September jobs report: a weakly positive report consistent with an inflationary expansion

 

 - by New Deal democrat

For most of this year my Big Theme has been that the AI Boom (or possibly bubble) has been counterbalancing a stagnant or even shallowly recessionary rest of the economy. In the last few months that has moved towards the economy being in an inflationary expansion, including a slight pick-up in employment and downward drift in unemployment.

This month came in at the weak end of that range.

Below is my in depth synopsis.

HEADLINES:

• +29,000 jobs gained. Private sector jobs increased 46,000, while government jobs declined -17,000. The three month average declined to +47,000.
• The pattern of downward revisions to previous months reappeared this month, as July was revised lower by -43,000, and August was also revised lower by -17,000 for a total decline of -60,000.
• The alternate, and more volatile measure in the household report, rose sharply for the second month in a row, by +406,000 jobs. But on a YoY basis, this series which had been negative for six months in a row before last month, turned negative again and is now lower by -504,000.. 
• The U3 unemployment rate rose +0.1% to 4.2%. This was the result of rounding, as the rate carried one further deimal point rose from 4.14% to 4.17%.
• But the U6 underemployment rate declined another -0.2% to 7.5%, its lowest in over 18 months.
• Further out on the spectrum, those who are not in the labor force but want a job now rose +43,000 to 5.790 million, still close to the bottom end of its recent range.

Leading employment indicators of a slowdown or recession

These are leading sectors for the economy overall, and help us gauge how much the post-pandemic employment boom is shading towards a downturn vs. rebounding. These were mainly positive for the third month in a row.

• The average manufacturing workweek, one of the 10 components of the Index of Leading Indicators, rose 0.3% to 42.0 hours, the highest reading of this entire post-pandemic expansion.

• Manufacturing jobs rose +9,000, the 6th increase in the last 12 months.
• Truck driving reversed its recent decline for the third month in a row, increasing by +2,600.
• Construction jobs rose +11,000.
• Residential construction jobs, which are even more leading, rebounded their recent 3 year low, up +3,000.
• Goods producing jobs as a whole rose +18,000. 

• Temporary jobs, which had declined by over -650,000 since late 2022, but has reversed higher most of this year, declined by -10,900.
• The number of people unemployed for 5 weeks or less rose +98,000 to 2.098 million, still low compared with the last 3 years.

Wages of non-managerial workers 

• Average Hourly Earnings for Production and Nonsupervisory Personnel increased $.07, or +0.2%, to $32.60, for a YoY gain of +3.3%. Except for July’s +3.2%, and several months affected by pandemic shutdowns, this equals the lowest since December 2019. This is slightly less than the 3.4% YoY inflation rate as of August.

Aggregate hours and wages: 

• The index of aggregate hours worked for non-managerial workers was unchanged, and is up 0.9% YoY, about average for the past 12 months.

• The index of aggregate payrolls for non-managerial workers rose +0.2%, and is up 4.2% YoY, and up 0.8% above the YoY inflation rate through August.

Other significant data:

• Professional and business employment declined for the first time in six months, down -9,000. These tend to be well-paying jobs. This remains above its low from last October, and remains higher YoY as well.
• The employment population ratio reversed its recent declines, rising another +0.1% to 59.2%, vs. 61.1% in February 2020.
• The Labor Force Participation Rate rose +0.2% to 61.8% , vs. 63.4% in February 2020. IMPORTANT: both the EPOP and LFPR are greatly affected by the retiring Boomer population. In the prime age 25-54 demographic, they are virtually unchanged.

SUMMARY

This was a weakly positive report, consistent with an inflationary expansion, but with reasonably good internals. While the employment gain was weak, and taking into account the revisions to the previous two months was on net negative, and after rounding the unemployment rate rose slightly, the broader measures of employment, most especially including all of the goods producing leading sub-sectors, were positive. And while aggregate hours stalled and aggregate payrolls almost certainly declined adjusting for inflation, they remained higher on a YoY basis.  The one more significant negative was that nonsupervisory wages continue to grow at a rate less than recent inflation. 

The inflationary AI data center Boom continues, while the rest of the goods producing economy appears recessionary

 

 - by New Deal democrat

As per usual, we start out the month with the ISM manufacturing report (for September) and the construction report (for August). Since these are two of the sectors that lead the economy, they give us a good first look at the remainder of the year.

The story mainly continues to revolve around the AI data center Boom. Manufacturing is expanding, but with widespread inflation, while construction in everything except AI data centers is recessionary.

Let’s start with the ISM manufacturing report (note: any number above 50 means expansion). The headline number declined -0.1 to 54.5. For forecasting purposes, I average the last three months, which comes out to 54.9:

The more leading new orders subindex rose +1.6 to 55.3, and the three month average is 55.2:

Both the headline and new orders numbers indicate continued expansion at about the same pace as earlier this year. This means the expansion in this sector can be expected to continue at least a few more months.

There was further if more subdued positive news in the employment subindex, which rose +1.5 to to 52.7. The three month average also rose slightly further into expansion at 52.2:

Where there was very negative news was in the prices paid subindex, which rose 6.8 to 77.9, close to its highs from earlier this year as well as during the immediate post-pandemic inflation, indicating very widespread price increases upstream. The three month average was 73.4, rising closer to its post-pandemic inflationary peak:

In short, the manufacturing rebound this year continues, but there are signs that producer inflation is accelerating again.

Keep in mind that for forecasting purposes, I weigh manufacturing at 25%, and the other 75% from the ISM services report, which will be updated next week, but has been running quite positive for many months.

A decidedly different story was told by construction spending. Nominally total construction (blue in the graph below) rose 0.9% in August, but remains down -1.7% YoY. The more leading residential construction sector (red) rose 1.1% for the month, but is down -4.8% YoY. Since the price of construction materials (not shown) rose 0.3% in August and was up 10.1% YoY, the increases were more subdued:

And in YoY% terms, “real” construction spending remains negative. In the below graph I’ve also included the YoY% change in construction materials costs discussed above (gold, right scale), since the monthly changes are not seasonally adjusted:

The news was “less dismal” in manufacturing construction spending, which rose for a change, up 0.2% for the month, but remains down -19.2% YoY and about 30% from its September 2024 peak:

Both residential and manufacturing construction remain recessionary, plain and simple.

Nonresidential construction, which includes but is by no means limited to AI data center related construction, rose 0.7% for the month, and is up 0.5% YoY:

But while unfortunately I can’t show you a graph, while total nonresidential construction spending nominally increased $6,543 millions YoY, AI data center related construction spending rose $5,943 for the month alone and was higher by $35,898 million YoY. In other words, take out the AI data center Boom and nonresidential construction is declining as well.

Basically the goods-producing sector of the US economy includes the AI Boom, which is continuing, and everything else, which is somewhere between outright recessionary to slightly expanding. And the expansion is very inflationary.

Jobless claims continue to decline near historic lows

 

 - by New Deal democrat

We’ll get important information on manufacturing and construction later this morning, but in the meantime let’s take our regular weekly look at jobless claims, which continued to be *very* positive.

Initial claims for the week declined -1,000 to 197,000, and the four week moving average declined -2,500 to 200,000. With the typical one week delay, continuing claims declined -11,000 to 1.701 million:

Again, these are in the vicinity of historical lows. In particular, continuing claims are not just the lowest in 3.5 years, but they are the lowest they have been since the early 1970s with the exception of 2018-19 and 2022 through early 2023.

On the YoY% basis more important for forecasting purposes, initial claims were down -13.4%, the four week moving average down -14.5%, and continuing claims down -11.5%:

Note that the YoY comparisons are becoming even more positive than they have been in the last 15 months, a very positive sign for the economy.

Finally, before the September jobs report tomorrow, let’s take our last look at where the unemployment rate is likely headed over the next several months:

Both initial and continuing claims are back down in their late 2022 and early 2023 ranges. This suggests that not only is the unemployment rate likely to decline to 3.9% or 4.0%, but it might even decline into the 3.7% range in the next few months.

August personal income and spending: suddenly, the downturn in real pesonal incomes has vanished; everything is sunny and upbeat

 

 - by New Deal democrat

Along with the employment report, real income and spending are probably the most important monthly reports as to the health of the average American household. But at the moment, for purposes of both the nowcast and the short term forecast, it is probably *the* most important set of data of all, along with stock market prices. That’s because personal consumption, which is about 70% of the entire economy, (at least until this morning’s drastic revisions - much more on that below) appeared currently to be driven by the wealth effect from stock market gains among the uppermost incomes. Unless something else comes out of the shadows to pick up the baton, if and when this wealth effect reverses, and if and when the average consumer pulls back, are the most crucial things to watch for.

Last month I opened my summary by saying that “This year there has been a real split between the income and spending sides of that ledger,” and further indicated that while real spending continued higher and was expansionary, “Real income declined, and continues being recessionary.”

This month that was entirely obliterated by revisions that went back over a year. Suddenly, instead of being negative and recessionary, we find out that consumers have really been basking in the sunshine of record high real incomes all this time. So, for the first time ever, to show you how dramatic the revisions have been, I am re-running last month’s graphs from the personal income side. And it is so jarring that, frankly, for the first time ever, I am waiting to see if there will be leaks from the bureaucracy about political meddling (but this report will give you the numbers straight up as always with a detached eye).

Here’s the more in-depth look.

Real Income:

Nominally income rose 0.2% in August, and was up 4.3% YoY. But after adjusting for the price deflator (blue), which increased 0.3%, they actually declined -0.1% for the month but rebounded to a gain of 0.8% YoY. Further, once we take government transfers into account (red), while the monthly change on a nominal basis was only +0.1%, and a decline -0.1% in real terms, but on a YoY basis they were also up 0.8%. This is in marked contrast with last month, which gave rise to the following graph:

Now here is this month’s updated graph that includes the revisions:

As you can see, the entire weakness since early 2025 is utterly gone.

Here is what the post-pandemic YoY% graph looked like last month:

And here is the updated graph this month:

The negative YoY numbers have vanished.

Real spending:

While the income side of the ledger was suddenly revised to the sunny side of the street, the spending side remained positive also. Nominally spending rose a sharp 0.9% and was up 6.1% YoY, which means that in real terms (blue) it was up 0.6% for the month, and up 2.6% YoY. As I’ve noted many times in the past, the leading indicator in this data has to do with spending on goods (red), as real spending on services (gold) frequently increases all the way through recessions. In August real spending on goods rose 1.3%, and was 2.7% higher YoY, while real spending on services rose 0.2% for the month and is up 2.5% YoY. As you can see, the trend in all three continues to be higher this year compared with last year:

Real spending on durable goods historically tends to peak even before goods spending as a whole. As shown in the below graph, real spending on durable goods (blue) rose 1.9% in August, while real spending on nondurable goods (gold) rose 1.8%:

All of these are at record highs. No sign whatsoever of any weakness here.

Again, the trend this year vs. last year is higher. On a YoY% basis (not shown), real spending on durable goods was higher by 4.6%, and on nondurable goods higher 2.5%.

Savings, real sales, and profits:

The difference between income and spending is what is saved. Once again, the revisions to income were important. Last month, the number was an increase from 2.6% to 3.0%. After revisions, this month for August, the saving rate declined  a sharp -0.5%, but revisions meant this was a decline from 4.6% to 4.1%. While this is still very low historically, the saving rate, while low, now is higher than much of the dotcom bubble era as well as the era of the housing bubble and in 2022 (note: graph subtracts -4.1% from the rate to show the current number at the “0” line for easy comparison):

To reiterate: unless and until we see a retrenchment by consumers indicated by a higher savings rate, the party goes on. 

Finally, this morning’s report also enables the update of real manufacturing and trade sales, one of the other important coincident markers used by the NBER to date recessions. This was unaffected by the income revisions. They increased 0.6% for July, continuing their uptrend to yet another record high:

They are higher 2.3% YoY (not shown).

To sum up: this morning’s report was an utter blockbuster; but one that relied upon heavy revisions to the last year’s data. While spending is in line with earlier reports, the big hit to real personal income which I have been reporting on all this year has suddenly vanished, replaced with substantial and continuing increases. I will look into the issue of where the revisions came from further, update as necessary if and when I find out anything substantial (probably not until next week).

August JOLTS report shows a labor market stabilized in a sideways low hire, low fire trend

 

 - by New Deal democrat

This morning’s other economic report was the JOLTS labor market report for August. This parses turnover in the market by hires, layoffs, and quits, among other things. It is a minor indicator, but let’s take a look.

The first graph below shows the “soft statistic” of job openings (blue), actual hires (red), and quits (gold), all normed to 100 as of just before the pandemic:

What is most noteworthy about this is that both hires and quits have had a sideways trend for the past two years - and that trend continued in August, as both numbers were basically right in the middle of that trend. Only the “soft” metric of openings has had a slightly increasing trend this year (although I didn’t run the historical graph, the fact is that outside of recessions, openings have had an upward trend for the past 25 years). Additionally, note both hires and quits have been running below their level of 2019.

Here is the same graph for layoffs:

Layoffs have been running at close to their lowest post-pandemic levels over the past nine months, and their declining trend since late last year is of a piece with what we have been seeing in the very low level of initial and continuing jobless claims each week.

Also, let’s update the comparison of the quits rate (blue, right scale), which has been suggested to lead YoY average hourly wages (red, left scale):

Like the number of hires and quits, for the past year the quits rate has been close to completely flat. That argues that *nominally* average hourly wages YoY should be stabilizing at a 3.4%-3.5% rate. In fact, if YoY average hourly earnings continue to decelerate in this Friday’s jobs report, that would call into question the above relationship.

But the general takeaway from this report is a jobs sector that has normalized in a low hire, low fire trend.

Repeat home sales price indexes continue recent trend of increasing YoY

 

 - by New Deal democrat

The current economic cycle has not just, as I explained yesterday, busted the infallibility of the inverted yield curve as an indicator, it has also blown up Prof. Edward Leamer’s theory “housing *is* the economic cycle.” But while no metric is perfect, it remains the case that housing is an important long leading sector, and house prices are an important component of the economy, particularly as they affect the official CPI measure downstream (not to mention their impact on ordinary buyers and sellers). And the repeat home sales indexes, by S&P Case Shiller and the FHFA, are the best indicator of prices in the 90% of the market that is existing home sales. same.

In the last few months, prices in both indexes have appeared to be firming, and that continued to be the case in this morning’s reports. After several months of decline, the seasonally adjusted Case-Shiller National index (blue in the graphs below) rose 0.1% for the three month period ending in July, while the FHFA index (red) rose 0.3%. Significantly, the FHFA index, which typically slightly leads the Case Shiller one, has been relatively “hot” compared to the latter. [Note: FRED has not yet updated the Case Shiller data]:

Earlier this year I noted that “there is something of a divergence showing in the YoY comparisons of the two national indexes,” as the Case Shiller national index had increased less than 1% YoY, while the FHFA Index had accelerated to a 2.0% increase. In the past several months, however, the Case Shiller index has also “warmed up” somewhat. In July, on a YoY% basis, the Case Shiller national index increased from 1.6% to 1.9%, a 12 month high, and the YoY% change in the FHFA index also increased from 2.3% to 2.6%, a 10 month high:

In stark contrast, as I wrote last week, the three month average of the YoY% change in the median price for new homes has declined to -2.6%, the biggest three month average YoY decline in two years. This is almost certainly a byproduct of homebuilders “meeting the market” while individual home sellers continue to resist taking losses (although I note a number of stories in the past month indicating that the percent of price reductions in existing homes for sale has been increasing dramatically).

Next, let’s take a look at how new (purple, right scale, averaged quarterly to cut down on noise (thick) and monthly (thin)) and repeat home prices (left scale) compare with households’ buying power, by adjusting for average hourly nonsupervisory earnings in the graphs below (median household income would be better, but is updated only once a year, and average wages are reasonably close for these purposes).

Last week I wrote that, deflated by average weekly earnings, the purchase price for new homes was less in “real” terms than at any point in the last 15 years except for one month during the COVID lockdowns. Applying the same deflator, existing homes as measured by both the Case Shiller and FHFA indexes, have become “less unaffordable” over the past 24 months. In July, average weekly earnings increased 0.3%, meaning that the “real” Case Shiller index declined further, while in “real” terms the FHFA index remained steady. The former has declined -4.6% in real terms since its recent peak in January 2025, while the latter has declined -3.1%. Nevertheless, as I wrote last month, it will take considerably more inventory on the market to bring existing homes down to just their average affordability compared with the past 30 years.

Finally, a reminder that the Fed’s interest rate hike and the similar increase in mortgage rates suggests that the recent equilibrium in the market is probably going to be yanked to the downside.

Another infallible economic indicator bites the dust: inverted yield curve edition

 

 - by New Deal democrat

No big economic data today. So let’s take a look at a significant forecasting tool. Some leading indicators are very good. But none is infallible. And the last few years have slain one of the most-touted for infallibility: the inverted yield curve.

For those who need a brief refresher, typically the longer the term of a bond, the more interest an investor will require for taking the risk of lending their money to the borrower for a longer period of time. Thus we typically expect a 30 year Treasury to yield more than a 10 year, and both to yield more than a 5 year, all three to yield more than a 2 year, and all of the above to yield more than a 3 month Treasury bill. That is called a normal, or regularized, yield curve.

But sometimes - typically but not necessarily always when the Fed is raising interest rates - the yield curve behaves abnormally, with some shorter duration Treasuries like the 3 month or 2 year durations paying more interest that longer duration Treasuries like the 10 year. That is called an inverted yield curve.

And for the past 50 years, it has been thought that an inverted yield curve means recession ahead.

Until now.

First, let me show you the historical record of the two most common measures of the yield curve: the 10 year minus 2 year treasury yield (blue) and the 10 year minus 3 month yield (red) since 1980:

The former measure doesn’t go back further, but the 10 year minus 3 month can be charted all the way back to the early 1960s:

So let’s summarize what we see. With the exception of the COVID lockdowns, which may never have been a recession otherwise, before every recession in the past 60 years, both measures of the yield curve above had inverted. About half the time, the 10 year minus 2 year had also re-normalized before the onset of the recession. And so in most cases had the 10 year minus 3 month comparison.

Astute observers will note that on one historical occasion there was a failure: during the 1966 “guns & butter” economy of LBJ, there was an inversion but no recession for another 3.5 years.

And, as I will document below, there has been a failure in this decade as well. 

The 10 year minus 2 year metric has almost always inverted first. The shortest period of time between inversion and recession has been 10 months (1980) while the longest has been 19 months (1990). The median period has been 15 months.

But the 10 year minus 2 year spread inverted in July 2022, a full 50 months ago (and counting). That’s almost triple the amount of time between the longest previous period of inversion to recession.

Similarly, going back to the 1960s, the shortest period of time between inversion of the 10 year minus 3 month spread has been 8months (2001) while the longest has been 25 months (1980). The median period has been 13 months.

But this decade the 10 year minus 3 month spread inverted in November 2022, 38 months ago, 1.5* more than even the longest such time in the past 50 years.

But, some have said, the real danger point is after the spread has re-normalized, pointing to all of the past times when a recession has begun after that normalization.

To begin with, about half the time no such re-normalization has occurred. So this metric appears to be random noise. But further, it relies upon an unstated assumption that the yield curve inversion is infallible. In other words, once there is an inversion, the recession will come either before or after a re-normalization, but it *must* come.

Well, certainly there is always another recession in the future *sometime,* but let’s see how this metric pans out as well.

The 10 year minus 2 year spread only un-inverted 3 times before recessions: by 2 months in 2001, 4 months in 2007, and 9 months in 1990. But this decade it un-inverted in August 2024, 25 months ago - almost 3* the maximum previous length of time.

And the 10 year minus 3 month spread un-inverted 7 previous times, with a duration of 1 month before at the least (2001) and 6 months before at the most (1990 and 2007). But this time around, measured monthly, it un-inverted in December 2024, 21 months ago, more than 3* the maximum previous duration. Even measured weekly it has been un-inverted for almost 12 months.

Only if we include the 1966 inversions and date them in comparison with the 1970 recession can we find any analogy, for the 10 year minus 3 month spread could be said to invert 42 months before the recession, and un-invert 25 months before. At present we are already past the former period, and only 4 months away from the latter. I submit that a “leading indicator” that takes almost half a decade to come to fruition is not an indicator at all.

So, if the inverted yield curve is busted as an infallible indicator, do I still plan to track it? Yes, because its past record is still good, and I use it only as part of a constellation of such indicators with good records that go back many years.

In which regard, let me give you the same graph, but over the past two years:

While the 10 year minus 3 month spread has continued to widen, the 10 year minus 2 year spread has narrowed again, most recently briefly to only 0.25%. Keep in mind from the above that the 10 year minus 2 year has almost always inverted first. And once it has narrowed to only 25 basis points, it typically has gone on in the near future to an inversion.

In other words, while the yield curve, like every other indicator, is not infallible, in a few months the clock might start ticking again.