cannot be acquired, only built.

Catastrophe and retroactive reinsurance: Berkshire writes covers no competitor can match because its financial strength lets it pay immediately and in full. Retroactive contracts trade an upfront, known charge for float usable for decades — inferior to profit-generating float, but decent business at the right price.

Common Misconceptions

Misconception 1: Any insurance float is valuable. Float from undisciplined underwriting — combined ratios persistently above 100% — is expensive capital, not free capital. Because loss costs must be estimated, insurers have enormous latitude in reporting, and reserve errors flow straight into earnings; apparent premium growth can disguise float costing more than market rates for money, destroying value rather than creating it.

Misconception 2: Insurance is a commodity business. The rare skill is charging adequate premiums across the cycle — walking away in soft markets while competitors cut prices to hold share. GEICO's structural cost advantage lets it price low and still profit; General Re's failure was accepting inadequate premiums for volume. The discipline is cultural, not actuarial.

Misconception 3: The float model is easily replicable. Every insurer wants an underwriting profit, and that collective wish creates competition so intense that the industry as a whole usually operates at an underwriting loss — the price the industry pays to hold its float. Replicating Berkshire requires scale, financial strength, investment skill, and decades of underwriting culture in combination; no parallel exists.

Thought Evolution

Related Concepts

Case Companies