The rate of growth of control retail sales was significantly stronger than expected during January and the revisions through January did not provide a significant offset. Along with the modest strength in the auto SAAR for the month, this suggests that real PCE growth will have run at 44 basis points, if you will forgive the false precision which avoids a tedious discussion of rounding.
In working up this estimate, I need to make an assumption for the 65% of PCE that is not informed by the retail trade report or the auto SAAR. The typical approach here is to extrapolate the recent growth momentum. But I would be the first to concede that this causes some hesitation this month. The reason is that the 3-month growth rate there fell from 3.9% in December to just 0.9% in January, which is why the January real PCE print missed expectations, even though expectations had been forced lower by the earlier reported weakness in retail sales and autos. Going for less wrong and an attempt to guess what the consensus will have factored in by the time the broader Income and Consumption report comes around, I assume that the 65% of PCE that is basically a guess will have grown sequentially at an annualized rate of 2.5% during February. There is some error around that estimate, but it is not a huge deal, especially when thinking about what the consensus will factor in before the income and consumption report, because most of the action remains in retail sales and the auto SAAR.
Assuming my estimate is right, that there are no revisions, and that real PCE is up at an annualized rate of 2.5% in March, the Q1 real PCE growth rate is on track to print at 1 1/2% (ar). The Atlanta Fed’s estimate is about a percentage point lower, so I think there is a downward bias there and that they should revise up. But it is not odd — or even totally unreasonable — for them implicitly to forecast revisions in the data and thus to stick with a number that looks implausible, even quite late into the quarter. So, we shall see. The real trouble with the Atlanta Fed estimate is that they are wrong in their treatment of the trade sector (and probably its interaction with inventories) for reasons I have been over and I assume are now well circulated.
The quarterly bean count of real PCE growth can be arbitrary because of base effects and noise introduced by auto sales, among other things. Trying to control for those two specific distortions, I like to pass a trend line through the level of real PCE ex-autos, as in the right panel of the chart below. Prior to the January retail trade report that told us that real PCE growth in January would be weak, that trend line was looking like 3%. I subsequently eyeballed it at 2 1/4%, and decided to stick with that, even though the real PCE report in January was even weaker than expected. But with February looking firm, that 2 1/4% trend line now seems quite plausible. This is inconsistent with the notion that the US economy has already dipped into recession, although growth has clearly cooled.
Source: BEA, FH estimate
Data are actual to January and estimated to February as described in the text.
The imposition of aggressive tariffs along the lines of what Trump has proposed can easily tip the US economy into recession, which is probably roughly consensus. And this lingering uncertainty encourages businesses to hesitate, which acts as an accumulating headwind, even if tariff heavy is still somehow avoided. But we are not yet at the point where recession is inevitable or even likely if another option is chosen on the tariff side. Recession or not remains a policy choice, not at the Fed but within the Administration. Congress is MIA.
