Why a re-rating lags the climb

A company can finish the hard part of moving up the value chain — the new plant running, the qualification won, the specialty share past half of revenue — and still trade on the multiple of the commodity business it used to be. That gap is the whole opportunity in value-migration investing, and it is worth being precise about where it comes from, because “the market is slow” is not an explanation. The delay is produced by four specific mechanisms, each with its own clock, and each of them is observable while it runs.

The classification clock

Most of the machinery that surrounds a listed company sorts it into exactly one bucket. Classification schemes assign a single sub-industry based on where the preponderance of revenue or earnings sits. Screens are built on those buckets. Comparable-company analyses are drawn from them. Benchmark constituents are defined by them. None of this is unreasonable — a company has to go somewhere — but the consequence is that a business with, say, two fifths of its revenue in a higher-value line is still formally a member of the peer group it is trying to leave, and is still being valued against multiples earned by firms with different economics.

Reclassification, when it comes, tends to arrive after the fact rather than in anticipation of it. That produces a specific and readable asymmetry: the operating evidence of a climb shows up in segment disclosure well before the company’s formal category changes. An investor reading the filings sees the shift; a screen filtering on sub-industry does not.

The coverage clock

Sell-side research is organised by sector, and an analyst’s peer set, model structure and internal credibility are all built around a particular kind of business. When a company begins behaving like a different kind of business, the analyst covering it has three unattractive options: rebuild the model around economics they do not specialise in, hand the name to a colleague, or keep the existing model and treat the shift as a variance. The third is the path of least resistance, and it is also the one that keeps published estimates anchored to the old business design.

There is a second-order effect. Consensus estimates are an average of models that mostly share a template. If the template assumes cyclical mean reversion, then a margin that refuses to revert reads as an unexplained beat rather than as a level change, quarter after quarter, until enough consecutive quarters accumulate that the template itself has to be rebuilt. That rebuild is discrete and late.

The trailing-history clock

Valuation multiples are usually computed against trailing or near-term earnings, and those windows contain the old business. A company three years into a transition still has two of the three years in its trailing window drawn from the commodity era. Returns on capital, computed on a capital base that includes the newly built asset but earnings that do not yet reflect its mature output, look worse during exactly the period when the transition is most real.

This one is mechanical and therefore forecastable in shape if not in timing: as the pre-transition years roll out of the trailing window, the same business with no further improvement will show rising trailing quality. Some of what looks like a re-rating is simply arithmetic finally catching up with a change that happened years earlier.

The proof clock

The last mechanism is the one that matters most and is least discussed. A margin gain is ambiguous on arrival. The structural-versus-cyclical question cannot be settled with one good year, because a commodity business at the top of its cycle produces the same picture. The only thing that resolves it is a down-cycle survived: the company holds a visible margin floor while its old peer group does not. Until that test has been run, a cautious investor is right to withhold the higher multiple, and the market’s reluctance is not error but appropriate uncertainty.

This is why the re-rating so often arrives in a downturn rather than a boom. The evidence that distinguishes a climb from a cycle only becomes available when the cycle turns.

What the four clocks imply for how you watch

Mechanism	What you can observe while it runs	What resolves it
Classification	Segment revenue mix diverging from the assigned category	A formal reclassification, typically after the mix has already shifted
Coverage	Repeated beats against estimates built on a mean-reversion template	Models rebuilt around a new peer set, or coverage handed to a different desk
Trailing history	Pre-transition years still sitting inside the trailing window	Those years rolling off, with no change in the business required
Proof	A margin gain not yet tested by a downturn	A down-cycle the company’s margin floor survives

Read together, these give a rough ordering rather than a date. The mix evidence is first and is available to anyone reading segment disclosure. The trailing arithmetic is second and is purely a function of elapsed time. The proof arrives whenever the cycle next turns, which is outside anybody’s control. Formal reclassification and rebuilt coverage generally come last, and by then the multiple has usually moved.

Limitations, honestly stated

None of this tells you the gap will close. A business can migrate up the chain and still deserve its old multiple, because the higher-value position turns out to be undefended, the end market disappoints, or the capital intensity of the climb eats the margin benefit. “Underpriced relative to what it is becoming” is a hypothesis about the business, not a promise about the share price.

Nor is a lag always present. Transitions that arrive through acquisition, or in businesses with unusually clean disclosure, can be recognised almost immediately — and a company whose migration story is well covered may already carry a multiple that assumes the climb succeeds, which is the mirror-image risk. The four clocks describe why a delay is common, not why one is guaranteed.

Finally, the reasoning above is about mechanisms, not magnitudes. We do not publish expected re-rating sizes or timelines, because any such number would be an invention dressed as an estimate, and the framework is worth more without one.

The bottom line

The market is not ignoring a migration; it is processing it through systems that update on schedules of their own. Classification follows the mix rather than leading it, coverage models change only when the template breaks, trailing windows clear by the passage of time, and the structural-versus-cyclical question cannot be settled until a downturn tests it. Knowing which clock is still running on a given name tells you what would have to become true for the gap to close — which is a far more useful question than asking when. The evidence we use to answer it is set out in the method and applied name by name in the US research book, and the forces that start a climb in the first place are catalogued in the field guide.