American factory production fell 0.3 percent in August, following an unrevised 0.2 percent rise in July. Economists polled by Reuters had forecast a 0.3 percent increase. It ends seven straight months of gains. Output was still 0.9 percent higher than a year earlier. Durable goods production led the decline, down 0.5 percent, with business equipment also falling. Utilities rose 1.8 percent and mining edged up 0.1, leaving overall industrial production unchanged on the month and up 1.4 percent year on year.
Manufacturing is about 9.4 percent of the American economy, and the Reuters account contains the detail that explains the whole shape of the year. Production in prior months received a boost as businesses rushed orders to avoid shortages and higher prices from the escalation of the war in the Middle East.
That is pull-forward, and it is the most misread pattern in business.
When a cost shock is announced or anticipated, buyers do not reduce demand. They accelerate it. Everybody orders early, fills the warehouse, locks the price, and the resulting surge looks exactly like growth in the data. It is not growth. It is next quarter’s demand arriving in this quarter, and the only thing it guarantees is a hole later.
A cafe runs this cycle in miniature every time a supplier announces a price rise with four weeks’ notice. You buy three months of beans, cups and syrup. Your supplier has a magnificent month. Then for the following two months you order almost nothing, and your supplier, looking at their own figures, concludes that demand has fallen off a cliff and starts worrying about losing your account. Nothing about your business changed. They simply got paid early and are now experiencing the absence of it.
August is where that absence showed up in American manufacturing, and the cost side arrived at the same time. Samuel Tombs of Pantheon Macroeconomics expects output to rise a little further but not to match the first-half pace, and the reason he gives is the one that matters: some manufacturers will find demand softens as they pass on higher energy prices to consumers. That is the whole pass-through problem in a sentence. You can move the cost along or you can absorb it, and the first option shrinks the order book while the second shrinks the margin.
The offsetting force is real and worth stating clearly. The AI infrastructure build-out has cushioned the blow from tariffs and continues to support semiconductors, electrical equipment, power generation and construction. Bernard Yaros of Oxford Economics expects manufacturing to pick up through next year, with the demand case for AI strong enough to shrug off higher-for-longer rates and geopolitical risk, and greater defence spending as a further tailwind.
But that spending does not reach every factory. A company making motor vehicles or business equipment is not in the data centre supply chain, and the 0.9 percent annual figure suggests a loss of momentum rather than a contraction.
The other reading in the Reuters piece is political. Some economists took the modest year-on-year number as evidence that aggressive trade policy has not rejuvenated the industrial base as intended.
One month is one month. The seven before it were partly borrowed.
Source: Reuters, “US factory production falls in August; outlook clouded by rising costs”
