- by New Deal democrat
This morning’s important data on housing construction contained two apparently contradictory trends: on the one hand, it continues to be - even more intensely - consistent with an imminent or ongoing recesssion. On the other hand, it suggests that the sector is bottoming, meaning that (except for the lunacy of the T—-p Administration) a recovery should be near.
Let me address this by taking things in reverse chronological order, by which I mean focusing on the most coincident economic metrics first, and then working backward towards the most leading.
The most important of the more coincident (really, short leading) aspects of the report is housing units under construction. These both peak and trough significantly after permits and starts. These declined another -11,000 to 1.365 units annualized, the lowest number in over 5 years, and -26.1% below their 2022 peak (red in the graph linked to below), which also shows the same metric in YoY% terms (blue):
Only once in the past 50 years has such a decline not *yet* given rise to a recession: in 1991, when the decline was -28.2%. On the other hand, while units under construction are down -9.6%, this is less of a decline than one year ago, when the YoY decline was -15.9%. With the important exception of the 1991 recession, when the YoY decline became “less bad,” it has typically occurred after the recession has ended.
Typically, the final housing-related metric to decline before a recession actually begins has been employment in housing construction (red in the graph linked to below):
As you can see, this has indeed happened, but only by -1% (vs. a more typical -5% decline prior to past recessions), and although you will need to zoom in on the above graph, that is about a 1% improvement since last August — again, a recessioin marker on the one hand, but a potential sign of incipient “green shoots” on the other.
Now let’s turn to the more leading permits and starts. The latter (blue in the graph linked to below) are noisier and slightly less leading. These increased 100,000 annualized for the month to 1.487 million, and they are 9.5% higher than they were 12 months ago, although the below graph really shows them as being mainly volatile with an overall flat trend. Permits (gold) are less noisy and more leading, and these declined -79,000 to 1.376 million, and were down -5.8% YoY. The trend here has been slightly negative, as the current reading is only higher than last July and August’s. Finally, the least noisy indicator of all is single family permits (red, right scale), and these declined another -8,000, but are higher than they were in late 2022 into 2023, and in June and August last year:
Nevertheless, as the graph shows, permits and starts remain in territory below their peaks sufficient to be consistent with a recession. On the other hand, in the past it has taken a more severe decline of greater than -10% to be consistent with with a recession:
Only single family permits, down -11.6% YoY, meet this criterion.
Finally, the most leading component of all is mortgage rates, currently just above 6%, just above their 3.5 year low of 5.98% made last month:
Unsurprisingly, the gradual decline of mortgage rates from their 2023 highs (with a peak of 7.79%) has led to an increase, albeit a tepid one, in mortgage applications over the past 15 months:
Again, the decline in mortgage rates has given rise to “green shoots” in purchase mortgage applications.
So let me sum up: in the longest leading data, we see signs of *relative* easing, which has given rise to some improvement off the bottom in mortgages (in 2023), mortgage applications (in 2024), and permits (in summer 2025). But among the less leading data, in particular housing units under construction and employment in housing construction, we see declines consistent with a recession now, or imminent, along with signs that, if there were not other complicating factors (like the Iran war), any such recession if it occurred at all would be a short and shallow one.
