Final August durable goods orders: more evidence that the AI related buildout is keeping the economy afloat

 

 - by New Deal democrat

Oddly, even though the government shutdown has crippled most reporting, the Census Bureau did update the durable goods report for August today, and it does reveal an essential bifurcation in the state of the economy.

First, here are the topline leading indicators: total durable goods orders (blue), core capital goods orders (red), and consumer durable goods orders (gold, right scale):

The difference between the first two and the last one is striking, and even more descriptively so when we measure them YoY:

As of the last report, consumer durable goods orders were dead in the water, down -0.1% YoY. By contrast, headline durable goods orders were up 7.6% YoY, and core capital goods orders up 4.0%.

The difference between the core and headline numbers is almost all driven by transportation orders (i.e., Boeing) (blue), not defense (red):

And a more detailed breakdown of new orders by industry reveals that the biggest increases have been computer-related (red) and communications (gold), although other sectors have grown significantly as well:

This is yet more evidence that the AI buildout in data centers and electronics updates are the driving force behind the continued expansion in the economy, fueling outsized stock price gains in the sector (and providing the fuel for wealth effect consumer spending); while the larger consumer sector is not growing at all.

EDITED TO ADD: And just to put an exclamation point on the data, here is Joe Wiesenthal with a breakdown of spending on AI data centers and everything else:

A note on initial jobless claims

 

 - by New Deal democrat

It’s Thursday, which typically means it’s time for the weekly initial and continuing jobless claims update. As we all know, the Federal government has shut down, because there is no budget for this fiscal year that began yesterday.

BUT, I went back and checked, and during the extended government shutdown in 2013, initial jobless claims were reported, and they were also reported at least for one week during the 2018-19 shutdown as well. 

Which makes sense, because they are simply compiled by adding up reports from the 50 States plus DC and Puerto Rico, and then using a seasonal adjustment.

But there has been no report this morning. 

As usual, my impulse is to find a workaround, which in this case is to get the raw data from the various States’ Departments of Labor or similar.  While I have neither the time nor the patience to do this for all jurisdictions, if enough of the bigger States publish their information, we should be able to arrive at a fairly reasonable estimate.

So far this morning none of the States I have checked, have updated their information for the last week. I will check later today, and if by then they have updated with claims through the week of September 27, I will follow up with a second post. Here’s hoping . . . .

In the meantime, via Carl Quintanilla, here is the latest graph on how the AI Boom (or bubble) in the stock market is fueling spending by the top 10% of consumers, which is what is keeping the economy afloat:

ISM manufacturing index continues to show slight contraction, as new orders retreat

 

 - by New Deal democrat

Typically the new month begins with important manufacturing and construction reports; but this morning the construction spending report for August became the first casualty of the government shutdown. 

The ISM manufacturing report has been a recognized leading indicator for the past 60+ years, although of diminished importance since the turn of the Millennium and China’s accession to regular trading status. While any number below 50 indicates contraction, the ISM itself indicates that the number must be under 42.8 to signal recession. 

Because of the report’s diminished importance, for forecasting purposes, I use an economically weighted three month average of the manufacturing and non-manufacturing indexes, with a 25% and 75% weighting, respectively. That briefly justified a “recession watch” during the summer, before the strong August rebound mainly in the services sector.

Today’s report continued the string of contractionary readings, although it rose slightly to 49.1. The more significant news is that the more leading new orders subindex, which rebounded to 51.4 in August, sank back into contraction at 48.9. Here is a look at both the total index (blue) and new orders subindex (gray) for the past three years (via Tradingeconomics.com):

Note that both remain slightly better than their low points in 2022-23, which is noteworthy because there was no recession then.

Hare the last six months of both the headline (left column) and new orders (right) numbers:

APR 48.7. 47.2

MAY 48.5. 47.6

JUN. 49.0. 46.4

JUL 48.0.  47.1

AUG 48.7. 51.4

SEP. 49.1. 48.9

The current three month average for the total index remains at 48.6, while the new orders rose slightly to a still contractionary 49.1. This is in accord with the recent regional Fed reports, which turned positive during August, but retreated somewhat in September.

As I indicated above, for the economy as a whole the weighted index of manufacturing (25%) and non-manufacturing (75%) indexes is more important. In the non-manufacturing report, the average of the last two months for the headline and new orders numbers has been 51.0 and 53.2, respectively. Pending the ISM report on services next Monday, the economically weighted headline number is 50.4, and the new orders average is 52.2.

If the ISM services report next week does not indicate any further downturn, this means that the economy as a whole continues to expand, albeit just barely.

August JOLTS report was weak, but foreshadows little

 

 - by New Deal democrat

In the past year, in contrast to much other data in the jobs sector, the JOLTS reports have been very much consistent with a “soft landing” jobs scenario. In the August report released this morning, the trend weakened slightly.

As a quick refresher, this survey decomposes the employment market into openings, hires, quits, and layoffs. So to begin, here are job openings, hires, and quits all normed to 100 as of just before the pandemic:

I regard openings are “soft” data. While they have trended down for several years, they have remained above their pre-pandemic levels, and are not of much concern to me. They improved slightly this month. The trend for the past 15 months has been flat to slightly downward. Meanwhile both openings and quits declined, the latter to the lowest level since December 2024, but the former came in at the lowest level since June 2015 except for June 2024 and the pandemic lockdown months! These are both “hard” data, and were both weaker readings.

Now let’s look at several components are slight leading indicators for jobless claims, unemployment and wage growth.

Layoffs and discharges, which have trended slightly higher since last summer, but have been rangebound since last autumn, remained so again, although the three month average was the highest in the past 12 months:

This generally accords with both the increase in the unemployment rate in 2023-24, as well as its plateauing this year (red, right scale), as well as the recent trend in continuing jobless claims.:

Next, the quits rate (left scale) typically leads the YoY% change in average hourly wages for nonsupervisory workers (red, right scale):

In August the quits rate declined slightly, tying its lowest in the past 12 months. This suggests that nominal wage growth may decelerate slightly further in the next several months.

Finally, I want to discuss why I don’t pay much attention to the “Beveridge curve,” which is the relationship between job openings divided by the number of unemployed, and the unemployment rate.

In the past several months there has been some modest hysteria about the number of unemployed exceeding the number of job openings (i.e., the ratio has fallen below 1:1), leading to speculation that the unemployment rate will increase.

But the historical view shows that there is no magic ratio of the Beveridge curve which is consistent with rising or falling unemployment. Rather, it is the *trend* in openings vs. the number of unemployed which has generally correlated with the *trend* in the unemployment rate. As shown below, with the Beveridge curve inverted for ease of comparison, the ratio was *always* below 1:1 for the entire period before 2018, and yet there were two extensive recoveries during which the unemployment rate declined:

Now here is the post-pandemic view. Again, we see that the *trends* correlate well, but there is nothing magic about the 1:1 level:

Just like the layoffs and discharges metric, this suggests that the unemployment rate may increase slightly. 

In short, this was a weak report. But it was a report for August, and we already have the August jobs report. It tells us very little about what to expect for September (if it is released, given the likelihood of a government shutdown before then), except that possibly the unemployed ent rate may increase.

Repeat home sales as measured by Case Shiller and the FHFA confirm price deflation

 

 - by New Deal democrat

Last month the indexes of repeat home sales from the FHFA and S&P Case Shiller were the final confirmation that the housing market was in deflation.

That continued in this morning’s reports for July. On a seasonally adjusted basis, in the three month average through June, both the Case-Shiller national index (light blue in the graphs below) and the FHFA purchase index declined -0.1%. The peak for the FHFA index (blue in the graphs below) was in March, while that the Case-Shiller Index (gray) was in February. (note: as per usual, FRED hasn’t updated the FHFA information yet):

The actual *de*flation in the house price indexes has been -0.7% in the FHFA Index and -1.1% in the Case Shiller Index:

On a YoY basis, price gains in both indexes continued to decelerate, at 1.7% for the Case Shiller index, and 2.3% for the FHFA index; but these were the lowest YoY% increases since 2012 for both indexes excluding 5 months in 2023 for the Case Shiller index:

Because house prices lead the shelter component of the CPI by 12 - 18 months, this also indicates that they will continue to decelerate over that period. Here is the same graph as above (/2.5 for scale) plus Owners’ Equivalent Rent from the CPI YoY (red):

The last time the Case-Shiller and FHFA Indexes were in this range, excluding the Great Recession, was in the 1990s, during which time Owners Equivalent rent was in the 2.5%-3.5% range (vs. 4.1% as of the most recent CPI report).

Similarly, the latest “National Rent Report” from Apartment List for August, released yesterday, showed a decline of -0.8% YoY. While this is only new leases, and the combined experimentsl all rent index published by the Census Bureau does not show a decline yet, after two years of YoY declines the trend continues to be that of sustained deceleration:

My conclusion this month is the same as that of last month: all phases of the housing market are either at or near their low points (sales, permits, starts), or declining (prices, construction, employment, and new spec units for sale). What is different is that the manufacturing side of the goods-producing sector has shown renewed vigor in the last couple of months. If that fades as well, it is hard to see how we avoid a recession in the next 12 months.