2023-03-08 15:12:31 ET
Summary
- Housing is typically an important market by which the Fed’s interest rate hikes get translated into a slowdown.
- But despite steep Fed rate hikes, housing under construction remains just under record levels, and residential construction employment has continued to increase.
- The logjam which started with pandemic-related supply bottlenecks resulted in record backlogs of housing authorized but not started, leading to a delay in the peak of housing under construction.
- The situation is magnified because of the near-record share of multi-unit dwellings under construction vs. single dwelling units.
Introduction
Last week, I wrote how the logjam in the Fed's housing transmission mechanism was one of three big reasons that no recession has begun. Let me elaborate on that, while I give a general update on the housing market.
Interest rates lead sales, which lead prices
The above mantra is something I have written many times in the past 10 years. Let's update the data.
Long-term interest rates, which are reflected in mortgage rates (inverted, *10 for scale in the graph below), lead housing sales, in the below reflected by permits, by an average of 3-6 months. I have also included 10-year Treasury yields (inverted, *10 for scale) to show they generally track closely with mortgages:
Mortgage rates vs. treasuries and housing permits YoY ((FRED))
It takes a number of months after sales go down for sellers to get the message. This is something I was pounding away at all last year, and within the last few months, it has shown up in the data, as shown by the YoY% change in house prices as measured by the FHFA index:
Housing permits vs. FHFA housing price index ((FRED))
To summarize: because we already know that mortgage rates are significantly higher than they were a year ago, we can expect the YoY comparisons in housing permits to continue to suffer, and further declines in house prices in the year ahead beyond the slight ones which have already begun.
A closer examination of the transmission logjam
Monday morning, Nobel Laureate Economist Paul Krugman tweeted the following:
Housing starts vs. construction employment (FRED via Twitter)
But, far from puzzling, it's an easy "mystery" to solve. As I wrote last week, the pandemic-induced supply chokepoints in housing materials like lumber led to a huge backlog in housing units which were authorized but not started. Thus, even though permits went down promptly as mortgage rates rose (blue in the graph below), followed shortly thereafter, as is typical by starts ( GRAY ), it took until September for housing units permitted but not started (red) to catch up, and housing units under construction ( GOLD ) until November:
Housing permits, starts, units not started vs. under construction ((FRED))
To elaborate on the point: in the past, construction has peaked with very variable lags after permits:
Housing permits vs. units under construction ((FRED))
Next, let's compare housing authorized but not started, vs. housing units under construction, first for single dwelling units:
Single-family housing not started vs. under construction ((FRED))
And here for multi-dwelling units:
Multi-unit housing not started vs. under construction ((FRED))
Note two things: (1) construction normally peaks only *after* units authorized but not started has already peaked; and (2) it normally takes longer for multi-dwelling units under construction to peak after units authorized but not started, vs. single-dwelling units.
So, the progression in the Fed transmission mechanism through permits, then starts, then units authorized but not started, and only then to units under construction, has a long historical track record.
Because housing under construction is the actual economic footprint of the housing sector, it is only once construction goes down that housing has a depressing impact on the economy; thus explaining much of the lag time between Fed rate hikes and their dampening effect on the economy as a whole.
When will construction go down significantly?
As I noted last week, housing units under construction are only down -0.6% from their peak last October. With the exception of the pandemic and the 2001 recession, in the past recessions have only begun once construction is down at least 5%, and typically 10% or more:
Housing units under construction ((FRED))
Further, as noted above, single-family units typically are completed more quickly, so such units under construction typically turn down before multi-unit construction, sometimes quickly, and sometimes with a significant delay (Note: shown YoY for ease of comparison):
YoY% change in single-family vs. multi-unit construction ((FRED))
That is particularly significant because the number of multi-unit developments is presently at a record in absolute terms and also (not shown) close to a record share of the market, as compared with single-family units:
Single-family vs. multi-unit housing under construction ((FRED))
So while single-family units under construction have gone down -9% since peaking last April, multi-unit developments have continued to rise by over 12%.
It appears in 2022 the turn in construction began to take place in October. Recessions have begun as few as 6 to 9 months thereafter, and as long as 27 and even 47 months(!) later.
Conclusion
So, to answer Professor Krugman's question, only once construction has peaked, as it did in October, will employment in construction - a leading sector for payrolls as a whole - go down with a lag:
Housing under construction vs. employment ((FRED))
While the process has begun, initially because of the supply chain bottlenecks during the pandemic, and the record number of multi-unit developments being built, it is likely to take longer than usual. And it is only once that happens that the logjam in the Fed's tightening transmission mechanism through housing into the broader economy will be broken.
For further details see:
A Closer Look At The Logjam In The Fed's Housing Transmission Mechanism