Why Wage Inflation Matters More Than CPI

Here is the framing we use, and it earns its keep across every inflation regime: the wage floor. In the labour-intensive part of the economy — most of services, which is most of the consumption basket in a developed economy — prices are, over any horizon that matters for valuation, markups over unit labour cost. Unit labour cost growth is wage growth minus productivity growth. Inflation in that part of the basket can print below the floor for a while, dragged down by goods deflation or an energy unwind, and it can print above it when margins are expanding. But it cannot settle below the floor, because no population of service businesses will absorb rising labour costs into shrinking margins indefinitely. CPI tells you where inflation has been. Wage growth net of productivity tells you where it can stay.

The identity underneath the framework

The reason this is a floor and not merely a correlation is an accounting identity, not a behavioural theory. Every unit of nominal output resolves into labour cost, non-labour cost, and margin. When wage growth runs persistently above productivity growth, unit labour costs rise, and the economy as a whole has exactly three ways to absorb that: prices rise, margins compress, or productivity accelerates. The first is inflation. The second is a transfer from shareholders to workers that shows up directly in earnings. The third is the only benign exit, and it is the one nobody controls on a policy-relevant timetable. This is why a central bank reads wage data the way an engineer reads a fuel line rather than the way a statistician reads a survey — it is upstream of the thing being targeted.

It also explains an asymmetry that confuses CPI-watchers. A soft inflation print with firm wage growth is not good news that arrived early; it is a margin squeeze in progress, and its resolution — either prices catch up or earnings give way — is deferred, not avoided. A firm inflation print with decelerating wage growth is the reverse: the printed number is living off pricing that the cost base will not support, and the path of least resistance for inflation is down even before policy does anything. The wage floor turns individual CPI releases from verdicts into observations about where the price level sits relative to its resting point.

Why wages move last and stop last: three layers of why

That last step is the compressed version of the whole framework. Sterling's Embedded Intelligence treats a wage release as two separate inputs — a persistence signal routed to the discount-rate view, and a cost signal routed to the margin view of labour-intensive sectors — rather than as one labour-market headline. Sterling's discipline is to ask, on every wage print, a single question: is this number above or below the productivity trend, and who is absorbing the difference?

The illustration: indexation versus anchoring

The strongest case against watching wages

The serious objection runs like this: wages are a lagging indicator by your own description, so watching them is watching the past. Empirically, the wage Phillips curve has looked flat for long stretches — labour markets tightened for years in the pre-pandemic expansion without producing an inflation problem — and in the post-pandemic episode a respectable body of analysis attributed the initial price surge more to margins and supply conditions than to labour cost. On that view, wages did not cause the inflation, so they cannot be the master variable, and the investor who waits for wage confirmation will be late to every turn.

We accept most of that and reframe the conclusion it points to. Wages are rarely the spark of an inflation episode — sparks come from energy, supply chains, fiscal impulses, pandemics. Wages are the ratchet: the mechanism that decides whether the spark becomes a regime. An inflation shock that never enters the wage base is a relative-price event that base effects will flush out of the data on their own schedule. The same shock, once it is in the wage base, has acquired a defender — an annual, backward-looking, downwardly-rigid repricing process that will reproduce it year after year. So the framework is not "wages predict inflation"; it is "wages decide inflation's duration." The flat-Phillips-curve years are consistent with this: no spark reached the ratchet, so the ratchet had nothing to hold. And the lateness objection dissolves once you are clear about what is being traded: the investor watching wages is not trying to forecast the next CPI print, which is mostly noise; they are trying to price the persistence of inflation, which is exactly what the policy path — and therefore the equity multiple — is a function of.

The three portfolio channels

The practical value of the wage floor is that it routes one release into three distinct portfolio questions. The first channel is the discount rate. Because the policy path keys off wage persistence rather than headline prints, the trajectory of wage growth relative to productivity is the cleanest single read on how long policy stays restrictive — and long-duration equity, where most of the present value sits in cash flows beyond five years, is levered to precisely that. A portfolio heavy in rate-sensitive growth assets is, whether its owner frames it this way or not, short the wage floor.

The second channel is margin incidence. When wage growth runs above output-price growth, the gap is compressing margin somewhere, and it does not compress uniformly: labour-intensive businesses with weak pricing power — much of consumer services, staffing-heavy healthcare delivery, hospitality, parts of retail — absorb it directly, while capital-intensive or pricing-power-rich businesses shed it. The wage–price gap is therefore a sector-rotation variable hiding inside a macro release. The third channel is real income. Wage growth relative to CPI is the purchasing-power line for the household sector: when it is positive, consumer volumes have a tailwind that can coexist with tight policy; when it is negative, the consumer is being squeezed even if the labour market looks strong on quantity measures. Note the tension between the channels — the same firm wage print that is bad for the multiple is good for consumer volumes and bad for labour-intensive margins. A single number, three signs, three different exposures. Collapsing them into one "hot or cold" verdict is how wage releases get traded backwards.