Two instruments, two questions
A survey asks. It is the right instrument for attitude and for motive: what people think, and the reasons they give for it. Cadence reads one every quarter — the harmonised consumer survey is where our confidence, savings-intent and big-ticket figures come from, and the position every market sits on is two of those measures.
A behavioural measure observes. Volumes through the tills, nights actually spent, what a market searched for. Nobody is asked anything, it is dated, and it records what happened rather than what someone intended.
Neither is the better instrument. They are answering different questions, and the answers do not always agree.
A level and a rate of change
The difference that matters commercially is not reliability, it is shape. A survey gives you a level: where a market sits, as of a fielding date some weeks back. A behavioural series gives you a slope: what is moving now, and how fast.
It also has an open vocabulary. A panel can only return what was on the form; a search basket or a chart can surface the brand, the artist or the theme nobody thought to ask about. That is a real advantage and it is a narrow one — it tells you what is moving, not why.
Where they part company
France and Germany read the same position this quarter: both below their own confidence norm, both pulled back on big-ticket intent. Underneath, the two instruments are in different places in one of them and the same place in the other.
| France | Germany | |
|---|---|---|
| Big-ticket intent vs own mean | −1.0, steady | −3.4, suppressed |
| Retail volume, year on year | +1.3% | −0.4% |
| Non-food excluding fuel | +2.8% | −0.6% |
| The two instruments | Disagree | Agree |
French households report easing off a large purchase while French retail volumes rise. German households report the same direction more sharply, and German volumes are flat to falling. One market is saying something its tills do not confirm; the other is not.
The quarter where the two instruments disagree is the quarter worth looking at.
What the disagreement is not
It is not evidence that the survey is wrong. Intent to make a large purchase and total retail volume are different constructs measured by different instruments on different frames, and a gap between them is a gap between two measurements before it is anything about consumers. Cadence reports it and infers nothing from it.
Nor is it a forecast. We pre-registered five tests of whether our signals predict demand and published all five results; none cleared the bar. A rate of change is still a description of the present.
Why it matters
A brief written from a stated measure alone is written from a level that was true a few weeks ago. A brief written from behaviour alone knows what moved and not why anyone moved. Read together, the moments where they separate are the ones worth a strategist’s attention — and they are visible only if you are reading both.
References
- Grounded in the revealed-preference tradition: read sentiment from what a market does, alongside what it says. See Field Note 0001, and Edmans, A., Fernández-Pérez, A., Garel, A., & Indriawan, I. (2022). Music sentiment and stock returns around the world. Journal of Financial Economics, 145(2).