- by New Deal democrat
As I reiterated the other day, housing is the one long leading signal of the economy that has been soundly negative - indeed frankly recessionary - for the past year. That trend continued in March, although there are signs of it bottoming out. Without having caused a recession.
Let me start with mortgage rates (blue, left scale) compared with single family permits (red, right scale). Typically housing permits and starts follow mortgage rates, and for most of the past several years they had been declining from their post pandemic high of over 7%. In February they made a new 3+ year low of 5.99%. The decline was sufficient to reverse the trend in single family permits (red, right scale), which began to rise from their bottom early last summer:
Of course mortgage rates did rise in March, and this appears to have put a damper on permits (blue in the graph below) which declined -166,000 to 1.372 million annualized, the second lowest reading since 2020. On the other hand, the more noisy starts (gold) rose 146,000 to 1.502 million annualized, the highest in 15 months. But the metric that is the least noisy of all and conveys the most signal, single family permits (red, right scale), declined -35,000 to 895,000 annualized, nevertheless continuing its general uptrend since the beginning of last summer:
On a YOY% basis, starts are (noisily) up 10.8%, while permits are down -7.4% and single family permits are down -7.9%. Typically all three have been down 20% or more at the onset of recessions in the past, although in the 1991 and 2001 recessions, they were only down about -10%:
Note that the downtrend has not been worsening for many months. Further, there have been a number of times, for example 1966, 1987, and 1995, where construction has been down -10% or more YoY without a recession occurring.
Let’s turn next to the number of housing units under construction. As I have written many times in the past several years, it is the best “real” measure of the economic impact of housing (blue in the graphs below). In March they were unchanged at their five year low of 1.264million annualized. They are also down -26.3% from their peak:
The above graph shows how they have followed single family permits (red), as expected. More often than not in the past by the time a decline in units under construction had declined by this much, a recession had already begun. The only two exceptions were the late 1980s, where the pre-recession decline was -28.2%, and 2007, where the pre-recession decline was -25.6%.
Now let’s compare housing units under construction with the typical final shoes to drop before recessions, houses for sale (gold) and residential construction employment (red), all normed to 100 as of their respective post-pandemic peaks. Uniquely, employment in residential construction has risen slightly since last August, even though the number of units under construction has continued to decline. Meanwhile housing units for sale have continued to decline:
Now here is the same data presented in YoY% change format:
In the past 40 years, houses for sale and employment in residential construction had turned down YoY before the recessions had begun. After making a trough at -1.4% YoY, residential building employment is now only down -0.4%; while housing units for sale are down -4.0%, the weakest YoY reading since summer 2023.
So we end with a quandary. With the sole exception of employment (which may be reflecting, and distorted by, the ICE raids on construction workers), all of the indicators in the housing sector are giving classic signs that a recession should be underway — and they have been since late last year.
And yet there are a number of signs that the situation has been bottoming, without a recession having occurred. As I wrote in my note earlier this morning about new capital goods orders, this may be because the Boom in AI data center construction and operation has more than overbalanced declines in things like real income, or real personal spending on goods, and the complete stall in employment growth since early last year. We should get more information about this tomorrow from the release of personal income and spending in March, as well as Q1 GDP.
