Over the long term, it’s hard for a stock to earn a much better return than the business which underlies it earns. If a business earns eighteen percent on capital over twenty or thirty years, even if you pay an expensive-looking price, you’ll end up with one hell of a result.
— Charlie Munger
Download the Slides Deck: Business_Economic_Duration
Listen to the deep-dive discussion – Why Spreadsheets Miss Business Duration
A bond has a duration: a measure of how long its money takes to come back, and how sensitive its value is to time. A business has one too, though no one prints it on a page. Call it the economic duration of the company — how long the enterprise will go on producing, and how much of its worth lies in the near future rather than the far. Munger’s eighteen-percent business held for thirty years is a long-duration asset; a business that collects this year and builds nothing is a short one. Most of investing is, in the end, an attempt to read this single quantity before a single ratio is opened.
It is not found in any statement. It is read from the shape of the business — from two things the company does with its capital: how fast it converts the cash it earns into a base that will earn more, and how long it can keep absorbing capital into that base before the base is full. Fast and long together make a company’s duration. This is not the equity duration the textbooks compute from a discounted-cash-flow model, which measures only where a company’s projected cash flows fall in time. It is a structural reading, not a calculation: a way of seeing how a business produces its flows, not merely when they arrive. What follows is a way of seeing where a company sits, built from a handful of questions asked in order, each of which sorts and frames the next.
I. The Toll and the Wave
Every business is a flow of capital between two reservoirs. The A is the toll — the cash collected now at a structural chokepoint; short duration, cash in the present. The B is the wave — the growth that makes tomorrow’s toll larger than today’s; long duration, cash in the future. Almost every business carries both in some proportion, and the proportion is its dual duration. The A is protected, or it is not: a strong toll sits behind a moat — a brand, a network, switching costs, scale — and collects reliably year after year; a weak toll is exposed and must be re-won. The moat decides whether the toll holds or leaks.
What matters most is not that both exist but the direction in which capital flows between them. In a business of quality, the A absorbs the B: the toll generates more than the growth consumes, the base is fed from within, and the surplus overflows to owners. In a deteriorating one, the direction reverses — the B absorbs the A, costly growth devouring the toll faster than the toll can refill it, the company consuming its own surplus and, at the limit, approaching the barrier where the capital runs out. Conversion is this flow from short into long; its direction decides whether value is created or destroyed, and its speed and duration decide how much.
II. Not All Growth Is the Same Price
The B comes in two kinds, and the difference is the difference between an ordinary compounder and a rare one. A mechanical B is growth that must be bought — funded out of the toll, steered by management, paid for at full price. A free B arrives from outside, carried by a secular wave, at no cost and requiring no decision. In the notation of this collection, the mechanical business is an A plus a B — two things added, the second purchased; the free-growth business is an A times a B — the toll and the wave fused, multiplying at zero marginal cost.
Union Pacific is the mechanical case in its purest form. It cannot carry more freight without laying more track and buying more locomotives; each dollar of new revenue demands a dollar of new asset. The growth is real, the moat is wide, but every increment of the base is bought at market price, and the return on that new capital drifts toward its cost. Visa is the other kind. It carries a new fintech’s transactions across rails already laid and already paid for; almost nothing is added to the balance sheet, so the numerator rises while the denominator stands still. The digitization of payments is a wave Visa did not build and does not fund — it simply collects on it, because the transactions run through its chokepoint rather than around it.
That is the test for a free B: is the wave captured by the moat? The growth must pass through the toll, not beside it. Visa and Mastercard capture the move to digital payments because the payments run on their networks; S&P Global and Moody’s capture the growth in debt because new issuance must pass through their ratings; MSCI captures the migration to passive because the assets track its benchmarks. These are the free-growth tolls, and their growth asks nothing of management — there is no allocation to get right, no runway to judge, no acquisition to price, because the growth was never the company’s to steer. Google shows the imperfect case: its search advertising is captured by the moat and rides in for free, but its cloud and AI growth is contested, fought for on open ground against Amazon and Microsoft, won only by spending. Part of the wave passes through the moat; part must be bought. Capture by the moat is the test, and it is rarely total.
III. What Kind of A, Before Any B
Before asking what kind of B a business has, there is a prior question about the A itself, and it sorts faster than anything else: is the toll recurring or one-off? A recurring toll arrives on its own — Visa’s fraction of every transaction, the railroad’s levy on freight that must move, the brand’s earning on every bottle sold. It is collected year after year without being re-won, and it is the foundation on which any B can be built. A one-off toll must be regained project by project: the consulting mandate that ends, the legal file that closes, the deal that must be replaced by the next. Its cash is not secure even in the present, and it rests on the continued presence of particular people.
This distinction decides much before the second question is asked, because a one-off toll is a short-duration asset by nature, whatever it earns. A law firm or a solo practice is a recurring need met by a non-recurring toll: the demand returns, but the cash must be won again each time, and nothing accumulates into a base that survives the people who produce it. Such a business cannot convert internally — there is no durable base to build from a flow that must be perpetually reconstituted. Its economic duration is short not because its market is small but because its toll is structurally impermanent.
And there is a trap that wears the costume of growth. A business with a one-off toll can reinvest its cash into buying more one-off tolls — a consulting firm acquiring other consulting firms, a practice adding partners and associates. This reinvests without converting. It grows the short rather than building the long: more billed hours, not a base that compounds. Accenture is the large-scale instance — a firm whose record revenues and bookings are an afflux of projects that must each be re-won, whose growth is bought by acquisition, and whose return on that purchased growth drifts toward its cost. The reinvestment is real; the conversion is not. The A is poured back into more A, and the economic duration does not lengthen, because stacking short upon short never produces long.
IV. Where the Conversion Stands
If a B exists and the conversion runs, it runs along a curve, and where a company sits on that curve is the last reading of its duration. Early, the base is small against its opportunity and absorbs everything the toll throws off; the conversion runs at full intensity and nothing overflows. As the base fills, each additional dollar finds less to do; the conversion slows, and cash begins to arrive faster than the base can absorb it. The company overflows — returning the excess in dividends and buybacks, not from generosity but because the base is full. This overflow is not a policy chosen; it is the visible sign that the absorption has reached its limit.
Coca-Cola is the pure case of a toll that has run its curve. It once converted at full intensity, building a global base for decades; now that base is full, it converts slowly and overflows the rest — a magnificent recurring toll that has become an instrument for returning cash. See’s Candies is the same shape in miniature: a superb A, a tiny B, overflowing enormously and compounding slowly. Nothing is wrong with either. A recurring toll whose base has filled is a short-duration asset again, not by failure but by maturity — and the owner, receiving the overflow, becomes the one who must now convert it, redeploying elsewhere what the business can no longer absorb.
But a company at the end of its first curve can do something other than overflow: it can spend the surplus to build an entirely new base, beginning a second conversion where the first has ended. This is the dual-duration company that contains both durations internally — Alphabet pouring the free cash of Search into cloud and artificial intelligence, Microsoft spending the overflow of software to build the capital-heavy base of a new era. The light toll builds its railroad. It is a wager, and the framework states its condition precisely: the new base must absorb capital at a return above its cost, or the spending is not investment but consumption — the B beginning to absorb the A. Where a mature toll could have overflowed into the owner’s hands, it reaches instead to regenerate its own duration, and the question is only whether the new base genuinely absorbs, or merely appears to.
V. What a Standard DCF Hides
The finance that measures duration with a number does so from a discounted-cash-flow model: it projects the cash a company will produce and weighs each year by when it arrives. A business whose value lies in distant flows is long; one whose value is near is short. The calculation is real and useful, and it does, in a buried way, contain conversion and absorption — they are dissolved into the growth rate and the return on reinvested capital the model assumes. But they are dissolved, not named: the number carries them agglomerated into a single trajectory of flows, and never separates how fast a company converts from how long it can absorb.
This is why a single duration figure can hide two opposite businesses. Imagine two companies a standard model would call long-duration, their value sitting in flows years out. One is Visa: a free-growth toll whose wave is captured by its moat, earning a return on capital near thirty percent. The other is a fast-expanding consulting firm, its revenue and bookings climbing, a model projecting that climb far into the future. On the page, their durations may nearly match. Structurally, they are opposites. Visa earns so much on so little capital that it cannot reinvest most of what it makes — its base absorbs only a fraction at that rate — so it overflows, paying a fifth of its cash flow in dividends and returning the rest in buybacks while it grows at ten to twelve percent on the wave. Its overflow is the overflow of wealth: more return than its base can hold. The consulting firm overflows nothing. It reinvests its cash into buying more firms, stacking A upon A — more billed hours, no base that compounds — and the return on that purchased growth drifts toward its cost. Its growth is the growth of poverty: motion that produces no durable base.
A discounted-cash-flow model sees two rising streams and may assign them the same duration. It cannot see that one rises by an excess of return the business cannot absorb, and the other by stacking short-duration tolls that never become long. The difference — free conversion overflowing, against mechanical reinvestment eroding — is exactly what the number agglomerates and loses. It is qualitative and quantitative at once: the components can be measured, but the nature that distinguishes them cannot be reduced to a single figure without destroying the distinction. This is why the economic duration of a business is read, not computed. A number that treats a Freesurfer and a stacker as the same asset is precisely wrong; a reading that calls one a free-growth toll that overflows and the other a one-off toll that stacks is approximately right — and in investing, as Munger and Buffett have long held, it is far better to be approximately right than precisely wrong.
VI. Reading the Duration
The readings compose into one judgment. What kind of toll — recurring or one-off, the question that sorts fastest. Whether it can become a base, and how — not at all, as with the one-off toll that only stacks; mechanically and dearly, as with the railroad buying its track; or freely, as with the toll whose wave is captured by its moat. And where the conversion stands on its curve — climbing, overflowing, or wagering on a second base. Three questions, asked in order, place a company on the spectrum that runs from the magnificent toll with no growth, through the free-growth Freesurfer, to the mechanical compounder that buys its base — and the position on that spectrum is the economic duration.
And the reading turns, at the last, toward the one doing it. A short duration is not a flaw; it is a choice with a consequence. A magnificent toll that overflows — Coca-Cola, See’s — hands its owner cash in the present, which is exactly what some owners want. But the overflow becomes the owner’s problem: the business has stopped converting, so the shareholder must now reconvert what he receives, redeploying the cash into a new base himself, or letting it sit idle. The short-duration business transfers the work of conversion from the company to its owner.
This is why an investor may prefer a business that allocates its own overflow — one that reconverts internally into a long base, whether one it already owns or one it is building. Alphabet pours the cash of Search into cloud and artificial intelligence; Berkshire reinvests the float and the earnings of its subsidiaries into new acquisitions and redeployable capital. These are permanent dual-duration companies: they never quite overflow to the owner, because they keep absorbing internally, converting short into long without end. To hold one is to delegate the reconversion to an allocator rather than to perform it oneself — which is worth doing only when that allocator converts at least as well as the owner would, and the base it builds absorbs capital above its cost. Where the magnificent toll makes the owner the converter, the permanent dual-duration company remains the converter itself, and spares him the work for as long as it allocates well.
This is the difference between knowing the parts of a business and seeing its duration. One stops asking what a company earns this year and begins asking how long it will go on earning, and how much of that worth lies near or far: which kind of toll, with which chance of becoming a base, at which point of its curve. A one-off toll that only stacks is short whatever its bookings; a magnificent toll whose base has filled is short again whatever its margins; a free-growth toll still climbing its curve, or a mechanical compounder with a long runway ahead, is long. Answer the three in order and the economic duration stands revealed — and with it most of the decision, before the first statement is opened. It is the quantity Munger was pointing at: the rate the business earns on capital, and the years over which it can keep earning it, are what the holder finally receives — and no entry price, high or low, changes what that duration will deliver.
Watch the Video