Value Investing

Scope. This article describes what value investors believe, what they actually do, and what four decades of evidence has found — including the period in which the evidence went badly against them. It does not recommend the approach, rank it against any other, or supply a screen, threshold or criterion to apply. No price, multiple or valuation appears anywhere. Nothing here is an argument that value investing works or that it does not.

Everything else in the tradition follows from taking that claim seriously. If price and worth were always identical, there would be nothing to estimate. If the gap existed but could never be measured, there would be nothing to do. Value investing is the position that the gap exists and is occasionally measurable — which is a narrower and more defensible claim than it is usually given credit for, and a weaker one than its adherents sometimes make.

What practitioners actually do

Estimate what the business is worth, by some combination of the methods in Pillar 25 — discounted cash flows, comparison against similar businesses, or the value of the assets themselves.

Compare that estimate with the price.

Insist on a gap before acting, rather than acting whenever the estimate exceeds the price at all. This is the margin of safety, and it is the tradition's most distinctive idea.

Wait. The approach supplies no mechanism by which price converges to value and no timetable for it to happen, which means holding periods are indefinite and unpleasant stretches are structural rather than exceptional.

The arithmetic the margin of safety implies

Treat the valuation as unbiased with a normally distributed error. The discount required to hold the chance of overpaying to a given level is then computable — and it depends entirely on how uncertain the estimate is.

How value is identified in practice, and the definitional problem

Formal research needs a rule, and the rule that became standard sorts companies by the ratio of book value to market price. That choice — made because book value was available and comparable across a large sample over a long history — has consequences that took decades to surface.

Practitioners rarely used it. Working investors have generally used earnings, cash flow, dividends, or a full valuation, and treated any ratio as a starting filter rather than a definition. So the thing the academic literature measures and the thing practitioners do are related but not identical, and evidence about the first does not transfer cleanly to the second.

What the evidence shows, including the part that went badly

The value premium was one of the most robustly documented patterns in empirical finance, observed over long periods and across developed markets, and it became one of the three factors in the model that reshaped asset pricing in the 1990s.

Then it stopped working, for a long time and by a wide margin. Measured by the standard book-to-market construction, value underperformed growth from 2007 onward, producing a drawdown of 55% by mid-2020 — the largest observed since 1963. That is not a rough patch; it is a period long enough to exhaust most careers and most patience.

The three explanations on offer, and this portal adjudicates between none of them. The premium was never real. One line of work argues the original construction involved several apparently innocuous choices that could reasonably have been made otherwise, and that the measured premium is smaller than first reported once those alternatives are considered — which is the multiple-testing argument from Pillar 22 applied to the most famous finding in the field. The measure broke, not the idea. A second line argues that book value fails to capture intangible assets in an economy where intangibles dominate, so the standard ratio stopped identifying cheap companies; capitalising intangibles, or using earnings and cash-flow-based definitions instead, restores much of the effect. The premium is intact and the starting valuations moved. A third decomposition attributes the entire drawdown to the widening spread between what value and growth companies were priced at, rather than to any deterioration in the underlying relationship. These are not compatible with each other, all three are argued by serious researchers with the same data, and the disagreement is live. A reader who leaves believing value investing has been refuted, or vindicated, has taken a side the evidence does not currently support.

Two properties that hold whatever the evidence settles on

The approach requires tolerating being wrong for years at a time. Not as a psychological aside but as a structural feature: an approach that buys what others are avoiding will, by construction, spend extended periods holding things that continue to be avoided. Thirteen years of underperformance is inside the range the historical record contains.

And the approach is unusually sensitive to a single judgement. Everything rests on the estimate of worth, which is a construction with disputed inputs — as the article on the cost of capital shows, a defensible range of one input alone can move a valuation by tens of percent. A margin of safety protects against error in the estimate; it does not protect against the estimate being the wrong kind of thing.