The Philadelphia Fed's six-month outlook index is the highest since August 1983. Its current price gauges fell — but not one firm reported paying less.

by Tomas Ferreira · August 21, 2026

Markets Reporter, US Market Current

The Federal Reserve Bank of Philadelphia released its August Manufacturing Business Outlook Survey on Thursday, August 20, at 8:30 a.m. Eastern, and the headline was strong enough to obscure the more interesting parts of the document. Current general activity rose to 47.4 from 41.4 in July, which the bank described as "a five-year high" and which is the strongest reading since April 2021. The forward-looking measure did something rarer: the future general activity index printed 73.6, characterized in the release as "its highest reading since August 1983."

That is a 43-year comparison, and it deserves scrutiny rather than applause. Diffusion indexes measure breadth, not magnitude. A reading of 73.6 means 74.5 percent of respondents expect activity to increase over the next six months against 0.9 percent expecting a decrease. It says almost nothing about how much activity they expect. Near-unanimity of direction among a regional manufacturing panel is a statement about sentiment, and sentiment indexes have historically run hot relative to what the hard data subsequently delivered.

The current-month internals are more equivocal than the headline. New orders fell to 30.1 from 37.0 and shipments fell to 27.7 from 33.7; the release says both "moved lower but remained elevated," which is a fair description. Unfilled orders eased to 14.4 from 18.1 and delivery times to 3.7 from 9.6 — both consistent with supply chains loosening rather than tightening. Inventories turned negative at minus 3.7 from 0.3 in July. What carried the headline was labor: the employment index jumped to 27.9 from 10.0 and the average workweek index to 26.5 from 14.0.

So the composition of August's improvement is hours and headcount rising while incoming demand slowed. That is a defensible configuration if firms are working through a backlog, but the backlog measure itself came down. It is worth naming that tension rather than smoothing it over.

The price story is where the release repays close reading. Both price indexes fell. Prices paid dropped to 40.9 from 53.9, a 13-point decline, and prices received to 17.7 from 27.4. The release notes that both "declined this month but remained elevated." Taken alone, that looks like the beginning of a cost-pressure reprieve.

The response distribution says otherwise. The prices paid diffusion index is the share of firms reporting increases minus the share reporting decreases. In August, 40.9 percent of respondents reported paying higher prices and 0.0 percent reported paying lower prices. The index equals the increase share exactly because there was nothing on the other side of the ledger. A prices-paid index can fall for two very different reasons: because more firms are seeing costs decline, or because fewer firms are seeing them rise. This month it was entirely the latter. Nobody in the panel got cheaper inputs.

The forward price indexes point the same way with more force. Future prices paid stands at 62.9 and future prices received at 59.8 — respectively about 22 and 42 points above the current readings. On the same breadth logic, 63.7 percent of firms expect to pay more six months from now and 60.6 percent expect to charge more, against 0.8 percent and 0.9 percent expecting declines. The current spread between prices paid at 40.9 and prices received at 17.7 is 23 points; the forward spread between 62.9 and 59.8 is about three. Those are gaps in breadth rather than in dollars — the survey does not measure the size of any price change — but the direction is unambiguous: far more firms report paying more than report charging more today, and the panel expects that discrepancy to nearly close over the next six months.

The survey's special questions this month address exactly that intention, and they are the most useful thing in the release. Asked how customer sensitivity to price increases has changed, 37.5 percent of respondents said customers had become more sensitive, 58.3 percent reported no change and 4.2 percent said less sensitive. Asked whether they could clearly anticipate any changes in their industry's costs over roughly the next six months, 43.5 percent said yes and 56.5 percent said no. Of the firms answering yes, 80.0 percent expect competitors to raise prices, 20.0 percent expect competitors to hold prices steady and 0.0 percent expect competitors to lower them. The median expected timing for those competitor increases was four months.

Four months from August is December. If that median is even directionally right, the panel is describing a pass-through impulse that has not yet reached the price data and does not expect to until late in the fourth quarter. That is a specific, dateable expectation from firms about their own pricing environment, and it sits uncomfortably beside a current prices-received index that just fell nearly 10 points.

Two caveats belong on all of this. First, the Philadelphia Fed's panel is regional and manufacturing-only, in an economy where manufacturing is a minority of output; the survey is a coincident temperature check, not a forecast of the CPI. Second, the 80 percent figure is a share of a share rather than a share of the panel: it is four-fifths of the 43.5 percent who anticipate any cost change at all, or roughly 35 percent of respondents. The follow-up was put only to that subset, and the release does not print a separate response count for the special questions, so the number of firms standing behind the 80 percent is smaller than the number behind the headline indexes and is not disclosed.

The macro backdrop makes the cost question live rather than academic. Crude oil moved higher during Thursday's session, with West Texas Intermediate trading above $86 a barrel and Brent holding past $91, according to a market recap of the day that cited tensions around Iran and the Strait of Hormuz. Those crude figures come from that recap, not from the Philadelphia Fed, whose survey does not track commodity prices. Energy is one of the more reliable inputs into manufacturing cost expectations, and it moved in the direction the survey's forward price indexes anticipate on the same day the survey was published.

What the August report actually supports is narrower than the 1983 comparison suggests. Regional manufacturers are hiring and lengthening the workweek; incoming orders slowed; nobody is getting cheaper inputs; and a meaningful minority expects an industry-wide price increase around the turn of the year. That is a coherent picture, and it is not the same picture as a five-year high in a headline diffusion index.