Money You Hold That Isnt Yours

Ask most people how Warren Buffett grew his money faster than almost anyone alive, and they will say "he picked good stocks." That is half the story, and the smaller half. The bigger half is that he invested with a huge pile of money that was not his - money he held, put to work for himself, and only had to hand back much later, if at all. In the insurance business this money has a name: float.

Here is the whole idea in one simple picture. An insurance company takes money from customers today - this payment is called a premium - and promises to pay out if something bad happens, like an accident. But the bad thing may not happen for months, years, sometimes never. In the gap between taking the premium and paying out a claim, the company is holding a big pool of cash. It does not own that cash - it belongs, one day, to the customers - but it gets to invest it in the meantime and keep whatever that investing earns. That held-in-the-middle pool is float. For a big insurer it can grow to an enormous amount, and it tends to grow rather than shrink, because new premiums keep coming in.

Now the twist that turns float from a footnote into an engine. Suppose the insurer's main business just breaks even - the premiums it collects roughly equal the claims and costs it eventually pays. Then the float has cost it nothing to borrow. It is like a giant loan with no interest, that nobody can suddenly demand back, and that keeps topping itself up as fresh premiums arrive. Invest that free money wisely for decades and you have a money machine most investors can only watch from the side. This study is about reading that machine honestly - including the ways it can quietly turn against you.

Buffett has been unusually open about this, because it is the part of his company, Berkshire, that people who think of it as just a stock-picking fund miss completely. Berkshire's insurers grew their combined float from a few million dollars in the 1960s to well over a hundred billion. In his letters he calls float money that does not belong to Berkshire but that Berkshire gets to invest, and he is careful to name its cost: the cost of float is simply how the insurance business itself did. Write policies at a profit and the float costs less than nothing - you are actually being paid to hold other people's money. Write them at a loss and the float carries a cost, like any other loan.

Two things made this work instead of blowing up, and both are matters of record. First, Berkshire kept strict discipline in its insurance - it was willing to write less business, even shrink, rather than cut prices to win customers who would later cost more in claims than they paid in premium. Buffett has again and again praised managers who walked away from cheap, badly priced business, because the other path - chasing more premium at any price - is exactly how insurers destroy themselves. Second, Berkshire held the float as permanent money inside a company that could never be forced to sell its investments to give cash back in a hurry. That meant a market panic was a chance to buy, not a threat. The float and the calm temperament were one single system.

Follow one rupee of premium through the machine and the whole idea becomes clear.

The number that decides whether float is a blessing or a slow poison is the combined ratio. It is simple: total claims plus costs, divided by the premiums collected. Below 100% the insurer makes a profit on its insurance - so it is being paid to hold the float, and the float's cost is less than zero. At exactly 100% the float is free. Above 100% the insurer loses money on its insurance, and that loss is the interest it is paying on the float. So the whole question "is this float a gift or a trap?" comes down to discipline: does this insurer keep its combined ratio at or below 100% year after year, or does it chase more premium into losses?

The second thing to read is that float behaves like borrowing without the dangerous part. Normal borrowing can be called back on a bad day - the lender says "give my money now," often at the worst possible moment. Float has no such lender. That is why Buffett could hold float-funded investments right through market crashes, and even buy more, while a person who had borrowed to invest was being forced to sell at the bottom. Float is patient in a way borrowed money almost never is - as long as the insurance stays honest and the promises set aside for claims are real.

Take a made-up Indian insurer to see the power of it. illustrative Say it holds ₹1,000 crore of float, backed by ₹400 crore of the owners' own money (this owners' money is called capital). It writes at a combined ratio of 99% - a small insurance profit, so the float is truly free. It invests the whole ₹1,400 crore pool at a steady 9% a year.

The owners put in ₹400 crore. The business invests ₹1,400 crore. At 9%, that is ₹126 crore of investment income in a year - plus the small insurance profit - on an owners' base of just ₹400 crore. So the return on the owners' own money is about 32%, not 9%, because the float did most of the heavy lifting. That gap between the 9% the investments earned and the roughly 32% the owners earned is the float acting like a lever. Now let it run for ten years, with the float slowly growing as more policies are written, and the owners' money races far ahead of what plain 9% investments could ever have done alone.

Now reverse the discipline and the machine reverses too. Let the insurer chase growth and slip to a combined ratio of 108% - an 8% insurance loss. Now the float costs 8% a year, the investments earn 9%, and the tiny gap between them almost vanishes. The "free lever" has become an expensive lever on a pool far bigger than the owners' own money - and a couple of bad claim years can turn that 32% dream into a loss that eats into the owners' capital. Same float, same investments - opposite ending, decided entirely by the discipline of the insurance.

Float is not the owners' own money, and treating it as free capital is the classic mistake. The money will, in the end, be paid out; the insurer is holding it, not keeping it. An insurer that invests its float in things it cannot sell quickly, and then faces a wave of claims, can be forced to sell at the worst time - the very trap float was supposed to avoid - if the claims come due faster than expected. Free to hold is not the same as free to keep.

Discipline is fragile, and losing it is invisible for years. An insurer can boost its premiums and its cash today by charging too little for risky policies, and the accounts will look wonderful - right up until the claims on those cheap policies arrive in later years. So a beautiful combined ratio today can be the mark of careful pricing, or a bomb quietly ticking. Telling which requires looking at whether enough money is being set aside for future claims, not just the headline profit.

The money set aside for claims is a guess, and guesses can be too hopeful. The claims not yet paid are booked as reserves - management's best guess at future payouts. Set aside too little, and today's profit and float both look better than they really are; the shortfall shows up later as "reserve strengthening," a polite phrase for "we were wrong and it costs us now." Claims that settle years later (like liability cases) are far easier to misjudge than claims that settle fast (like motor accidents).

Disasters, all at once. Float quietly assumes claims are fairly steady and spread out. A single big disaster, or many claims arriving together, can demand cash suddenly - and if the float was invested for the long term, that mismatch bites hard.

This is the part of Buffett's edge that transfers least, and honesty means saying so loudly. Float is available to an owner of an insurance company, run with strict discipline, holding permanent money. A normal investor cannot make float. The nearest thing they can reach is borrowing against their shares - and that is its dangerous opposite, because it can be called back on the worst day, which is exactly the feature float does not have. Anyone who reads "Buffett used borrowed money" and thinks "so I should borrow to invest" has learned exactly the wrong lesson.

What does transfer is the reading skill. When you look at an insurer as a possible investment, you now know to judge it on discipline (its combined ratio over many years), on honesty about reserves (does it keep having to add more, a sign of past hope over care?), and on the quality and easy-to-sell nature of what it has invested in - not on premium growth, which is the easiest and most dangerous number to puff up. And you know to treat float as borrowed money whose cost equals the insurance result, never as the owners' own free money.

• Find the combined ratio, over many years. Steadily at or below 100% means the float is cheap or free; steadily above means it is costly borrowing.
• Distrust premium growth on its own. Fast growth with a falling combined ratio can be discipline; fast growth won by cutting prices is usually a bill arriving later.
• Read how reserves change. An insurer that keeps having to add more to its reserves has a history of setting aside too little - today's profit may be borrowed from tomorrow.
• Check that the investments can be sold in time to pay the claims. Long-delayed claims funded by hard-to-sell investments is where float turns into a forced seller.
• Never confuse float with borrowed money. Float is patient; a loan is not. Do not copy the lever without the structure that makes it safe.