Wednesday, November 18, 2026

The autumn skies have turned gloomy for housing

 

 - by New Deal democrat


The housing market, along with corporate profits and real retail sales per capita, is one of the three nonfinancial, or “real world” long leading indicators, that is of added importance for whether the US economy continues in expansion or falls into recession.

Earlier this year housing construction had been most consistent with an economy coming out of a recession than going in to one. That was derailed by the decision of the Buffoon in Washington to start the Iran war, which caused inflation, and interest rates to increase. In particular, mortgage rates came close to a new multi-year high last month:



Which, unsurprisingly, has led to a sharp decline in mortgage applications, housing sales, and new permits and starts. While this morning’s report indicated that the relatively noisy housing starts (blue) for October increased 9.4% to 1.309 million units annualized, the three month average was the lowest since this past winter. The less volatile and slightly more leading permits (gold) declined -7.4% to 1.295 million. The least volatile metric of all, single family permits (red, right scale) declined -8.7% to 772,000:



Total permits made a new post-pandemic lows, and perhaps even worse, single family permits are the lowest in almost 10 years except for the two months of the pandemic lockdowns and two months at the end of 2022. Further, permits are down -15.% from their recent peak in February this year, and single family permits are down -16.9% from their similar peak. 

All three of these metrics had been trending essentially sideways since the summer of last year. But this month, permit are down -8.9% and single family permits are down -12.9%. Only the volatile starts measure is higher, by 2.8%:



This doesn’t mean housing is forecasting a recession yet since, while I won’t bother with the graph this month, in the past these three metrics have generally been down -10% or more YoY to be consistent with the onset of a recession. Per the above, however, single family permits are down more than -10% from their recent peak, so this raises the first yellow caution flag that we are probably going to have to restart watching for a recession signal in the near future.

Now let’s update the several “last shoes to drop” in the housing sector. In particular, housing units under construction (blue), which had been giving a recovery signal earlier this year by rising from its lows, something that it typically does after the end of a recession, has now stalled out again, declining -1,000 to 1.272 million units annualized. And the even later shoe to drop, employment in residential construction (orange, right scale), may have reversed its recent increasing trend as well:



In summary, while real retail sales continue to sail on, and the outlook for corporate profits is only partly cloudy, to paraphrase an apt seasonal lyric from the late Gordon Lightfoot, for housing the proverbial skies have turned gloomy.


Tuesday, November 17, 2026

Despite a decline in October, the AI data center driven growth trend in production remains intact

 

 - by New Deal democrat


The Boom in AI data center construction has been the centerpiece of one of the most important trends of the last year; namely, the rebound in manufacturing. This has been reflected in the outsized but very volatile growth of the utilities component of production. 


That volatility continued in October, as headline industrial production (blue in the graph below) declined -0.4%, driven by a -4.3% decline in utility production (gold, right scale). Meanwhile, manufacturing production (red) increased 0.3%:



Note that the above graph is normed to 100 as of just before the pandemic, showing that manufacturing production is up less than 1% since then, while utility production, up nearly 10%, has driven the headline growth, up 2%.

On a YoY% basis, headline production is up 2.1%, manufacturing production up 2.6%, and even after October’s big decline, utility production is up 1.7%:



In short, this month’s decline is not significant enough to signal any change in the underlying trend. I would need to see a sustained break in trend in utility production for me to think that the AI data center driven Boom, or bubble, is ebbing.


Real retail sales decline slightly in October, but forecast continued slow employment growth

 

 - by New Deal democrat


As per usual, real retail sales is one of my favorite forecasting tools, since it tells us a lot about the consumer on a timely basis, and since consumer spending leads employment, and real retail sales per capita are a good non-financial long leading indicator.


On a nominal basis, retail sales have tended to be very positive most of this year, in part due to depressed comparisons post-“Liberation Day” last year. Although I won’t show the graph this week, they have tended to track the weekly data published by Redbook, which has shown YoY sales up in the 8%-9% range nominally.

To cut to the chase, October was weakly positive nominally but negative on an inflation-adjusted basis. Sales were higher nominally by 0.2% m/m and 6.8% YoY. But consumer inflation was up 0.4% m/m, so after rounding real retail sales declined -0.1%. Here’s what real sales (red), and real sales per capita (blue) look like for the past several years (courtesy of Steven Miller and the T—-p Administration, US population growth has slowed to only 0.2% YoY, one of the lowest such rates in decades, so this year the lines are virtually identical) :



Note that one month ago, September, was also their peak, so neither one is even close to forecasting a recession at this time.

With inflation estimated (due to the government shutdown one year ago) at 3.9% YoY, real retail sales YoY increased an estimated 3.0%, and on a per capita basis increased 2.8%. Note that the above graph, for the same reason, cannot show this month:



This remains very positive.

Since real retail sales tend to lead employment, here is the update on that relationship:



Again, this forecasts continued employment gains over the next few months. And again, almost all of this traces back to the AI data center Boom - or bubble.


Monday, November 16, 2026

If the AI data center Boom is a bubble, how quickly will the stock market and consumer spending follow any burst?

 

 - by New Deal democrat


I have written a lot this year about how the wealth effect from soaring stock market prices has been driving consumer spending, especially among the top-most incomes. It may be a Boom. But if it is a bubble, how soon after it bursts can we expect consumer spending to follow suit?


Unfortunately in the modern era for which we have decent statistics, there is only one example; namely, the dotcom bubble of the late 1990s. In that case, it appears that consumers did not react until it was clear that the bubble had burst, which took several months:



Which is important, because consumer spending, as shown in the YoY% change in real retail sales, below, typically peak on a per capita basis about a year before a recession, and turn negative YoY shortly before it begins:



There has been some choppiness recently, with some AI related stocks actually lagging the headline S&P 500 index (via Carl Quintanilla this morning):



And stocks are only up about 1% from six months ago:



Although on a YoY basis they remain higher by over 10%.

In the past, although it is a commonplace that the stock market anticipates corporate profits, on a quarterly basis the average YoY change in stocks has followed, rather than led, corporate profits. Since FRED isn’t allowed to post the full S&P 500 series, here is the NASDAQ composite going back 50 years compared with corporate profits:



Aside from the 1980s, typically (again, on a quarterly basis averaged), stocks have lagged corporate profits by about one quarter.

Here is the last 10 years, including the S&P 500:



So if the AI data center buildout is a bubble, it will probably show up in earnings first - to which the stock market will then react.

While “official” corporate profits as part of GDP for Q3 won’t be reported until the end of this month, over 90% of the S&P 500 have already reported earnings to Wall Street. Here’s what that looks like:



In Q2, not only did we have the effect of the AI Boom (or bubble), but we also had record profits in the Oil Patch thanks to the surge in prices due to the Iran War. At the moment, reported profits are down about -9% from Q2, although they are higher than any other preceding Quarter. Typically profit estimates for the next several quarters are lower than what the actual results will be - unless a recession starts, in which case all hell breaks loose!

Which brings me back to the current status of retail spending. Here’s what the recent YoY comparison looks like:



No recession risk showing there.

And here is the recent absolute value of real retail sales:



If the AI data center buildout is in fact a bubble, I would expect corporate warnings and substantial misses compared with estimates. Which the stock market will sniff out; and if the sell-off got serious and persistent, then (all else being equal) I would expect consumer spending to follow in short order. 

As to which, retail sales for October will be reported tomorrow.



Saturday, November 14, 2026

Weekly Indicators for November 9 - 13 at Seeking Alpha

 

 - by New Deal democrat


My “Weekly Indicators” post is up at Seeking Alpha.

This week’s theme is the same as the past few weeks: the long leading time horizon has become considerably more problematic, while the coincident present continues to feature strong production, transportation, and spending.

As usual, clicking over and reading will bring you all of the details in organized fashion and reward me a little bit for my efforts.



Friday, November 13, 2026

October producer prices confirm pressures spreading to services upstream

 

 - by New Deal democrat


I don’t normally pay that much attention to producer prices. Back in the day, producer prices used to lead consumer prices by several months or more. But since the turn of the Millennium, and in particular with “just in time” inventory management, somewhat attenuated since the pandemic, producer prices and consumer prices have risen more or less in sync.


But at very least producer prices as reported for last month confirm continuing inflationary pressures, with final demand prices (gold) rising 0.3% for the month, although raw commodity prices (red) declined by -0.2%, as compared with headline consumer prices (blue), which earlier this week were reported up 0.5%:



On a YoY basis, final demand prices were up 6.5%, and commodity prices up 12.8%, vs. an estimated 4.0% for consumer prices:



Decomposing producer prices further into prices for goods (purple) vs. services (orange), the former rose 0.5% and the latter rose 0.2%:



Note that on a monthly basis, for most of this year producer goods inflation is as bad as it was during the post-pandemic bottlenecks, while producer services inflation is up much less. On a YoY basis, the former is up 10.2%, and the latter up 5.0%:



Again, I don’t think this gives us a lot of predictive informatioin like it might have once upon a time. But at very least it confirms that there is widespread inflation at all levels, and to the extent there may be a leading/lagging relationship, I would worry that goods inflation has been spreading to services, and might further show up in the next several months in consumer prices as well.

Next week we’ll begin to get the November regional Fed reports, which will give us considerably more insight into whether further consumer price hikes are on the horizon.


The 3+ year equilibrium in the existing home market has broken - to the downside

 

 - by New Deal democrat


I’ll discuss this morning’s producer price index release later. First, though, let me catch up on yesterday’s report on existing home sales for October.

For the last three years, the market for existing homes has been rangebound, first in sales, then in prices, and finally this year in inventory as well, in what I have described several times as a “suboptimal equilibrium,” since existing homes have been unaffordable compared with any time outside of the housing bubble before the pandemic. Last month I concluded that, downstream of the increase in interest rates, September might be the last month for this equilibrium.

And it was, as existing home sales broke out of their three year+ range of between 3.85-4.30 million units annualized, declining a seasonally adjusted -5.5% monthly and -8.8% YoY to 3.75 million annualized, the lowest level in over 10 years:



Downstream of sales, prices on a YoY basis (which are not seasonally adjusted, so this is the best comparison) have been relatively calm as well. Since February of 2025 there has been no YoY comparison higher than 3.0%, and in October there was the lowest such comparison since, up only 0.4% YoY from $420,200 to $421,900: 



The most lagging metric, inventory, is still increasing, although tepidly. In October, the YoY% change in existing home inventories increased by 4.6%. As shown in the below graph, this year’s inventory has been the highest since the pandemic, but still below its 2016-19 baseline by about 500,000 units:



As I have been saying all year, until this inventory shortage is taken care of, the housing market will continue to be tight, and largely unaffordable for new buyers.

But the big takeaway has to be that the equilibrium has broken to the downside, along with most of the financial long leading indicators. It remains to be seen whether the crucial non-financial long leading indicators (corporate profits and real retail sales per capita) follow suit.