Growth

Retail sales rose eight tenths of a percent in September — and seventy-one cents of every new dollar was spent at the gas pump.

The headline is real, and it’s solid: the margin of error doesn’t reach zero. Sales rose five point seven billion dollars. Gas stations took four point one billion of it. They’re about nine cents of every retail dollar.

Now the gallons. Gas station receipts rose six point four percent. The consumer price report the day before had pump prices up nine. Paying more, buying less — about two and a half percent fewer gallons.

Take gas out, and sales rose two tenths of a percent. The Census Bureau’s own note says that cut, without autos either, can’t be told apart from zero.

So, volume. Adjust the whole month for prices, and spending rose about a tenth of a percent. In August, it rose nine tenths. The dollars held up. The stuff people actually bought nearly stopped growing.

Breadth narrowed too. In August, all thirteen categories rose. In September, eight. Restaurants were still up one point one.

Two housekeeping notes. Census rewrote the monthly history on September twenty-eighth, so the dollar levels we cited for August are lower now. And the savings cushion we said was funding spending turned out bigger in the government’s revision — we’ve corrected that page.

Here’s what that means for the Fed. A rate hike works on demand. In September, the part of spending that grew fastest was fuel — the part we said in our consumer price read a funds rate can’t reach. Measured in things, not dollars, demand barely moved.

One month is one month, and August’s volume was strong. But if the case for more tightening rests on spending, September’s version of it is mostly gasoline.

Watch volume, not dollars. October’s report lands November seventeenth.

Growth hit three point four percent last quarter, and the story you’ll hear is that AI carried it. Here’s the line that story skips: imports.

First, what added up. Consumers added two and a half points. Business investment, about one and two thirds. Inventories, one and six tenths. Trade took away three and a third.

Now the build — business equipment and software, where AI spending lands. It added one point six. Take it out and growth was one point eight. On that count, about half the quarter.

But look at what the equipment was. The agency says the gain was led by computers, measured from import data. Business computer purchases added seven tenths of a point. Computer imports took out one point one.

That’s how GDP works. A server built abroad adds in investment and subtracts in imports. It counts what’s made here — the software, the buildings, the installation. So the AI cycle’s net push on output is smaller than its investment line looks.

Now the consumer, the biggest piece. In dollars, spending grew about six percent, annualized. In volume, three point eight — the same pace as spring. Prices took two points, down from five. Volume held up.

Services led — health care and restaurants. Goods slowed. And as our PCE read put it, strong fits the summer: September paused, paid for out of savings.

Last time we said watch demand, not the headline. Consumer spending plus fixed investment rose four point nine percent, up from four point six.

Prices inside this report: core consumer inflation ran two and a half for the quarter, down from three point three. Our PCE read found the last three months at two point nine. The average hides a pickup.

So here’s what it means for the Fed. This isn’t an AI-only economy — the consumer did more than the build, and much of the hardware nets out. If the case for a second hike rests on demand, this report strengthens it. Its weak spot is the funding: savings, not income.

The second estimate lands November twenty-fifth — the same morning as October’s PCE.