General insurance: the combined ratio and the float

A general insurer - motor, health, property - can pay out more in claims and running costs than it collects in premiums and be a thoroughly good business anyway. That is not a paradox to explain away; it is the ordinary state of affairs, and it turns on a fact with no equivalent in most industries. The insurer collects premiums today and pays claims months or years later, so at any moment it holds a large pool of other people's money - the float - which it invests in the meantime. It has two engines: the insurance itself, and the investment income on the float. It can run the first at a loss and still make money on the second. illustrative

The number that reads the first engine is the combined ratio: claims paid plus expenses, as a percentage of premiums earned, with 100 the dividing line. Below 100 the insurer collected more than it paid out - an underwriting profit; above 100, an underwriting loss. But a combined ratio of 102% does not mean the business lost money. It means underwriting lost 2%, which the float's investment income may more than cover. A reader trained on ordinary businesses - where the P&L is the whole story - will misjudge exactly this case, treat the insurer as unprofitable, and reject one of the best structures in finance.

This module reads that two-engine business, and it exists because judging a general insurer on its bottom line, or worse on its combined ratio alone, misses both how the money is made and where the risk sits. Three ideas carry it. The combined ratio measures whether the underwriting makes or loses money. The insurance float - premiums held against future claims, invested for the insurer's own account - is where most of the profit and nearly all of the compounding really come from. And reserve adequacy is the honesty check that sits behind the reported claims figure the way provisioning sits behind a bank's: a combined ratio is only as honest as the reserves behind it, and an insurer that sets aside too little for claims already incurred can manufacture an underwriting profit that reverses a year or two later.

Take the two engines in turn, then the judgement behind both.

The underwriting engine - the combined ratio. Add the claims ratio (claims paid as a percentage of premiums earned) to the expense ratio (the cost of running the business, similarly). Their sum is the combined ratio. Below 100 the insurer made an underwriting profit; above 100, a loss. A combined ratio of 102% is an underwriting loss of 2% of premium - real, but small, and not the whole story.

The investment engine - the float. Because premiums come in before claims go out, the insurer permanently holds a large pool of money that belongs, in effect, to future claimants but is the insurer's to invest until the claims fall due. Invested at even a modest yield, a float several times the size of annual premium throws off income that can dwarf the underwriting result. Profit before tax is the underwriting result plus this investment income - so an insurer with a 102% combined ratio and a strong float can be more profitable than one with a 98% combined ratio and a weak float. You never read the combined ratio in isolation; you read it against what the float earns, and ask which engine is doing the work.

This is why the greatest insurance investors care more about the size and cost of the float than about the combined ratio in any single year. A combined ratio slightly above 100 that lets the insurer hold a large, growing float cheaply can be worth far more than a 95 on a small one. The float, not the underwriting margin, is where the compounding lives - and an underwriting loss of a few percent is a small price for holding it.

The lever - reserve adequacy. Here is the number that most often deceives. When a claim has occurred but not yet been paid or even fully assessed, the insurer must estimate it and set aside a reserve. Set the reserve too low and the claims ratio looks better than reality, the combined ratio improves, and this year's underwriting flatters - until the claims are actually settled and the shortfall lands as a top-up. This is the general-insurance twin of a bank's provisioning: the reported underwriting profit is only as honest as the reserves behind it, and an improving combined ratio built on thinning reserves is a warning wearing the costume of an achievement.

The two-engine structure is specific to insurance, and it reads differently even from a life insurer, let alone a manufacturer. It is worth seeing exactly which parts have no equivalent elsewhere.

The distinction to hold is that a general insurer profits from holding other people's money, and can run its actual insurance at a loss while doing so. No manufacturer has a float; its money is its own, tied up in inventory and plant. The insurer's float is the opposite - a large, cost-free (or better than cost-free) pool that grows as the business writes more premium, with a second profit engine bolted onto the first. A life insurer holds a float too but reads nothing alike, its profit spread over decades rather than settled within the year. And the bank row makes the finer point: reserving here is the exact structural twin of provisioning there - in both, the biggest judgement sits behind the reported profit and can flatter it.

Read the composite general insurer's fifth year. Its combined ratio is 102% - a claims ratio of 72% plus an expense ratio of 30% - so underwriting is a loss of ₹180 crore on ₹9,000 crore of net earned premium (the premium it keeps and has actually earned over the year). Read that number alone and you would call the business unprofitable. illustrative

Now read the second engine. The insurer holds an investment float of ₹14,000 crore - premiums collected against claims not yet paid - and earns 8% on it, or ₹1,120 crore. That comfortably covers the ₹180 crore underwriting loss and leaves a profit before tax of ₹940 crore. The business makes money; it just makes it on the float, not on the insurance. And the float is durable: as the insurer writes more premium each year, it grows, and the investment income grows with it. That compounding pool, not the combined ratio, is the heart of the value.

Then look at the weak year, the composite's third, to see both engines wobble at once. A claims spike pushed the combined ratio to 104% - an underwriting loss of ₹308 crore - and a soft investment year cut the float yield to 6.5%, so income fell to ₹767 crore. Profit before tax nearly halved, to ₹459 crore. That is the risk of the model laid bare: in a bad year the underwriting loss widens and the float earns less at the same time, and the two engines that usually offset can both misfire together. Reading only a good year would miss how the profit behaves under stress.

The instrument makes both truths physical. Push the combined ratio above 100 and the underwriting result goes negative - yet raise the float yield and overall profit stays positive, because the float covers the loss. Then tick under-reserve, and the combined ratio drops a few points while a deferred top-up appears alongside it: the reported improvement is borrowed from a future year. That deferred top-up is what the reserve-development triangle exposes over time.

The habit to build: read a general insurer as two engines, never one. Judge the underwriting by the combined ratio, but treat a figure a little above 100 as normal, not as a loss. Read the float - its size relative to premium and the yield it earns - because that is where much of the profit and nearly all of the compounding come from. And before trusting either, check the reserves behind the claims ratio: an improving combined ratio on thinning reserves is a top-up waiting to happen.

The combined ratio tells you whether underwriting made or lost money this year, but not whether the reserves behind it are adequate - and that is the number that most often deceives. Reserves for claims incurred but not reported, and for claims still being assessed, are estimates; an insurer that estimates them low reports a lower claims ratio and a better combined ratio with no change in the actual claims experience. The visible number cannot certify its own honesty. Only the reserve-development disclosures - how previous years' reserves turned out against what was set aside - can, and they lag by years.

Nor does the float's current investment income tell you how durable it is. A float earning steady interest on high-grade bonds is repeatable; a float that produced a big number this year through equity-market gains is not, and the two look identical on the profit line. An insurer can paper over weak underwriting with a good market year, and a reader who does not decompose investment income into steady yield versus one-off gains will mistake market luck for a durable second engine.

And neither number captures catastrophe risk - the tail that defines general insurance. An insurer can report an excellent combined ratio and a fat float for years and then face a single flood, earthquake or pandemic that overwhelms both, a loss so large it dwarfs the accumulated underwriting profits. The combined ratio is an average of ordinary years; the real risk lives in the extraordinary one - in how much catastrophe exposure the insurer has written and how much is reinsured, passed to another insurer to carry in exchange for a share of the premium. That sits in the risk and reinsurance disclosures, not in the combined ratio or the float, and it is the part that can turn a steady compounder into a solvency event - a loss big enough to threaten the insurer's ability to pay its claims.

When a combined ratio improves, a sharp analyst asks whether it was underwriting or reserving. The question sounds like this: "Your combined ratio improved 300 basis points, mostly on a lower claims ratio. How much of that is genuine loss-ratio improvement versus favourable reserve development, and what did prior-year reserves develop to this year?" The analyst is checking whether the improvement is real underwriting or a release from reserves set high before - or, worse, thin reserves now.

A good answer splits it and points to the evidence:

"Fair question. Of the 300 basis points, about 200 is genuine - we re-priced the motor book and exited two loss-making segments, and the current-year loss ratio is down accordingly. About 100 basis points is favourable prior-year development: reserves we set in earlier years proved a little conservative and released. Our reserve-development triangle, on page 180, shows consistent small favourable development, not a reversal, and our IBNR is up in line with premium. So the improvement is mostly current-year underwriting, and the reserving remains prudent."

It separates current-year underwriting from prior-year development, points to the triangle, and confirms the reserving stance. Contrast the evasion:

"We're very pleased with our combined-ratio improvement, which reflects our underwriting discipline and pricing sophistication. Our reserving philosophy has always been prudent and is reviewed by our appointed actuary. We're confident this reflects the structural quality of our book and expect the trajectory to continue."

Reassuring and unusable. It never splits the improvement, never mentions prior-year development, and gives no triangle figure. "Prudent reserving philosophy" is the same claim an under-reserving insurer would make, and "expect the trajectory to continue" implies a re-priced book and a one-off reserve release are the same durable thing, which they are not. The follow-up that settles it - "show us the reserve-development triangle: how did each of the last five years' reserves develop, and was any of this year's profit a release from them?" - forces the reserving history into daylight. The silence is the tell: either the development is unflattering, or the current reserves would not survive being shown year by year.

The first trap is reading a combined ratio above 100 as a failing business. It is a genuine underwriting loss, but only one of two engines, and the investment income on the float can more than cover it. Some of the finest general insurers run combined ratios slightly above 100 for years, because holding a large, growing float cheaply is worth more than squeezing out an underwriting profit on a smaller one. A reader who rejects every insurer with a combined ratio over 100 is discarding exactly the businesses whose value is in the float.

The second trap is admiring a falling combined ratio without checking the reserves. A lower claims ratio can mean better underwriting - or too little set aside for claims already incurred, which flatters the ratio now and forces a top-up later. The improvement looks like skill and is sometimes just thin reserving, and the two are indistinguishable on the face of the number. The check is the reserve-development history: an insurer whose prior reserves consistently prove adequate is improving honestly; one whose reserves keep developing adversely is manufacturing today's combined ratio at tomorrow's expense.

The third trap is treating investment income on the float as uniform quality. Steady interest on high-grade bonds is a durable, repeatable engine; a large equity-market gain in a good year is not, and can reverse. An insurer with weak underwriting and a big one-off gain can report the same profit as one with solid underwriting and steady float income - the profit is identical, the durability is not, and only decomposing the income tells them apart. Before you conclude a falling combined ratio proves a better insurer, name in advance the figure that would flip your read: prior reserves developing adversely, an improvement driven by thin reserving rather than pricing, or a profit leaning on one-off gains. If you cannot name it, you are admiring the combined ratio, not interrogating it.