What “intrinsic value” really means — and three ways to estimate it

The price of a share is a fact. You can look it up, and two people who look it up get the same answer. Intrinsic value is not a fact. It is a claim about what a business is worth, and the only way to hold a claim like that to any standard is to state the reasoning well enough that somebody else can disagree with a specific part of it.

Most of what goes wrong in valuation goes wrong at that step. A single number gets published with no visible assumptions, and there is nothing to argue with — so it either gets believed or ignored, and neither is analysis.

A definition worth arguing with

The most useful working definition: intrinsic value is what a business is worth to an owner who never gets to sell it. Not what someone will pay you next year — what the enterprise itself will produce for whoever holds it.

The reason to define it that way is disciplinary. It removes the exit from the argument, so you cannot value a company on the basis that someone else will want it more later. What is left is cash the business generates and assets it owns, which are the only two things an owner who cannot sell actually receives.

Three ways to estimate it

1. Discounted cash flow — the most rigorous and the most abusable

Project the free cash flow the business produces, discount each year back at the return you require for the risk, and add a terminal value representing everything beyond the forecast. Conceptually it is the correct method, and every other approach is a shortcut to it.

In practice, two things make it dangerous. First, the terminal value is usually more than half the answer, so the bulk of your valuation rests on a growth rate you assumed for a period you did not model. Second, it is exquisitely sensitive: move the discount rate by a single percentage point and the output can shift by a fifth. A DCF is the most honest method available and the easiest one to make say whatever you already believed. Both are true at once.

2. Normalized earnings power, priced at a multiple

Take a normalized earnings figure — an average across a full cycle rather than the most recent good year — and apply a multiple appropriate to the kind of business. This is cruder than a DCF and considerably harder to abuse, because it has exactly two moving parts and both are visible. Someone who thinks you are wrong can point at the multiple or at the averaging window, which is precisely what you want from a model.

Its weakness is real: a fixed multiple applied across a whole sector is a blunt instrument, and a decade of history may describe a business that no longer exists. But an assumption you can see and adjust beats an assumption buried in a spreadsheet.

3. What the balance sheet is worth on its own

Total assets minus total liabilities gives you book equity; price that at what the market typically pays for that sector's net assets and divide by the share count. This ignores earnings entirely, which is exactly why it is worth doing — it is a genuinely independent second answer to the same question, built from a different part of the same filings.

It is at its most informative for banks, insurers, industrials and property, and at its least informative for a business whose real assets are a brand, a codebase and a customer base that never appear on a balance sheet at anything like their worth.

Why you want more than one estimate

The instinct is to run several methods and average them, which throws away the most valuable output. The spread between the methods is a measurement of your own confidence. Two independent approaches landing within 15% of each other is a genuine signal. Two approaches disagreeing threefold is not noise to be smoothed away — it is the model telling you that the answer depends almost entirely on which lens you picked.

One hypothetical company, three methods

The company below does not exist and the figures are round on purpose. Everything is illustrative; the point is the shape of the disagreement, not the numbers.

Three methods, one hypothetical company, and a spread from roughly $46 to $96. If that looks like a failure, it is not. It is the honest output, and it happens to be instructive: in this example the terminal value is a little over half of the DCF result, the earnings method depends entirely on whether 20.2× is a defensible multiple for the sector, and the balance-sheet method assumes book equity is worth four times what it is carried at.

Notice also that reported net income of $4.0B sits above free cash flow of $3.0B. That gap is not an error in the example; it is the normal state of a capital-intensive business, and it is the sort of thing a single blended number would have hidden.

What to do with a spread that wide

1. 01Work from the low end rather than the middle. A floor is a usable number; an average of three methods is a number nobody computed and nobody can defend.
2. 02Name the assumption you are least sure of, out loud. In the example above it is the multiple, and everything downstream of it inherits that uncertainty.
3. 03Let the spread size the discount you demand. A wide disagreement between methods is a direct argument for requiring a larger margin of safety — the methods are telling you they do not know.
4. 04Recompute when the filings change, not when the price does. If a valuation moves because the stock moved, it was never a valuation.

What intrinsic value is not

• It is not a price target. It carries no view on what the stock will do, and no date attached to anything.
• It is not a number the market has agreed to respect. A price can sit below a well-argued estimate of value for years, and may never converge on it at all.
• It is not precise. A range honestly derived is worth more than a point estimate spuriously derived, and anyone quoting a value to the cent is telling you something about their process rather than the company.
• It is not a substitute for understanding the business. Every method above assumes the earnings power persists. Nothing in the arithmetic checks that assumption.

A valuation is not a prediction. It is a statement of what you would have to believe.

How Hety handles this

Hety values every company in the S&P 500 and the Dow 30 two independent ways — a decade of its own reported earnings priced at a fixed sector earnings multiple, and its balance sheet priced at a fixed sector price-to-book — and shows both ends as a range rather than a single figure, working from the lower of the two as its floor. The multiples are constants, published in full, so two people running Hety on the same company get the same range and can see exactly which assumption to discount. For the first method it also gives you a discounted cash flow model you drive yourself, with your own free cash flow, growth, discount rate, terminal growth, net debt and share count, because a DCF built on somebody else's assumptions is not worth reading. And where a source is missing, Hety shows an explicit blank instead of estimating one.