Data_Note: Finally, a hard data point.
Retail sales, and unfortunately, it clarifies very little.
For those of us obsessing about if and when the very bad soft data start to infect the hard data (soft: consumer/biz surveys; hard: jobs, spending, etc.), any hard data point right now is deserving of scrutiny. This morning, the Census released retail sales for February, which came in well below expectations at 0.2% vs. 0.6% expected.
If that leads you to think, “aha! evidence of weakening consumers pulling back based on the pessimism they’ve been reflecting in confidence surveys,” you might be right, but unfortunately, it’s not that simple.
Here’s why:
—A key stat from this report is called “core sales,” which is the topline retail sales net of a few volatiles categories (food services, gasoline, autos, building materials). Just like core inflation gives us a better signal on where inflation is heading, core sales is a better predictor of consumer spending for the quarter. It was expected to rise by a mere 0.2% in Feb but it popped up to 1% (all there numbers are nominal but commodity inflation was just 0.1% in Feb, so that’s a hefty real gain).
—So, all good, right? Not so fast. January’s core sales were revised down to -1%, so the strong Feb is a bounce back and the first two months of Q1 look pretty flat (also, for technical reasons, Jan gets a heavier weight in the quarterly spending calculation than Feb). Online sales were strong in Feb (+2.4%) but restaurants were weak, down 1.5%.
—How do we get strong core and weak overall retail sales? Clearly, because the categories we take out to get from headline to core did badly. Along with food services, cars and car parts were down 0.4% and gasoline spending fell 1%.
—By definition, retail sales leaves out most services, so it comprises just about a third of all spending; it’s a noisy series that provides more of a glimpse than a clear signal re consumers’ activities.
I don’t like this on-the-one-hand-on-the-other any better than you do, but that’s data for you. Bottom line, the big topline miss is suggestive of some consumer pullback, but the core number says maybe not.
As readers know, I worry that the sharp, quick negative turn in consumer sentiment/confidence will soon appear in hard indicators, and I think I see a touch of that here. But we also know that vibes and real outcomes have diverged for a long time, and it will take a lot more data than this report to see if my worries bear out, which for the record, I hope they do not.
If I understand this correctly, the core represents essential spending, whereas headline represents discretionary. If that is the case, then one would expect headline to start showing weakness first, with core lagging, maybe by quite a bit of time considering that eating isn't optional.
Jared, you advised against reading too much into the Atlanta Fed's GDPNow model going sharply negative at the end of last month. But here it is almost three weeks later, and this morning's estimate is still -2.1 percent. What do you make of that?