February JOLTS report: “soft landing” so far, but indications of further weakness ahead

 

- by New Deal democrat

 Along with all the other reports, yesterday the JOLTS survey was updated for February. This survey decomposes the employment market into openings, hires, quits, and layoffs, and so gives a more granular view. The question over the past year has been whether they best describe a “soft landing,” or “hard” one ending in recession.

Additionally, several components are slight leading indicators for jobless claims, unemployment and wage growth.

In February the news was mainly good, as most categories stabilized.

First, here are openings, hires, and quits all normed to 100 as of just before the pandemic:

Openings, which are “soft” data and have generally uptrended going all the way back to the turn of the Millennium, remain above their pre-pandemic levels, but this is not terribly significant. Both hires and quits fell below their pre-pandemic levels at the beginning of last year. But the far right end of the graph suggests stabilization, which is even more apparent when we zoom in on the past 12 months:

Openings are virtually unchanged from their level 10 months ago. Hires have been very stable since last July, and Quits have been stable since August. In fact on a 3 month average basis, both Hires and Openings have remained within a 1% range.

This is good news. It says “soft landing” (barring political own-goals in Washington like massive tariff wars, of course).

One item of concern in the report was layoffs and discharges, which increased to their highest level in almost two years excluding last September:

This is of a piece with the uptick in new jobless claims (red, right scale) we have seen in February and March, which in turn suggests there may be some upward pressure on the unemployment rate in the coming months.

Finally, here is the update on the quits rate (right scale) vs. the YoY% change in average hourly wages for nonsupervisory workers (red, left scale):

While the quits rate has been stable since last August, it did downshift from earlier in 2024. Average hourly wages have not yet reflected that downshift, but the likelihood is that they will follow, down to a YoY growth rate of about 3.6%-3.7% in the coming months.

To sum up, the coincident reporting in the JOLTS data indicates “soft landing” so far, while the short leading components suggest weaker employment data in the next few months.

February construction spending: nominal vs. real makes all the difference

 

 - by New Deal democrat

Since the turn of the Millennium, a downturn in manufacturing has not been enough to tip the economy into recession. There must also be a decline in construction as well.

This morning’s construction spending report for February painted a substantially different picture depending on whether the data was looked at nominally or in real inflation adjusted terms.

Nominally total construction spending (blue in the graph below) rose 0.7% in the month to yet another new all-time record; while residential spending (red, right scale) rose a sharp 1.3%, close to a 12 month high:

But when we deflate by the cost of construction materials, the picture is not so rosy:

So adjusted, total spending actually declined -0.1%, and residential spending rose 0.4%. 

The picture for the leading residential sector is that spending has been trending sideways (much like permits, starts, and sales) for the past year, while total construction is slowing. In fact in real terms it has not advanced in four months.

Finally, the boom in manufacturing construction has also ended:

The overall picture for the entire goods producing sector of the US economy from this morning’s ISM and construction spending reports is that both manufacturing and construction are almost right at the juncture between expansion and contraction.

Heavy truck sales warrant a yellow caution flag

 

 - by New Deal democrat

After this morning’s contractionary ISM manufacturing report, it occurred to me to look at heavy truck sales to see if they confirmed the downturn. They did, but like the manufacturing index, 

While passenger vehicle sales are very noisy, heavy truck sales convey far more signal than noise, and usually turn down before car and light truck sales, as shown in the below historical graph (averaged quarterly for ease of viewing):

Here is the updated graph since the pandemic, monthly through the latest report from earlier this week for February:

Heavy truck sales were 438,000 annualized, the lowest monthly number since January 2022. The three month average, which does away with most of the noise, was 461,000 annualized, the lowest three months since spring of 2022.

When the three month average of heavy truck sales is down more than -10% from its peak, that has more often than not been a leading indicator of recessions within a year. But note the false positives in 1996, 2016, and 2022, so I am not at the point yet where I am sufficiently confident to say that they warrant a “recession watch,” just a yellow flag caution.

Typically auto and light truck sales have also declined before the onset of recessions, although as noted above they are much noisier. As late as December those made a 3+ year high. I would be looking for the three month average of those to fall below 15 million annualized as a confirmatory signal.

ISM manufacturing index returns to contraction as the front-running of tariffs has ended

 

 - by New Deal democrat

[Note: I’ll report separately on construction spending, also released this morning]

Although manufacturing is of diminishing importance to the economy, (it was in deep contraction both in 2015-16 and again in 2022 without any recession), the ISM manufacturing index remains an important indicator with a 75+ year history of accurately describing that sector and forecasting it over the short term. 

Any number below 50 indicates contraction. The ISM indicates that the number must be 42.5 or less to signal recession. I use an economically weighted three month average of the manufacturing and non-manufacturing indexes, with a 25% and 75% weighting, respectively, for forecasting purposes.

In the last few months, as most businesses likely figured that the new Administration would be laying more tariffs, it appears there was a rush to get their new orders in asap. That was confirmed by this morning’s report for March, in which the headline index declined -1.3 to 49.0, the lowest since November, and new orders declined -3.4 to 45.2, the lowest level since last June.

Here is a look at both the total index and new orders subindex since the Great Recession:

Including this month, here are the last six months of both the headline (left column) and new orders (right) numbers:

OCT 46.5. 47.1

NOV  48.4. 50.4

DEC 49.2. 52.1

JAN 50.9  55.1

FEB  50.3  48.6

MAR 49.0. 45.2

The current three month average for the total index is 50.1, and for the new orders subindex 49.6. While the headline number is similar to those of last summer and autumn, as noted above the new orders component is the lowest since last June. 

Last month I wrote that “The surge and then retreat in new orders in particular certainly looks like front-running potential tariffs.” This month’s further decline makes that look like almost a certainty. And the uncertainty about the level and extent of looming tariffs undoubtedly is also having an effect on new orders as well.

For the economy as a whole, the weighted index of manufacturing (25%) and non-manufacturing (75%) indexes is more important. Since In February the ISM services index came in at 53.5, and the more leading new orders subindex at 52.2, and the three month weighted average of each was 53.4 and 52.6 respectively, it would take a very steep decline in that index to signal recession now.

In summary, as of now the combined indexes continue to suggest that the economy is growing, albeit slowly. The non-manufacturing index for March will be reported on Thursday.

Why I am not concerned by the February increase in core PCE inflation

 

 - by New Deal democrat

The deluge of new monthly data starts tomorrow. Today there’s no significant data, so let me follow up on a point from Friday’s personal income and spending report.

I never look to see what others are saying before I finish my own analysis, because I want to be as unbiased as possible. After that I may check other analyses, and usually they aren’t too different (except for the perma-DOOOMers, who always gotta perma-DOOOM).

Friday was different, because almost everybody else who looked at the personal income and spending report focused on a perceived renewed inflationary pulse. So let me explain in a couple of graphs why I arrived at a different perspective.

A good example of the concern in other quarters was this take by Harvard Econ Prof. Jason Furman:

Core PCE inflation came in a little above the already high expectations in Feb. The pattern is the opposite of what you want to see--the shorter the window the higher the annualized rate (and still high at 12 months):

1 month: 4.5% 3 months: 3.6% 6 months: 3.1% 12 months: 2.8%

And he supplied in support the following graph:

The three month change in particular looks very worrisome.

But notice something else in the graph. Specifically, notice that there was a similar - even bigger - spike in the 3 month average at the beginning of 2024. And in fact there were lots of inflationary concerns expressed back then as well.

And then they completely faded away in the spring and summer.

The reason is apparent when we look at the monthly readings starting back in October 2023, shown in the below graph for both headline (light blue) and core (dark blue) PCE inflation:

There was a big spike in the monthly readings from January through April of last year. Then it simply stopped.

Which has the following effect on the YoY% numbers:

At its low point in the past year, YoY core PCE inflation was 2.63%. As of Friday it was 2.79%. In other words, an increase of all of 0.16%.

So what we have is a repeat of the monthly spike we saw 12 months ago, that has had a very small effect on the YoY comparisons. And is very much in line with the likely unresolved seasonality we have seen in a number of indicators since COVID, including weekly jobless claims and personal spending.

If the monthly number don’t back off in a month or two, and the YoY comparisons get significantly higher than they were 12 months ago, I’ll be convinced that there is a real problem. Unless and until that happens, I am somewhat skeptical.

Arctic sea ice makes new record low annual max

 

 - by New Deal democrat

On Sunday I occasionally post about topics of interest unrelated to economics. So today, let’s take note of a significant milestone in global warming.

Specifically, the arctic has just had its lowest peak ever for sea ice in modern history, at 14.33 million square kilometers. The next closest were 2017, at 14.41 million; 2018, at 14.47 million; 2016, at 14.51 million; and 2015, at 14.52 million. Here’s what the peaks look like graphically, including every year since the turn of the MIllennium:

Interestingly, this year was the third latest peak in the past 25 years, bucking the mild trend of earlier and earlier peaks over time.

Here is what the past several months have looked like in comparison with the four previous record low years noted above, as well as the historical trend:

Although at its peak, this year was higher than previous years on the same date, the peak was nevertheless lower than the peaks of the earlier years, each of which occurred earlier.

Finally, as you can see from the first chart, there has been a sharp decline in the week since this year’s peak, and it is now at an all-time low for this date as well. Here is a close-up of this year vs. the two closest runner-ups, 2006 and 2017:

I am frankly surprised that this page has not been taken down, which is one reason I wanted to make a record of the data here.