What Is an Economic Moat? Margin Proves It, Story Doesn't

An economic moat is a feature of a business that keeps its return on capital above the cost of that capital for years, even though competitors keep trying to take it away. That is the definition. The problem is that almost every management team claims one, and a moat is visible in only two places in the financial statements: a gross margin that stays above the industry's, and a price list that rises without volume falling. That is proof. Everything else, from brand to "market leadership", is a hypothesis waiting to be tested.

Economic moat definition, from the source#

Warren Buffett popularised the word. His best definition sits in the 2007 letter to Berkshire Hathaway shareholders: "a truly great business must have an enduring 'moat' that protects excellent returns on invested capital". His reason is in the same paragraph: "the dynamics of capitalism guarantee that competitors will repeatedly assault any business 'castle' that is earning high returns". The castle is the profit. The moat is whatever keeps the competition on the far bank.

It is the second half of the quote that matters more. A high return on capital is not a trophy, it is a signal: it tells everyone with money and engineers that this spot pays better than the alternatives. Capital always walks towards that signal. So the moat question is concrete: what stops a competitor who can see this margin from taking it within three to five years?

The market's usual answer is brand and market share, and it is not a silly one. A known brand lowers the cost of winning a customer, and a leader with, say, 40% of the market buys cheaper from suppliers and spreads its fixed costs over more units. Investor decks lean on both, and most readers nod along.

Except that brand and share are claims, and a moat is proof, which neither of them supplies on its own.

The makers of phones with keyboards had market share, and brands too, right up to the day someone showed a phone without a keyboard. So if the answer to the moat question is "they are just good and well known", there is no moat. Being good is the garrison on the wall, not the water in the ditch.

The five types of economic moats#

Only a handful of mechanisms genuinely keep rivals out, and knowing them by name makes it harder to mistake one great quarter for a wall. Morningstar groups them into five sources, and the classification maps closely onto what you find in an annual report once you start reading it from the competition section.

Three rows need a footnote. Scale is a cost advantage only when it lowers unit cost; a billion spent on advertising buys popularity, not scale. Know-how accumulated over decades (ASML, the lithography-machine maker nobody with billions has managed to copy) has no line of its own on the balance sheet, yet it is a moat from the intangible-asset family. Regulation and licences land there too, with one caveat: the regulator who guards your door also sits at your table, so a protected position often means a capped price and rules that change after a single election.

Switching costs and network effects get a piece of their own, because they are the two mechanisms most often mistaken for something else. From the outside, switching costs and plain customer inertia look identical until a rival shows up with a better offer and a one-click migration.

Every moat has an expiry date#

The most honest type of moat is a patent, because it is the only one with the end date written on it. In 2023, its first year of US biosimilar competition, Humira, AbbVie's flagship drug, lost 32% of its global revenue, which fell to $14.4 billion. Which means a third of the product's sales disappeared in twelve months, not because the drug stopped working, but because the piece of paper that kept rivals across the water ran out.

Every other moat works the same way, just without a date in the calendar. A brand ages with its customer, a switching cost shrinks the day someone writes a migration tool, a cost advantage disappears when a new technology resets the whole cost curve, and a network effect breaks when the same users move to a new network. So "does the company have a moat" is a worse question than "which way is its moat moving". You measure the water level every year, not once.

Gross margin gives the moat away#

A moat means a company can hold its price higher, or its cost lower, than a rival, and keep doing it for years. Gross margin, revenue minus the cost of making the product, is the first line in the accounts where both show up at once, before selling, research and overhead costs blur the gap. A real moat leaves a trace there. No trace, and the moat lives in the slide deck.

Say company A has posted a 60% gross margin for ten years while its three largest rivals sit in a 30–35% band. Which means A keeps 60 cents of every sales dollar for everything beyond making the product, the rival keeps 33, and the 27 cents in between is rent the market pays for something it cannot copy. Capital had a decade to take that rent and did not. How much margin any industry lets a company hold is settled by that industry's own ceiling rather than by a universal threshold. That is proof, as far as a set of accounts proves anything.

The number needs a comparator, or it says nothing. On its own, 60% sounds good and means nothing: in software it would be middling, in wholesale distribution a miracle. The readings that work are the comparative one, against rivals in the same industry, and the historical one, against the company's own decade. A 60% margin in an industry where everyone earns 60% is a property of the industry. A 60% margin that was 45% five years ago is a good run, and becomes a moat only if it survives the first year a rival cuts prices.

Gross margin is the first trace, not the verdict. Buffett wrote about return on invested capital (ROIC), and ROIC is the judge at the end: a moat exists when that return stays above the cost of capital for years. Between the two sit selling, research and overhead costs, plus the capital tied up in plants and inventory. Say a company with a 60% gross margin spends 45 cents of every dollar on sales and marketing because its customers have to be won again every year. The rent from the moat goes to the sales force, not the owners, and almost none of it reaches ROIC.

The trap is that gross margin comes off the income statement, so it bends to the same accounting choices as net profit. Move part of the production cost into selling expenses and gross margin rises without a single price going up. That is why gross margin gets read next to the path of one dollar through the whole company: if 27 cents of advantage at the gross level melts to three cents at the operating level, the moat is in the bookkeeping, not the business.

The price-list test: price up, volume flat#

Margin says a moat exists. The price list says who owns it. A company that raises prices 5% and ships the same number of units has pricing power: the customer has nowhere else to go, or no wish to go. A company that raises prices 5% and loses 5% of its volume has a price list the market writes, and what the market gives it takes back in the first soft year.

The test can be run from the filings, because management splits revenue growth into price and volume in its discussion of results, sometimes in a sentence, sometimes in a segment table. Say revenue grew 8%, six points of it price and two of it volume. Which means the company raised its prices and the customers stayed, and even bought a little more. Turn it around, 8% growth from volume alone at a flat price in a year when costs rose 5%, and the company is buying growth with its own margin. A full protocol for the reading, promotions and inventory clearances included, is in the pricing power test.

A third pattern is the most telling: the company cuts prices to hold its share. Say volume stands still, price falls 4%, and management calls it "investing in our market position". A rival has just crossed the moat dry, and the company is spending its own margin to hold the line. A moat that has to be topped up with discounts is a puddle with a marketing budget.

Claim, evidence, counterquestion#

Management describes its own moat in marketing language and usually believes every word. So every sentence about a company's competitive advantage gets broken into three parts: the claim, the number that would confirm it, and the question that would knock it over.

"Customers rarely leave." Evidence: retention rate, contract renewals, depth of integration. Counterquestion: will a customer whose budget gets cut stay, or settle for a cheaper substitute?

"The brand is strong" gets its evidence from the price premium over the unbranded product and the share of repeat customers, and the counterquestion is: is the brand ageing along with its customer?

"Scale lowers our costs." Evidence: unit cost falling with volume, visible in gross margin. Counterquestion: could a new technology zero out that curve, the way digital photography zeroed out scale in film?

The pattern is the same each time: a claim with no evidence in the filings is a hypothesis, and a hypothesis with a named counterquestion is already analysis. A moat you cannot describe a way of draining has not been researched. It has been admired.

A wall around an emptying town#

A moat tells you how well the profit is defended, and nothing about whether that profit is growing, flat or quietly shrinking. That part depends on where the industry is in its life, not on the height of the wall.

I paid tuition for this one. Early on I found a company that looked cheap: a low multiple, a "market is overdoing the pessimism" story, a tidy business with a real cost advantage. What I did not check was where its industry was going. It was a Kodak-shaped company, well run and well defended, in a market the world was walking out of. Its moat was real. Its town was emptying, and the wall guarded nothing but the exit.

The positive version runs like this. Say a maker of cable harnesses gets its product qualified for a server rack built for AI workloads. Switching supplier risks an outage in equipment worth tens of millions, so the customer stays for the life of the rack. That is a switching cost in its purest form. One difference from Kodak, and it decides everything: the market this moat guards is growing. Same wall, different town.

Hence the order, and there is only one: industry first, moat second, valuation last. Excel won't save you from a bad industry, and a moat won't save you from an industry that is ending. Valuation comes last because a cheap company in a dying town is cheap for a reason the multiple shows and does not explain.

Wide moat vs narrow moat#

The size of a moat is measured in years, not in how impressive it looks. Morningstar assigns a wide moat to a company whose advantage should last more than 20 years, and a narrow one to an advantage that should hold for about 10. Everyone else has no moat. Those thresholds are conventions, but the question points the right way: "how many years of margin above the industry can I see ahead", not "how good does this quarter look". The rating itself is an analyst's judgement, not a measurement, so it reads like a management sentence: a claim that still needs evidence from margin and the price list.

The third category, the one the classification leaves out, is the largest: the puddle. A first-mover advantage, a popular product nobody has copied yet, a lead in a category that is only just forming. In a good year for the market, a puddle and a moat glitter in the sun exactly alike. The difference shows in the first year a competitor cuts prices, and then gross margin says in one line what the presentations hid for three years.

I do not know how many software moats will survive five years of models that write and migrate code on the customer's behalf, because a switching cost built on "migration is too expensive" is getting cheaper. I do know where I will check: in gross margin and in the price-and-volume split of revenue, report after report, not in a presentation about "leadership in the AI era".

Where the evidence lives in the annual report#

For US-listed companies the annual report, Form 10-K, sits free in the SEC's EDGAR database. Two sections do most of the work, and they have to be read together:

1. Competition. The company names its rivals and says what sets it apart from them. Check in their reports whether they tell the same story about the market. When they do not, someone is wrong, and that is the first place to dig.
2. Risk Factors. Here the company itself writes where the moat could leak: a patent expiring in a given year, the largest customer accounting for a third of revenue, a regulator considering a fee cap. You read the moat together with the risks, because whatever is crumbling the wall sits in the filing right next to the wall.

Those two sections say nothing about competitors the company would rather not name. Then the map has to be built by hand, from the customer upwards. The earnings-call transcript adds a third signal. A management team that names its threats usually knows what it is doing. A team that waves off the competition question with "strong momentum" usually cannot see that someone is already tunnelling under the wall.

What to do with this in the coming week: pick one company you believe is well defended, and build a five-year gross-margin table for it and the three competitors named in its Competition section. Then find the price-and-volume split of revenue growth for the last three years, and read Risk Factors marking every sentence that says how this moat could run dry. If the table, the price list and the risks tell the same story as the presentation, you have a moat. If not, you have a presentation.

From the finished report an Investment thesis is built, and the moat goes into it as one assumption with a measurable condition, for instance "gross margin does not fall below 55% for two consecutive quarters". When the condition breaks, Monitoring says so after the results, before you have had time to read that "fundamentals remain strong".

The moat in the Full report#

That is exactly the reading the Moat & competition section of the Full report does: it names the mechanism and checks it against margin and the price list. Right next to it, in Key risks, sits whatever could drain that moat, with a citation from the filing beside every claim, so you can click through instead of taking the report's word for it. In the Full report pro, three models write that same section independently and a judge keeps only what can be grounded in a document; that is the Three models and a judge mechanism.

To see what that section looks like on a real company, open the sample reports.