Where is the margin in the world?
Companies everywhere sell about $72tn of goods and services a year and keep roughly $5.7tn of it. That profit is not spread across the world economy in anything like proportion to its activity: it collects in a small number of sectors, in a smaller number of countries, and inside those, in a very small number of firms. This piece sizes the pool, maps where it accrues, names where it is systematically destroyed, and sets out the four structures that let a margin survive. Editorial estimates; methodology in the appendix. The companion piece, When to do margin, is about when to reach for it.
From world output to company profit
The first thing to fix is the size of the prize, because it is smaller than people expect. World GDP is about $113tn. Most of that is wages. What is left over as corporate gross operating surplus is on the order of $29tn, and most of that is consumed by depreciation, interest and tax before anyone can call it profit. The residual — what a shareholder actually owns — is roughly $10.5tn across every firm on earth, of which about $5.7tn belongs to the listed universe that this page measures.
| Layer | Size | Share of world GDP | What is taken out at this step |
|---|---|---|---|
| World GDP | $113tn | 100% | — |
| Corporate gross value added | $65tn | 58% | Government, households, non-profits, informal activity |
| Corporate gross operating surplus | $29tn | 26% | Wages and salaries — by far the largest deduction |
| After-tax profit, all firms | $10.5tn | 9% | Depreciation, interest, corporate tax |
| After-tax profit, listed firms | $5.7tn | 5% | Private companies, partnerships, unlisted state firms |
Five percent of world output. That is the whole thing every listed company on the planet is competing over, and the point of the rest of this page is that it is distributed far more unevenly than the activity that produces it.
The margin map
Profit is the product of two numbers, so it is best read on the two axes that produce it. The chart below puts reported revenue on the horizontal axis and net margin on the vertical, and sizes each bubble by the profit pool itself. Because profit is revenue multiplied by margin, every combination that earns the same money lies on a hyperbola — the dashed curves. A sector is doing well if it sits above and to the right of a curve, and the route it took to get there is irrelevant to the shareholder.
Reported revenue against net margin, bubble area proportional to the profit pool. Editorial estimates for ~2025; the rows sum to a global listed-company pool of about $5.74tn.
What the sector map shows
- Banking is the largest profit pool on earth. Roughly $1.04tn a year, more than oil and gas, and more than software. It is also the least discussed, because banking margin does not look like a price — it looks like a spread between what deposits cost and what the central bank pays. Raise policy rates and the pool inflates without anyone doing anything.
- Revenue and margin are almost inversely related. The three biggest revenue lines in the world — retail, insurance and oil — run margins of 3%, 4.5% and 9.5%. The highest-margin lines — software, semiconductors, payments, banking — are comparatively small on the top line. Scale in units and scale in profit are different businesses.
- Autos produce $4.2tn of revenue and less profit than semiconductors. That single comparison is the clearest statement of where the world has moved. One industry ships ninety million complex machines a year for about $147bn; the other ships a few hundred thousand wafers for $286bn.
- The margin usually sits one layer away from the product. Jet engines are sold near cost and earn on spare parts. Printers earn on ink, razors on blades, consoles on games, cars on financing, phones on the app store. When a sector looks unprofitable, check whether the margin has simply moved to an adjacent layer someone else owns.
- Six sectors take 52% of the pool. Banks, Oil, gas & coal, Software & internet platforms, Industrials & capital goods, Consumer staples, Insurance. Everything else in the world economy shares the rest.
Where it accrues geographically
Switch the chart to geography and the shape is starker still. The United States earns about 40% of the world's corporate profit on about 26% of its revenue. China earns about 17% on nearly the same revenue base. The two countries are running comparable volumes of commerce at margins that differ by a factor of more than two.
That gap is not a productivity gap and it is not really a technology gap either. It is a composition gap plus a policy gap. The American listed universe is concentrated in the four highest-margin structures in existence — patents, licences, networks and brands — while the Chinese one is concentrated in manufacturing, property and infrastructure, where capacity is directed by policy rather than by return on capital. Where the state sets output targets, someone will always build the marginal plant, and the marginal plant is what destroys a margin. This is the same mechanism the capital-stock piece traces from the investment side.
| Region | Revenue | Net margin | Profit pool | Share of pool |
|---|---|---|---|---|
| United States | $19.0tn | 12.1% | $2.30tn | 40% |
| China & Hong Kong | $18.0tn | 5.3% | $954bn | 17% |
| Western Europe | $14.5tn | 6.2% | $899bn | 16% |
| Rest of Asia-Pacific | $7.5tn | 7.6% | $570bn | 10% |
| Japan | $6.0tn | 7% | $420bn | 7% |
| Middle East | $2.2tn | 13.6% | $299bn | 5% |
| Canada | $1.7tn | 8.8% | $150bn | 3% |
| Latin America | $2.0tn | 5.5% | $110bn | 2% |
| Africa | $0.9tn | 5% | $45bn | 1% |
Where the world loses money
A profit pool has a mirror image, and it is analytically more useful than it looks: the places where capital is systematically destroyed tell you what a margin needs in order to survive, by showing what happens when it is missing. Note that these are not small or marginal industries. They are some of the largest and most celebrated in the world.
| Loss pool | Where | Scale |
|---|---|---|
| Chinese property developers The largest corporate loss cluster in history: Evergrande and Country Garden built against presold cash flow that stopped arriving. | China | Well over $100bn of cumulative losses and write-downs |
| Solar module manufacturing Capacity was built to roughly twice global demand; polysilicon and module prices fell below cash cost and stayed there. | Mostly China | Combined losses across the top module makers since 2023 |
| Electric-vehicle makers outside the top two More than a hundred brands chasing one market with a price war on top; only BYD and Tesla have made the volume work. | China, US, Europe | Tens of billions per year |
| Airlines, across the cycle Perishable inventory, price-transparent distribution, fixed fleet costs and a demand curve that collapses in every shock. | Global | Roughly zero cumulative profit since the industry began |
| Streaming video other than Netflix Every incumbent bought subscriber share at a loss against one profitable scale player. A land grab entered too late. | US-led, global | On the order of $25bn of cumulative operating losses |
| Pre-revenue biotech This is not a failure: the losses are the R&D cost of the patent monopolies that show up as pharma margin a decade later. | US, Europe, China | $25-30bn burned per year, structurally |
| European bulk chemicals & steel A globally priced commodity produced on a regionally priced input. Energy cost is the whole margin. | Europe | Negative operating margins through the 2022-25 energy shock |
| Ride-hailing and delivery platforms Network effects turned out to be local, not global, so the land grab had to be won city by city against subsidised rivals. | Global | Over a decade of losses before the first GAAP profit |
Read down that list and one pattern repeats: losses cluster wherever capacity can be added faster than demand grows, by someone who is not paying for it out of his own return. Subsidised solar capacity, presold property, venture-funded delivery, state-backed EV brands, flag-carrier airlines. The mechanism is not incompetence. It is that a competitor with a cheaper cost of capital than yours will build the plant you decided not to build, and then you both have to live with the price it sets.
The four things a margin can be made of
Every durable margin in the tables above resolves to one of four structures. This matters more than the sector label, because the same sector contains both: Hermes and a contract garment maker are in “apparel,” and only one of them has a margin that survives contact with a competitor.
| Source | Mechanism | Examples | How it ends |
|---|---|---|---|
| Scarcity rent | You control a physical asset that cannot be reproduced: a cheap oil field, a high-grade ore body, a right of way, a leading-edge fab. | Saudi Aramco, BHP, the North American railroads, TSMC | Ends when the resource is substituted rather than exhausted — which is how it usually ends. |
| Legal exclusion | The state grants you the right to be one of few: a patent, a banking licence, spectrum, a regulated return, a pharmaceutical approval. | Big pharma, banks, exchanges, regulated utilities, tobacco | Ends by legislation or by the expiry date written into the grant. Patent cliffs are diarised years ahead. |
| Network and switching cost | Your value to each customer rises with the number of other customers, and leaving costs them more than staying. | Card networks, marketplaces, enterprise software, app stores | Ends when the network fragments or a platform shift resets the base. Slow to build, slow to lose, then sudden. |
| Brand and manufactured scarcity | Willingness to pay is attached to meaning rather than to function, and you refuse to supply all the demand you could. | Hermes, Ferrari, luxury spirits, prestige education | Ends the moment you meet the demand. The margin is destroyed by the volume that would seem to prove it. |
What is not on that list is worth as much as what is. Being good at something is not on the list. Being large is not on the list. Being early is not on the list. Each of those can produce a margin for a while, and none of them prevents a competitor from reproducing your position — which is Michael Porter's point in the argument about operational effectiveness, and the reason a well-run company in a badly structured industry still earns 3%.
Three consequences for anyone allocating capital
- Pick the structure before the sector. The margin map is a picture of structures, not of industries. A business with a licence, a patent, a network or a real brand will out-earn a better business without one, in the same industry, for as long as the structure holds.
- Ask who is paying for the marginal unit of capacity. If the answer is a state, a subsidy programme or a venture fund, the price in that market is not being set by anyone who needs it to cover cost. No amount of operational excellence beats that.
- Follow the margin one layer up or down. The profit in a value chain is rarely where the visible product is. Find the layer with the fewest credible suppliers and the highest switching cost; that is where the pool has collected, whatever the industry is called.
Appendix — methodology
The anchors
Three numbers hold the page together: nominal world GDP of about $113tn, a global listed-company revenue base of about $72tn, and an after-tax profit pool of about $5.7tn on it. The sector rows sum to $5.74tn on $71.8tn of revenue and the region rows sum to $5.75tn; the residual between them and the anchor is rounding.
What “revenue” means, and why the denominators are not comparable
Each sector is stated on the revenue convention its own accounts use: gross written premiums for insurance, net revenue (not gross interest income) for banks, gross merchandise revenue for retailers, gross billings for distributors. These conventions are not comparable to one another — an insurer's 4.5% on premiums and a software firm's 21% on licence revenue are measuring different things. They are added anyway because published accounts offer no other common basis, and the vertical axis of the chart should be read as “margin on this industry's own revenue definition,” not as a clean cross-industry ratio. The profit column, by contrast, is comparable: a dollar of after-tax profit is a dollar everywhere.
Scope
The pool covers the listed, investable universe of roughly 9,000 companies. It therefore excludes private firms, partnerships (including the large professional-services and trading houses), co-operatives, and unlisted state-owned enterprises — a group that adds roughly another $4.8tn of profit that is not mapped here. Saudi Aramco is included because it is listed; most national oil companies are not.
Assignment rules
- Sector. Each company is assigned to one sector by its dominant profit source, not its dominant revenue. Conglomerates with two genuinely large pools — a retailer with a cloud business, a manufacturer with a captive bank — are the main source of error here, and they are large enough to move a row by several percent.
- Geography. Companies are assigned by country of domicile, not of production or of sale. The American figure therefore includes profit earned worldwide by American-domiciled firms, which is the correct basis for “who owns the margin” and the wrong basis for “where the activity happens.”
- Losses. Loss-making firms are netted inside their sector rather than excluded, which is why pharma sits at 14% rather than the ~18% big pharma alone earns: the pre-revenue biotech tail is netted against it.
Caveats
Single-year snapshots flatter whichever cycle is peaking. The energy row would have been roughly twice this size in 2022 and roughly half in 2020; banking moves with policy rates; semiconductors are in the steepest upswing in their history and this row will look wrong in either direction within two years. The structural claims — that profit is far more concentrated than activity, that the United States earns a disproportionate share of it, and that the four sources in the table above are what separate a durable margin from a temporary one — are robust to those swings. The individual rows are not.
Sources consulted
- Company filings (10-K, 20-F, annual reports). Revenue and net income for the largest firms in each sector pool.
- S&P Global / Compustat-tradition aggregates. Index-level revenue and earnings for the listed universe.
- MSCI ACWI IMI sector weights. The shape of the investable global universe by sector and domicile.
- McKinsey Global Banking Annual Review. Global banking profit after tax and return on equity.
- Swiss Re sigma. World insurance premium volume and insurer profitability.
- IEA World Energy Investment / Oil & Gas industry data. Oil and gas industry net income across the price cycle.
- IATA industry outlooks. Airline industry net margin and cumulative profitability.
- OECD / UN national accounts. Corporate gross operating surplus as a share of GDP.
- IMF World Economic Outlook. Nominal world GDP, the denominator for every share on the page.
- De Loecker, Eeckhout & Unger (2020). Long-run markup estimates showing where the rise in margin is concentrated.