The Insurance Business Model: An Investor's Thesis on Float, Underwriting, and How to Value Insurers

The One-Sentence Thesis

Almost every business on earth must raise money before it can operate: it borrows from a bank, sells equity, or waits for customers to pay. An insurance company is the rare business that gets its customers to hand over cash first — and often gets paid to hold it. That single structural quirk is the entire reason Warren Buffett built the world's most valuable conglomerate on top of an insurer.

If you understand nothing else about insurance, understand this: an insurer collects premiums today and pays claims years later . The gap between the two is a giant pool of other people's money it gets to invest. Whether that money is free, profitable, or quietly toxic is the whole investment case — and it is decided by one number most investors ignore.

This article assumes you already understand the building blocks. If any of these are fuzzy, start there:

What You'll Learn

Why float is the most powerful and misunderstood asset in finance
How to read a combined ratio in five seconds and know if underwriting is healthy
Why property-casualty (P&C) and life insurance are two completely different businesses wearing the same label
The real moats in insurance: cost, data, distribution, and discipline
How to value a P&C insurer (price-to-book vs return-on-equity) and a life insurer (embedded value and appraisal value), with worked numbers
US worked examples: GEICO, Progressive, Chubb, Berkshire
India worked examples: ICICI Lombard, Star Health, HDFC Life, LIC

First, The Two Families of Insurance

Before we go further, one distinction runs through this entire article. Insurance splits into two broad families, and they are analysed in completely different ways.

Property & Casualty (P&C) — often called "general insurance" in India — covers damage to things you own and liabilities you might owe others. Property pays when your car, home, cargo, or equipment is damaged, destroyed, or stolen. Casualty pays when you are legally responsible for harming someone else or their property (auto liability, product liability, professional indemnity). Health insurance is usually grouped here too. The defining feature is that P&C is short-tail: claims are typically paid within months to a few years of the premium being collected.

Life insurance pays out on death, maturity, or over a long savings-and-protection contract. Its defining feature is that it is long-tail: a single policy can run for 20 or 30 years, so the profit emerges slowly over decades.

Tip: Keep this split in mind throughout. In this article, GEICO, Progressive, Chubb, ICICI Lombard, and Star Health are all P&C / general insurers. HDFC Life and LIC are life insurers. Because P&C claims are paid fast and life claims are paid over decades, the two are judged on almost entirely different metrics — which is exactly why we value them with different tools later on.

Where This Model Came From: Coffee, Fire, and the Mathematics of Death

Insurance feels like a modern financial product, but its core ideas were hammered out over three thousand years by merchants, a coffee-house owner, an astronomer, and a city that burned to the ground. Understanding how the model was invented is the fastest way to understand why it works.

The oldest instinct is simply spreading risk. Chinese river merchants thousands of years ago split their cargo across many boats, so that one wreck could never sink a trader — the first risk pool. Around 1750 BCE, the Code of Hammurabi in Babylon formalised "bottomry": a merchant paid a lender an extra sum, and if his ship or caravan was lost, the loan was cancelled. That extra sum was, in effect, the first premium.

But the recognisable modern industry — with float, risk-based pricing, and actuarial science — was born in a remarkable burst around London in the century after 1666.

How the modern model was invented

Three inventions — risk-based pricing, underwriting-on-data, and the mortality table — appeared within 80 years of each other and still define the business today.

1666–1681

Fire creates property insurance

The Great Fire of London destroyed roughly 13,200 houses. A physician-turned-economist, Nicholas Barbon, had been quietly keeping fire statistics — and in 1680 opened the "Fire Office," arguably the first fire insurer. Crucially, he priced by risk: a timber house cost 5% of its annual rent to insure, a brick house just 2.5%.

1688

A coffee house invents underwriting

Merchants, ship-owners and financiers gathered at Edward Lloyd's coffee house near the Thames, drawn by his reliable shipping news. To insure a voyage, a financier would sign his name under the details of the risk — which is literally where the word "underwriter" comes from. That coffee house became Lloyd's of London.

1693–1762

Maths tames death

The astronomer Edmond Halley (of the comet) built one of the first mortality tables in 1693. James Dodson extended it into age-based premiums, and in 1762 the Equitable Life Assurance Society launched — the first insurer to price life cover using actuarial mathematics. Life insurance became a science, not a wager.

1752

The model crosses the Atlantic

Benjamin Franklin co-founded the Philadelphia Contributionship, one of America's first fire insurers. It refused to insure the riskiest houses and used "fire marks" to identify covered buildings — early risk selection in action.

Lloyd's is still not an insurer but a marketplace of competing syndicates — the coffee-house structure, more than 330 years on.

Notice what happened in those decades. Barbon invented risk-based pricing (charge more for a timber house). Lloyd's invented underwriting on data (use shipping intelligence to price a voyage). Halley and Dodson invented the mortality table (turn death into a predictable average). Those three ideas — priced risk, informed underwriting, and the law of averages — still underpin how every premium is set today, as the next few sections will show.

Insurance in India: from Calcutta to liberalisation

India's story followed the same arc, then took a detour through the state. Life insurance began in 1818 with the Oriental Life Insurance Company in Calcutta, and dozens of private insurers grew over the next century. After independence, the government consolidated the industry: in 1956, more than 240 private life insurers were merged to create LIC, and general insurance was nationalised in 1972. For four decades, insurance in India was effectively a public monopoly.

That changed in 2000, when the IRDA (now IRDAI) reopened the market to private and foreign-backed players. The companies at the centre of this article — HDFC Life, ICICI Lombard, Star Health — are all children of that 2000 liberalisation, which is why India today has a fascinating split: a giant state incumbent (LIC) competing with nimble, high-margin private insurers.

Why This Matters: The whole industry is a story of turning fear into a priced, poolable, investable product. Once you see that risk-based pricing, data-driven underwriting, and actuarial averages were deliberate inventions — not obvious truths — you read every modern insurer as the latest expression of the same 350-year-old idea. The winners are simply the ones who price risk, use data, and pool most effectively.

Est. 1688

How An Insurer Actually Makes Money: Two Engines

Every insurer runs two profit engines bolted together. Most beginners see only the first. The professionals live in the second.

The two engines of an insurance company

Underwriting

Engine 1

Premiums minus claims and expenses

Investing the Float

Engine 2

Returns on money held before claims are paid

Engine 1 — Underwriting. The insurer prices risk. It collects premiums, and it pays out claims plus the cost of running the business (salaries, commissions, marketing, technology). If premiums exceed claims and expenses, it earns an underwriting profitUnderwriting ProfitThe money an insurer keeps from premiums after paying all claims and expenses, before any investment income. It exists only when the combined ratio is below 100%. An insurer with underwriting profit is being paid to hold its float.See all terms in the glossary. If not, it loses money on underwriting and hopes Engine 2 bails it out.

Engine 2 — Investing the float. Between the day a premium is collected and the day a claim is paid, the insurer holds a growing pool of cash. This is the floatInsurance FloatThe pool of premium money an insurer collects upfront but has not yet paid out in claims. It legally belongs to policyholders, but until claims come due the insurer can invest it. If underwriting breaks even or better, this is investable capital the insurer effectively holds for free — or gets paid to hold.See all terms in the glossary. A well-run insurer invests it in bonds, equities, or whole businesses and keeps the investment returns.

Float is not the insurer's money. It legally belongs to policyholders. But until claims come due, the insurer controls it — sometimes for decades.

Here is the magic. In an ordinary company, the money you invest is your capital or borrowed capital that charges interest. In insurance, if underwriting merely breaks even, the float is invested capital you hold for free. And if underwriting is profitable, you are being paid to hold hundreds of billions of investable dollars.

Why This Matters: Buffett's famous line is that Berkshire's float has often carried a negative cost — the company was paid to hold other people's money and then invested it. Over 60 years, that is the engine that turned a failing textile mill into a trillion-dollar enterprise. When you analyse any insurer, the first question is always: what does this company's float actually cost?

The Number That Tells You Everything: Combined Ratio

You can judge the health of an insurer's underwriting in one glance using the combined ratioCombined RatioAn insurer's claims plus expenses divided by the premiums it earned. Below 100% means the insurer made an underwriting profit (it kept money even before investing the float); above 100% means underwriting lost money.See all terms in the glossary. It is built from two parts:

Combined Ratio = Loss Ratio + Expense Ratio

Loss ratio — claims paid ÷ premiums earned. How much of every premium rupee goes back out as claims.

Expense ratio — costs of running the business ÷ premiums earned. Commissions, salaries, marketing, technology.

Below 100% → underwriting profit (the insurer keeps money before investing the float).

Above 100% → underwriting loss (the insurer is paying to hold float, and needs investment income to make up the gap).

A combined ratio of 92% means that for every ₹100 of premium, the insurer paid ₹92 in claims and expenses and kept ₹8 as pure underwriting profit — before a single rupee of investment income. A ratio of 105% means it lost ₹5 on underwriting and must earn that back (and more) by investing the float.

Here is how a few of our examples actually performed recently. Notice the gulf between world-class underwriters and the merely adequate.

Combined ratio — lower is better (recent full-year figures)

GEICO (US auto)2024 — best result this century81.5%

Chubb (global P&C)FY2024 P&C86.6%

Progressive (US auto)2025 companywide87.4%

Star Health (India health)FY25 (98.8% FY26)101.1%

ICICI Lombard (India GI)FY25102.8%

Below 100% = underwriting profit. Source: company disclosures. Always verify latest filings.

Key Point: A combined ratio above 100% is not automatically bad — Indian general insurers routinely run above 100% and still make money because investment income on the float more than covers the small underwriting loss. But it does mean the float has a positive cost of floatCost of FloatWhat an insurer effectively pays to hold policyholders' money. If underwriting runs at a loss, the float has a positive cost (like interest on a loan); if underwriting is profitable, the cost is negative — the insurer is paid to hold investable money.See all terms in the glossary. The best insurers in the world, like GEICO, are paid to hold their float. That difference compounds enormously over decades.

How Insurers Set Premiums: The Price of a Promise

A premium is the price of a promise to pay a future claim — so the insurer has to estimate a cost that has not happened yet. It does this by building the price up from parts, not by guessing a round number.

How a premium is built up

Premium = expected claim cost + expenses + cost of capital + profit

Expected claim cost — the heart of the price. Estimated as frequency × severity.

Expenses — commissions, admin, marketing, technology (the expense ratio).

Cost of capital — a charge for the money kept aside to back the risk.

Profit margin — the underwriting profit the insurer targets.

The core of it is the frequency and severityFrequency and SeverityThe two building blocks of an expected claim cost. Frequency is how often a claim occurs; severity is how much it costs when it does. Expected claim cost = frequency × severity, and it is the core of every premium an insurer charges.See all terms in the glossary of claims: how often a claim happens multiplied by how much it costs when it does. An actuary estimates both from years of historical data, separately for each risk segment.

This is why insurers ask so many questions. Every question — your age, your car, your postcode, your claims history — is an input that sorts you into a risk segment with its own frequency and severity.

A 19-year-old in a modified sports car has both higher frequency (more accidents) and higher severity (costlier repairs) than a 45-year-old commuting in a sedan — so the premium is far higher. This is exactly the risk segmentation edge that makes Progressive so formidable: price each driver precisely, win the good risks, decline or surcharge the bad ones.

Why This Matters: None of this works without the law of large numbersLaw of Large NumbersThe statistical principle that lets insurance work: any single claim is unpredictable, but across a large, diverse pool of policies the average claim cost becomes highly predictable. The bigger and more varied the pool, the more reliable the pricing.See all terms in the glossary. Any single claim is random and unknowable. But across a large, diverse pool of policies, the average claim cost becomes remarkably predictable — and a predictable average is something you can price. Insurance is not a bet on one person; it is the mathematics of a crowd.

Two forces then push on the actuary's clean number. Competition and the cycle drag the actual charged price up or down (in a soft market insurers cut below the technically correct rate to win share). And medical or repair inflation pushes the expected claim cost up over time, which is why insurers reprice — as Star Health repriced roughly 65% of its retail health book after its loss ratio crept up.

What Happens If Too Few People Buy In A Year

Because pricing depends on the law of large numbers, volume itself is part of the product. A weak sales year does real damage, but rarely the kind that kills an insurer outright.

Watch Out:

When too few policies are sold, three things go wrong at once. The pool shrinks and becomes unpredictable — with too few policies the averages stop holding, so a single large claim can wreck the year. Fixed costs spread over less premium, so the expense ratio rises and the combined ratio worsens. And the float grows slowly or shrinks, weakening the second profit engine — investing the float.

But here is the crucial cushion, and it is why an insurer is more resilient than a normal company: the in-force book keeps working. A life insurer's existing 20-year policies keep paying premiums and generating float for decades; a P&C insurer's policies renew every year. So a poor sales year mostly hits new-business value (VNB) and growth — not survival. The embedded value already on the books carries the company through.

Key Point: The question that matters is why sales are low. Selling less on purpose in a soft market because prices are inadequate is discipline — a good sign. Losing business because you are uncompetitive, or being left with only the risks nobody else wanted (adverse selectionAdverse SelectionThe danger that the customers most eager to buy insurance are the riskiest ones. If an insurer prices poorly or loses its good customers, it can be left with a pool of bad risks — claims rise, losses mount, and the book spirals.See all terms in the glossary), is the dangerous version — that is how an insurer quietly spirals: bad risks in, claims up, prices up, good customers leave, worse risks remain.

Volume Is Product

The tell is always why the number moved: a falling sales figure is only alarming once you know whether the company chose it or suffered it — the theme of the underwriting cycle we turn to next.

Why Insurance Is Cyclical: The Underwriting Cycle

Insurance pricing swings in long, punishing cycles that trap undisciplined companies. Understanding the underwriting cycleUnderwriting CycleThe multi-year swing in insurance pricing. When capital is plentiful, insurers compete and cut rates ('soft market'); after big losses drain capital, rates spike ('hard market'). Disciplined insurers write less business in soft markets and more in hard ones.See all terms in the glossary is how you separate a genuinely great insurer from one that simply caught a good year.

The underwriting cycle

Capital floods in when times are good and drains out after disasters. Rates follow capital, not logic. The best insurers do the opposite of the crowd.

Soft market

Everyone competes

Capital is plentiful, recent losses are low, and insurers slash rates to win market share. Combined ratios drift up. Disciplined insurers write less business and let rivals chase bad prices.

The shock

A catastrophe hits

A hurricane, wildfire, or pandemic drains industry capital. Weak insurers who under-priced are exposed. Some go insolvent. Capacity leaves the market overnight.

Hard market

Prices spike

With capital scarce, survivors raise rates sharply. Combined ratios fall. Disciplined insurers who kept their powder dry now write aggressively at excellent prices.

Repeat

Memory fades

High profits attract fresh capital, competition returns, and the cycle begins again. The tell of a great insurer is that it grows premiums in hard markets and shrinks them in soft ones.

In annual reports, compare premium growth against the rate environment. Double-digit growth in a soft market is one of the most common preludes to reserve trouble later.

This is why a single year's combined ratio can mislead. An insurer posting 90% in a soft market is doing something special; one posting 90% in a hard market may just be riding the tide. Always read the combined ratio against where the cycle is.

Part 1 — Property & Casualty (P&C): The Purest Version of the Model

P&C is the cleanest place to watch the two-engine model at work: because claims are paid fast, underwriting quality shows up quickly and the float turns over rapidly. The US market offers three masterclasses in how an insurer actually wins — one on cost, one on data, one on discipline.

The US masterclass: GEICO's cost moat

Cost Moat

GEICO's entire strategy is a single, ruthless idea: be the lowest-cost provider of car insurance and pass some of that saving to customers to win volume. It sells direct — no agents — which structurally strips out a layer of cost.

The result in 2024 was extraordinary. GEICO posted a combined ratio of 81.5% — its best result for any 12-month period this century — with an expense ratioExpense RatioThe share of premium an insurer spends running the business — commissions, salaries, marketing, technology. Loss ratio plus expense ratio equals the combined ratio. A low expense ratio (GEICO's is under 10%) is a durable cost moat.See all terms in the glossary of just 9.7% and a loss ratioLoss RatioThe share of earned premium an insurer pays out as claims (and claim-handling costs). A 70% loss ratio means 70 paise of every premium rupee went to policyholders' claims. It is the single biggest driver of whether underwriting is profitable.See all terms in the glossary of 71.8%. That produced roughly $7.81 billion of pre-tax underwriting profit, more than double the prior year.

Why This Matters: A sub-10% expense ratio is a genuine, durable moat. A rival cannot simply decide to have GEICO's cost structure — it would have to dismantle its agent network, rebuild its brand around direct sales, and match GEICO's scale in data and advertising. That takes a decade and billions. Low cost, once achieved at scale, is one of the few moats that actually widens over time.

Progressive: the data and segmentation machine

Progressive competes on a different edge: it prices risk more precisely than almost anyone. Its early bet on telematics (usage-based insurance) and granular customer segmentation lets it charge exactly the right price to each driver — winning the good risks and shedding the bad ones.

The scoreboard is the proof. In 2025 Progressive grew net premiums written by 12% and policies in force by 10%, all while holding a companywide combined ratio of 87.4%. Over five years it added nearly 13.9 million policies in force — profitable growth, the rarest combination in insurance.

GEICO

Edge: lowest cost, direct-to-consumer.

Weapon: ~9.7% expense ratio.

Risk: slower to modernise pricing tech; growth lagged in recent years.

Progressive

Edge: superior risk segmentation and telematics.

Weapon: price the good driver correctly, decline the bad one.

Risk: premium valuation leaves little room for error.

GEICO

Edge: lowest cost, direct-to-consumer.

Weapon: ~9.7% expense ratio.

Risk: slower to modernise pricing tech; growth lagged in recent years.

Progressive

Edge: superior risk segmentation and telematics.

Weapon: price the good driver correctly, decline the bad one.

Risk: premium valuation leaves little room for error.

Both are excellent. But note how they win is different — and a moat you can name (cost, or data) is worth more than a vague sense that a company is "well run."

Chubb: disciplined commercial underwriting at scale

Where GEICO and Progressive dominate personal auto, Chubb is the world-class commercial and specialty P&C underwriter, led by Evan Greenberg. Its edge is underwriting discipline — the willingness to walk away from underpriced risk.

In FY2024 Chubb delivered a P&C combined ratio of 86.6% on consolidated net premiums written of $51.5 billion (up 8.7%), with record net income of $9.27 billion. Book value per share reached about $159.77.

Tip: When you read a commercial insurer, the phrase to hunt for in the annual report is some version of "we did not chase business at inadequate rates." Chubb says it constantly. An insurer that brags about volume growth in a soft market is telling you something worrying; one that brags about discipline is telling you something reassuring.

Berkshire: where P&C becomes a compounding machine

Berkshire Hathaway is the ultimate expression of the model. Its insurers (GEICO, Berkshire Hathaway Reinsurance, General Re, and others) held roughly $171 billion of float at the end of 2024, and the insurance group generated around $9 billion of underwriting profit that year.

Read that again: Berkshire was paid ~$9 billion to hold $171 billion of investable money. That float funds the equity portfolio (Apple, Coca-Cola, American Express) and the wholly-owned businesses (BNSF Railway, Berkshire Hathaway Energy). The permanent capitalPermanent CapitalMoney a firm can invest with no fixed deadline to return it. Unlike a fund that must repay investors after ~10 years, permanent capital lets the holder buy and hold indefinitely and never become a forced seller in a downturn.See all terms in the glossary that float provides is why Berkshire is never a forced seller in a crash — it is usually the buyer. We cover this machine in depth in the Berkshire Hathaway analysis.

Part 2 — India's General Insurance: Float That Costs a Little, But Still Works

Indian general insurers look worse on the combined ratio than their US peers, and beginners often misread this as weakness. It is mostly a feature of the market, not a flaw in the company.

ICICI Lombard: the private general insurance leader

ICICI Lombard is India's largest private general insurer, with about a 9.0% market share of gross direct premium in FY25. It wrote ₹26,833 crore of gross direct premium (up 8.3%, ahead of the industry) and ran a combined ratio of 102.8%, a slight improvement on FY24's 103.3%.

Indian motor and health lines carry structurally thin margins — a big reason general insurers here sit near 100% while US auto sits comfortably in the 80s.

So why is ICICI Lombard a strong business despite a sub-par combined ratio? Its float — invested largely in Indian government bonds and corporate debt yielding 7%+ — throws off investment income that comfortably exceeds the small underwriting loss, so it still earns a healthy return on equity while growing premiums faster than the market. Its strength in commercial lines (roughly 13% share in fire, 17% in engineering, 21% in marine cargo, 19% in liability) gives it pricing data and scale advantages that the retail-heavy players lack.

Star Health: the retail health specialist

Star Health is the largest standalone health insurer in India, with about a 33% share of the retail health market. Health insurance is a harder game than auto: medical inflation is relentless, claims are frequent, and mispricing shows up fast.

That pressure is visible in the numbers. Star's loss ratioLoss RatioThe share of earned premium an insurer pays out as claims (and claim-handling costs). A 70% loss ratio means 70 paise of every premium rupee went to policyholders' claims. It is the single biggest driver of whether underwriting is profitable.See all terms in the glossary rose to 69.8% in FY25 (from 66.5% in FY24), pushing its combined ratio to about 101.1%. Management responded exactly as a disciplined underwriter should — repricing roughly 65% of its retail portfolio and tightening its mix — and the combined ratio improved to about 98.8% in FY26, with profit after tax rising 16% to ₹911 crore.

Watch Out: Health insurance carries a structural risk that auto does not: medical inflation can outrun your ability to reprice. If claims rise faster than premiums, the loss ratio drifts up quietly for years before management is forced to hike prices — and every hike risks losing the healthiest, most price-sensitive customers, leaving a sicker pool behind. When you analyse a health insurer, the loss-ratio trend matters more than any single year.

Part 3 — Life Insurance: A Completely Different Animal

Here is where most investors get lost. Life insurance uses the same word — "insurance" — but it is a fundamentally different business, and its accounting actively hides its economics.

The problem: a life insurer sells a 20-year policy today. It incurs all the selling cost upfront, collects premiums for two decades, and pays out much later. Standard accounting recognises the cost now and the profit slowly, so a fast-growing, highly profitable life insurer can look like it is barely making money. Reported profit and book value are almost useless for judging it.

The industry solved this with two purpose-built metrics.

How life insurers are actually measured

Embedded Value

Stock of value

Net worth + PV of profits from existing policies

VNB

Flow of value

PV of future profit from THIS year's new policies

Persistency

Quality of book

% of customers who keep paying, not lapse

Embedded Value (EV) is the life insurer's true economic net worth: its net assets plus the present value of all future profits already locked into policies it has sold. Think of it as the liquidation-plus-runoff value. This is the anchor you value a life insurer against — not book value.

Value of New Business (VNB) is the present value of all future profit expected from the policies sold this year. The VNB margin (VNB ÷ new business premium) tells you how profitable that new business is. Rising VNB is the single clearest sign a life insurer is compounding value.

Persistency is the percentage of policyholders who keep paying rather than lapsing. It matters because the future profits baked into EV only materialise if customers stay. The 13th-month and 61st-month persistency ratios reveal whether the book is real or a mirage.

Why This Matters: Book value and reported P/E are the wrong tools for a life insurer. A life insurer growing VNB at 15% with 87% persistency is compounding intrinsic value fast even if reported profit looks flat. Value the stock of embedded value, judge the flow of VNB, and stress-test both with persistency.

India life: HDFC Life vs LIC — same market, opposite businesses

APEAPE (Annual Premium Equivalent)A standardised measure of new-business volume: 100% of regular annual premiums plus 10% of one-time single premiums. It lets you compare life insurers whose product mixes differ, and is the base on which VNB margin is calculated.See all terms in the glossary — annual premium equivalent — is the standardised measure of new-business volume that lets you compare insurers with different product mixes. On top of it, the two Indian giants tell completely different stories.

HDFC Life (private)

Model: high-margin, profit-focused private insurer.

VNB margin FY25: ~25% — among the best.

Growth: individual APE up ~18%, EV up ~13%.

Persistency: 13th-month ~87%, improving long-term.

Market view: premium valuation (~7x book, well above EV).

LIC (state-owned)

Model: vast scale, market leader, lower margin.

VNB margin FY25: ~17.6% — rising but still modest.

Scale: ~57% overall FYPI market share; EV ~₹7.77 lakh crore.

Shift: non-par APE up ~50% — mix improving.

Market view: trades below its embedded value.

HDFC Life (private)

Model: high-margin, profit-focused private insurer.

VNB margin FY25: ~25% — among the best.

Growth: individual APE up ~18%, EV up ~13%.

Persistency: 13th-month ~87%, improving long-term.

Market view: premium valuation (~7x book, well above EV).

LIC (state-owned)

Model: vast scale, market leader, lower margin.

VNB margin FY25: ~17.6% — rising but still modest.

Scale: ~57% overall FYPI market share; EV ~₹7.77 lakh crore.

Shift: non-par APE up ~50% — mix improving.

Market view: trades below its embedded value.

This contrast is the entire lesson of life-insurance investing. LIC is astonishingly large — an embedded value of roughly ₹7.77 lakh crore and 57% market share — yet the market values it at less than its EV, because investors doubt its ability to grow high-margin business and worry about government influence. HDFC Life is a fraction of LIC's size but commands a premium far above its EV, because its VNB margin (~25%) and growth persuade the market that each new year adds a lot of value.

Two insurers. Same country. One trades below its net worth, the other at several times it. The difference is almost entirely the market's view of future VNB growth.

The Moats in Insurance: What Actually Protects Profits

Insurance is a commodity in disguise — anyone can promise to pay a claim. So durable advantage comes from a short list of real moats. When you evaluate any insurer, score it against these.

The five insurance moats

1. Cost advantage — the lowest expense ratio wins over time (GEICO's ~9.7%). Structural, hard to copy.

2. Underwriting data & discipline — better risk selection means lower loss ratios (Progressive's segmentation). Improves with scale.

3. Distribution — owning the customer relationship (bancassurance for HDFC Life, LIC's million-agent army) lowers acquisition cost.

4. Brand & trust — people buy protection from names they believe will pay. Trust compounds; it is why LIC still dominates.

5. Float scale & permanence — a large, low-cost, long-duration float (Berkshire, life insurers) is investable capital that never gets called.

The best insurers stack two or more of these. GEICO has cost plus scale plus brand. HDFC Life has distribution plus brand plus a high-margin book. A single-moat insurer is far more fragile than it looks.

How To Value An Insurer

Because P&C and life are different businesses, they use different valuation frameworks. Here is the practical toolkit, with worked numbers. Treat every figure as illustrative — always pull the current price and latest filings before you act.

Valuing a P&C insurer: Price-to-Book anchored to ROE

For a P&C insurer, book value is meaningful (claims are paid fast, so the balance sheet is roughly honest). The right lens is price-to-bookBook ValueA company's assets minus its liabilities, as recorded on the balance sheet — its net worth on paper. For most of Berkshire's history, growth in book value per share was Buffett's headline measure of value creation.See all terms in the glossary relative to return on equity. The intuition is simple:

An insurer earning a high, sustainable ROE deserves a high price-to-book. One earning a low ROE deserves a low one. A useful shorthand for the justified multiple is:

Justified P/B ≈ (ROE − g) ÷ (r − g)

where g is sustainable growth and r is the cost of equity. The higher the ROE relative to the cost of equity, the more the stock should trade above book.

Worked example — Chubb. Chubb earns an ROE around 15%. Assume a cost of equity of ~9% and long-run growth of ~5%. Justified P/B ≈ (0.15 − 0.05) ÷ (0.09 − 0.05) = 2.5x. Chubb actually trades near ~1.8x book — arguably undemanding for a world-class underwriter, which is part of the bull case for it.

Worked example — Progressive. Progressive earns a much higher ROE (mid-20s%), so the same formula justifies a far higher multiple — and indeed it trades near ~4x book. The premium is not irrational; it is the market paying up for superior, sustained profitability and growth. The risk is that at ~4x book, any slip in ROE compresses the multiple hard.

Key Point: The mistake beginners make is calling Progressive "expensive" at 4x book and Chubb "cheap" at 1.8x. Neither is meaningful without the ROE. A P&C insurer's price-to-book is only interpretable next to its return on equity. High ROE justifies high P/B; the question is always whether that ROE is sustainable through the cycle.

Valuing a life insurer: Embedded Value + Appraisal Value

For a life insurer, book value is useless and EV is the anchor. The market values a life insurer at:

Appraisal Value = Embedded Value + Value of future new business

The future-new-business piece is usually expressed as a multiple of this year's VNB (a "structural value" or "new-business multiplier"). So the market price implies a Price-to-EV multiple greater than, equal to, or less than 1.0 depending on how much future VNB growth investors expect.

Worked example — LIC vs HDFC Life. LIC's embedded value is roughly ₹7.77 lakh crore, yet its market capitalisation has often sat below that — a Price-to-EV under 1.0x. The market is effectively saying "the existing book is worth its EV, but we assign little value to future new business" because LIC's VNB margin (~17.6%) and growth are modest. HDFC Life, by contrast, trades at a large premium to its EV — a Price-to-EV of roughly 2–3x — because its ~25% VNB margin and faster growth mean each future year is expected to add substantial value.

When a life insurer trades below 1x EV, ask: is the market right that new business is worth little, or is it mispricing a turnaround (rising VNB margin, better mix)?

When it trades well above 1x EV, ask: is the implied VNB growth realistic, or is the premium pricing in a decade of flawless execution?

The gap between price and EV is the market's forecast of future VNB. Your job is to decide if that forecast is too optimistic or too pessimistic.

The Investor's Checklist

Before you buy any insurer, run it through this. It works for a US auto insurer, an Indian health insurer, and a life giant alike.

The six-question insurance checklist

1. Combined ratio — and where in the cycle? Below 100% is underwriting profit. Read it against soft vs hard market conditions.

2. What does the float cost? Negative (paid to hold) is world-class. Slightly positive is fine if investment income covers it.

3. Is growth profitable or bought? Premium growth with a stable or falling combined ratio is real. Growth with a rising ratio is a warning.

4. Life insurers: is VNB and VNB margin rising? And is persistency healthy — 13th-month above ~85%?

5. Which moats does it own? Name them: cost, data, distribution, brand, float scale. Vague "quality" is not a moat.

6. Is the valuation anchored correctly? P/B-vs-ROE for P&C; Price-to-EV and implied VNB growth for life. Never a bare P/E.

Red Flags: What Makes Insurance Dangerous

Insurance has destroyed more investors than almost any industry, because the losses are hidden until they are catastrophic. Watch for these.

Watch Out:

Reserve under-provisioning. An insurer sets aside "reserves" for claims it expects to pay. Under-reserving inflates today's profit and blows up later when claims come due. Falling reserve adequacy is the classic accounting time-bomb. Chasing growth in a soft market: rapid premium growth while rivals shrink usually means winning business everyone else correctly declined. Long-tail liability creep: in liability or reinsurance, claims can surface decades later (asbestos is the textbook horror story), so a cheap-looking long-tail insurer may be sitting on unquantified losses. Medical inflation outrunning repricing in health insurance is a slow, quiet march up the loss ratio. Investment reaching for yield: if underwriting is weak, management is tempted to take excessive risk with the float — converting an insurance problem into a solvency problem.

Handle With Care

The reason Buffett insists on underwriting discipline above all else is that the float is only an asset if the liabilities behind it are honestly priced. An insurer with cheap float and dishonest reserves is not a compounding machine — it is a leveraged bet waiting to detonate.

Key Takeaways

Insurance is the rare business that collects cash before it delivers, creating float — investable money the insurer holds, sometimes for free, sometimes at a profit.
The combined ratio tells you instantly whether underwriting is profitable: below 100% is a profit, above 100% means the float has a cost.
P&C and life are different businesses. P&C is short-tail and judged on combined ratio and ROE; life is long-tail and judged on embedded value, VNB, and persistency.
The best US P&C insurers (GEICO, Progressive, Chubb) earn underwriting profits — their float is free or better. Indian general insurers run combined ratios slightly above 100% but still earn good returns via investment income.
In life insurance, the gap between price and embedded value is the market's forecast of future VNB growth — LIC trades below EV, HDFC Life well above it, and the difference is almost entirely growth expectations.
Value P&C insurers on price-to-book anchored to ROE, and life insurers on embedded value plus appraisal value — never on a bare P/E.
The moats are cost, underwriting data and discipline, distribution, brand, and float scale. The red flags are under-reserving, growth in soft markets, and reaching for yield.

Frequently Asked Questions

What is insurance float in simple terms?

Float is the pile of premium money an insurer has collected but not yet paid out as claims. Because customers pay upfront and claims come later, the insurer holds a large, growing pool of cash in the meantime. It legally belongs to policyholders, but until claims are due the insurer can invest it and keep the returns. If the insurer's underwriting breaks even, this float is essentially free investment capital; if underwriting is profitable, the insurer is effectively paid to hold it. Float is the single biggest reason insurance can be such a powerful compounding business.

Is a combined ratio above 100% always bad?

No. A combined ratio above 100% means the insurer lost money on underwriting alone, but many profitable insurers — especially in India — run slightly above 100% and still earn strong returns because investment income on their float more than covers the small underwriting loss. What matters is the trend and the cost of float. A combined ratio drifting steadily higher is a warning; one holding stable while investment income is strong can be perfectly healthy. The world's best insurers, like GEICO, run well below 100% and are paid to hold their float.

Why can't I value a life insurer with a normal P/E ratio?

Because life-insurance accounting recognises selling costs upfront but profits slowly over the 15-to-20-year life of a policy. A fast-growing, highly profitable life insurer can therefore show flat or depressed reported earnings, making its P/E meaningless. The industry uses embedded value (net assets plus the present value of profits already locked into existing policies) as the anchor, and value of new business (VNB) to measure how profitable this year's sales are. You value a life insurer at its embedded value plus a multiple of its future new business — a framework called appraisal value.

What is the difference between P&C and life insurance for an investor?

Property & casualty (auto, home, health, commercial) is short-tail: claims are paid within months to a few years, so underwriting quality is visible fast and you judge it on the combined ratio and return on equity, valuing it on price-to-book. Life insurance is long-tail: policies run for decades, accounting hides the economics, and you judge it on embedded value, VNB margin, and persistency, valuing it on price-to-embedded-value. They share a name and the float concept, but almost nothing else about how you analyse them is the same.

Why does LIC trade below its embedded value while HDFC Life trades above?

Embedded value captures the worth of business already written. The market price also reflects expectations about future new business. LIC has enormous scale and a huge embedded value, but its VNB margin (~17.6%) and growth are modest, and investors discount government influence — so the market assigns little value to its future new business and prices it below EV. HDFC Life has a much higher VNB margin (~25%) and faster growth, so the market expects each future year to add significant value and pays a large premium over EV. The gap between price and embedded value is, in effect, the market's forecast of future VNB growth.

The One-Sentence Thesis

What You'll Learn

First, The Two Families of Insurance

Where This Model Came From: Coffee, Fire, and the Mathematics of Death

Fire creates property insurance

A coffee house invents underwriting

Maths tames death

The model crosses the Atlantic

Insurance in India: from Calcutta to liberalisation

How An Insurer Actually Makes Money: Two Engines

The Number That Tells You Everything: Combined Ratio

How Insurers Set Premiums: The Price of a Promise

What Happens If Too Few People Buy In A Year

Why Insurance Is Cyclical: The Underwriting Cycle

Everyone competes

A catastrophe hits

Prices spike

Memory fades

Part 1 — Property & Casualty (P&C): The Purest Version of the Model

The US masterclass: GEICO's cost moat

Progressive: the data and segmentation machine

Chubb: disciplined commercial underwriting at scale

Berkshire: where P&C becomes a compounding machine

Part 2 — India's General Insurance: Float That Costs a Little, But Still Works

ICICI Lombard: the private general insurance leader

Star Health: the retail health specialist

Part 3 — Life Insurance: A Completely Different Animal

India life: HDFC Life vs LIC — same market, opposite businesses

The Moats in Insurance: What Actually Protects Profits

How To Value An Insurer

Valuing a P&C insurer: Price-to-Book anchored to ROE

Valuing a life insurer: Embedded Value + Appraisal Value

The Investor's Checklist

Red Flags: What Makes Insurance Dangerous

Key Takeaways

Frequently Asked Questions

What is insurance float in simple terms?

Is a combined ratio above 100% always bad?

Why can't I value a life insurer with a normal P/E ratio?

What is the difference between P&C and life insurance for an investor?

Why does LIC trade below its embedded value while HDFC Life trades above?

What to Read Next

Related Articles

Valuation 101: When to Buy Great Companies (A Beginner's Guide)

Understanding P/E Ratios: The Complete Beginner's Guide to Stock Valuation

Semaglutide and the GLP-1 Wave: The Investment Thesis Behind Pharma's Biggest Boom