Confidence Falls to 89.4 (25 August 2026): Present Situation +6.8, Expectations Down to 68.2 — and a 16 August Cut-Off That Missed the Fed Minutes and the Tariffs

The Conference Board's August Consumer Confidence Index fell 0.8 points to 89.4, missing a 90.2 consensus and printing the weakest headline since January. But the composite is hiding the story. The Present Situation Index jumped 6.8 points to 121.2, snapping a three-month decline, while the Expectations Index sank 5.8 points to 68.2 — the widest split between how households read the present and the future in months. The labour differential widened to +7.5 from a revised +2.7. And the cut-off was 16 August, which means the survey closed before the Fed minutes, before the 30-year went back to 5.27%, and before the 50% Canada tariffs landed.

This page argued in July that the value of the release lies in reading its cut-off date rather than its headline. August proves the point twice over — once because the survey again closed a week before the news that moved the dollar, and once because the forecast this page made on the strength of that reasoning was wrong. Both are worth working through, because the error is more instructive than the hit.

What actually happened

The release landed at 10:00 a.m. Eastern on Tuesday 25 August, one day before core PCE and three days before the Jackson Hole keynote.

Chief economist Dana M. Peterson framed it as confidence having "moderated slightly in August for a second consecutive month", with "the Expectations Index slipped further into negative territory, which was offset by a moderate rise in the Present Situation Index after declining in the past three months". The full detail is in the Conference Board's release.

Note also that July was revised down on every line — the headline from 90.8 to 90.2, the Present Situation from 114.9 to 114.4, the Expectations Index from 74.7 to 74.0, and the labour differential from +3.1 to +2.7. That matters for the August comparison: some of the Present Situation's 6.8-point "jump" is measured from a base that was quietly marked lower a month after the fact. The direction is not in doubt, but the magnitude is flattered.

The split is the story, not the headline

A 0.8-point fall in a second-tier sentiment survey is noise. A 12.6-point gap opening between the two things the survey measures is not.

Households told the Conference Board that conditions right now are the best they have described in months, and that the next six months look worse than they have in some time. Both statements came from the same sample in the same fortnight. The composite averages them into a number that resembles a quiet month, which is precisely how a genuinely unusual release ends up ignored.

The labour internals are the cleanest illustration. In July, both sides of the jobs question fell — fewer said jobs were plentiful and fewer said they were hard to get, the signature of households drifting into uncertainty. In August both sides moved decisively and in the direction of strength: plentiful up 2.6 points, hard-to-get down 2.2. That is not a migration to the neutral middle. That is households reporting an actual improvement in hiring conditions, and it reverses the three-month cooling trend this page flagged in July as the thing to watch.

The cut-off was 16 August. Again, it closed early.

Here is the structural point that survives every month's noise. The Conference Board collected August responses through 16 August. Line that up against what has happened since.

Every one of those repriced something. None of them is in the data. The July minutes released on 19 August showed several participants had favoured an increase; the long-end selloff put the 30-year back at 5.27% two days later; the 50% duties on Canadian goods took effect on 22 August with talks suspended. Jackson Hole falls outside the window as well.

So the release is an accurate description of household sentiment in the first half of August and a poor description of the environment a trader is looking at on 25 August. That is not a criticism of the survey — it is what a monthly survey with a mid-month cut-off is. It is a criticism of reading it as though it were current.

The forecast this page made, and why it was wrong

In July this page argued that because Brent's spike above $100 on 23 July round-tripped entirely to $88.36 by 27 July, the August survey — the first to contain any of the energy episode — would plausibly show inflation expectations ease rather than spike, removing a stagflationary complication.

That call was wrong, and the way it was wrong is the useful part. The Conference Board reported that average and median 12-month inflation expectations were slightly more elevated in August.

The error was a category mistake: treating the front-month crude contract as the price households respond to. They do not see it. They see the posted number at the forecourt, and that number did not go down.

AAA's 20 August update put the national average at $4.10, up three cents on the week and the highest it has ever been on that date, and attributed it to crude staying in the $80 range amid continued instability around the Strait of Hormuz. Two mechanisms sit behind the disconnect. The first is that crude did not stay at $88 — it climbed back through August, so the "net move was down" premise expired within days. The second is the well-documented asymmetry in retail fuel pricing, where pump prices rise quickly with crude and fall back slowly, so a round trip in the futures market is not a round trip at the forecourt.

Worth flagging that the record is nominal, not inflation-adjusted, and a nominal record is a weaker economic statement than it sounds. But households answering a survey respond to the posted price, not to a deflated series — which is exactly why the nominal figure is the one that showed up in the expectations data.

What it does to the dollar

The release itself was close to a non-event for price. The dollar index sat near 98.96, essentially unchanged on the session, with EUR/USD at 1.1654 and USD/JPY at 159.29. Attributing any of that to a consumer survey the day before core PCE would be the classic error. The contribution is to the underlying factors, and there it is genuinely two-sided:

• Interest-rate factor: mildly hawkish. A 4.8-point widening in the labour differential is the survey's clearest signal, and it lands with the market carrying roughly a third to 40% odds of a September increase after the minutes. Households reporting that jobs are easier to find does not sit comfortably with the case for a cut, and it is the sort of input that firms up the hawkish side of a genuinely two-way September.
• Growth factor: negative on the forward leg. An Expectations Index at 68.2, deeper below the sub-80 threshold it has not cleared since February 2025, is a mark-down of the forward US consumption read — and consumption is roughly two-thirds of GDP. The present-conditions strength does not offset this for a currency framework, because exchange rates trade on expected differentials, not realised ones.
• Commodity factor: transmitting, and the wrong way for the disinflation case. The expectations line confirms that fuel is still passing through to household price views. That is the same channel behind the record diesel cracks, and it argues against the clean energy-led disinflation this page expected in July.
• Risk-sentiment factor: neutral. An orderly split print generates no haven flow.

The net is a dollar picture that is marginally firmer on rates and marginally softer on growth — which is, unhelpfully but honestly, roughly where it started. The information value is in knowing which factors moved and by how much, not in a directional conclusion the data does not support.

What to watch

Core PCE lands on 26 August, one day after this release, and Warsh's first Jackson Hole keynote follows on 28 August, both covered in the Jackson Hole preview. Either can overwrite the rate factor entirely; a consumer survey cannot. The Federal Reserve's calendar puts the decision itself on 15–16 September.

Two things from this print are worth carrying forward. The first is whether the Present Situation rebound holds — one month of a 4.8-point differential swing, off a base that was revised lower, is not yet a trend reversal, and the September survey will say whether households were reacting to something real or to a good fortnight. The second is the September window, which will be the first to contain the minutes, the tariffs and Jackson Hole. If forward expectations are already at 68.2 without any of that in the sample, the interesting question is not whether the number falls but whether the present-conditions leg follows it down. That is the point at which a split stops being a curiosity and starts being an inflection.

For the method behind the five-factor framework and how each input is scored, see how the meter works.

Educational macro context only — not investment advice.