Thursday, April 25, 2019

How increasing local oligopolization has distorted the housing market


 - by New Deal democrat

Earlier this week new home sales for March were reported, soaring to a new expansion high bar one month (November 2017). Something else that a few other writers picked up on: the median *prices* for new homes fell to a level not seen in the past two years, off -11.8% from their peak, also in November 2017: 


With mortgage rates also down at approximately where they were in January 2018, the carrying cost of a new house has declined by over 10% overall, enticing lots of new potential buyers into the market.

All well and good. But my reaction went a little beyond that: “Holy crap! Builders can slash their prices by almost. 12% and still make a profit?!?” 

That kind of pricing power smacks of a market that isn’t competitive.  In a truly competitive market, the kind of big price increases we saw until 2018 — at sales levels well below any time in the decade between 1995 and 2005 — would have called forth new supply at lower prices. It turns out, I’m not the only one who thought that. Two economists, Jacob Cosman and Luis Quintero, presented a paper, Fewer players, few homes: concentration and the new dynamics of housing supply, to the American Economic Association last winter, documenting just how much local oligopolization has distorted the market.

Let me cut to the chase. Here are the two graphs that show their primary points. The solid line in the first graph shows the number of firms that account for 90% of all homebuilding in the median local market (dotted lines are the first and third quartile of local markets):


This has declined from 6 to 4 firms in the decade since the bursting of the housing bubble.

The second graph shows the share of production accounted for by the largest 3 and 5 firms in typical housing markets:


Each has increased by roughly 10% since the bursting of the housing bubble.

The authors conclude that:
“[compared with] a counterfactual scenario where housing market competition remains at its high pre-recession level across the United States[, ...] market outcomes [are] very different. The annual level of new housing would be $106 billion higher (equivalent of 3.4% of private fixed investment or 0.6% of gross domestic product. Approximately 150,000 additional housing units would be built each year. Housing price volatility would decline by over 50%.”
To give you an idea how dramatic an effect this market concentration may have had on housing in the past decade, here is a graph showing the typical pricing premium for a new house vs an existing house: 



Historically, the premium was 10%. Since the bursting of the bubble, however, that premium went up to 30%! Even now, as of March 2019, although it has closed somewhat, the premium is still 14%:


 

An alternative explanation that has been offered for the decline in median house prices since the end of 2017 is the square footage of the median new house that builders are offering is smaller. But a comparison of median prices and median square footage suggests that, while there is an effect, the market power of local oligopolies appears to be the primary driver in both the big increase in new home prices through 2017, and the big decline since.

Here is a graph of the median sales price for new houses (blue) and the median square footage of new houses (red) on a quarterly basis. Each is normed to 100 as of their respective peaks:


As you can see, house prices continued to rise by over 15% (!) in the nearly 3 years after the median square footage of a new house began to decline after Q1 of 2015. This certainly *isn’t* a decline in square footage driving a decline in prices!

Further, there’s very little evidence that the rate of decline in square footage accelerated after prices peaked in Q4 2017, as shown by the same data graphed as YoY% change:


The square footage of the median new house has varied between unchanged and -3% in the past four years, both when prices were still increasing, and since prices started decreasing. For some reason FRED hasn’t gotten around to posting the Q4 2018 data for median house sizes (the last quarter for which data is available), but even that is only -2.4%. During that same period (Q4 2017 - Q4 2018) median house prices declined -3.9%. At the most, the decline in square footage explains no more than 60% of the decline in house prices.

An alternative explanation — and mind you, I am speculating here, I don’t have data — is that builders have been increasing their profits by decreasing lot sizes, thereby increasing the number of single family houses they can build on any given footprint of land.

Note further in the second graph above that median house prices appear to lead median square footage by about 1 to 2 quarters, suggesting that it is the change in prices which is driving the change in square footage, rather than changes in square footage driving prices.  A good test of this hypothesis will be if median square footage declines by more than -3% YoY by the 3rd quarter of this year.

The bottom line is that the increasing concentration, and attendant local market power, among home builders has distorted the housing market in the past decade, increasing substantially the cost of the typical new home while holding down the number of units constructed. Because new homes have been so much more expensive, this distortion has also been an important reason for the relative unaffordability of existing homes compared with historical norms, and by keeping both new and existing home prices higher than they would otherwise be, also been an important reason behind soaring rents, which are also at all time highs compared with median renters’ incomes. (UPDATE: In that vein, median asking rent for Q1 2019 was just reported, up +4.4% q/q and +5.5% YoY, continuing to show pressure in excess of income growth.)

Wednesday, April 24, 2019

Commercial and industrial loans: another sign of a slowdown?


 - by New Deal democrat

There are lots of cross-currents in the economy right now. At the absolute tip of the spear is the decline in interest rates since November, which has led to an improvement in some of the housing market metrics. In the shorter-term outlook, a simple quick-and-dirty metric of initial jobless claims (new 49 year lows) and the stock market (just made new all-time highs) suggests all clear. But there are contrary signs as well. For example, the weekly measure of temporary jobs by the American Staffing Association just fell to -1.8%, its worst YoY comparison since the 2015-16 shallow industrial recession. 

Here’s one other little tidbit. Yesterday I read an article elsewhere about how a near-term recession isn’t in the cards, citing among other things a declining delinquency rate for commercial and industrial loans. Here’s their accompanying graph:


True enough, although if you look carefully, in the lead up to both the 1990 and 2008 recessions there were only two quarters of significant increases off the bottom before the recessions began. Since the latest data in the graph is for Q4 2018, a similar pattern wouldn’t rule out a recession beginning as soon as Q3 of this year, i.e., July.

The value of commercial and industrial loans is a lagging indicator, typically not bottoming on a YoY basis until after a recession is already over:


But here’s the interesting thing. The same graph shows that YoY volume of such loans significantly *decelerated* from its YoY peak before 8 of the 11 recessions since 1945, as well as for all of the significant slowdowns, e.g., 1966 and 1995.

No big deal, right? At the far right of the graph, the YoY volume of loans was increasing as of March.

Except for this: when we zoom in on the recent weekly, rather than monthly data, we get this:


A very sharp deceleration over the past three weeks.

Of course, this could reverse with this Friday’s report. And since loans tend to follow bank tightening by about 5 quarters:


I tend to doubt the deceleration will go too far.

But nevertheless something to keep an eye on in terms of another sign of a slowdown.

Tuesday, April 23, 2019

New home sales suggest housing bottom is in


- by New Deal democrat

New home sales are extremely volatile, and extremely revised, but they do have the advantage of probably being the single most leading housing statistic, ahead of permits and starts.

So it is noteworthy that new home sales for March rose to 692,000, below only one month in late 2017 when they hit their expansion high of 712,000:



I have been looking for the bottom in housing, as mortgage interest rates have fallen in the past 5 months, and purchase mortgage applications have risen to new expansion highs:




Subject to revisions(!) —this morning’s data indicates we’ve already made the bottom in new home sales, and adds to my confidence that we either have just made or will shortly make the bottom in housing permits and starts.

My guess is that interest rates have now stayed lower long enough for new expansion highs to be set in all of the metrics, although because of continuing increases in the price of houses, there is not much room for advances beyond that.

Monday, April 22, 2019

When will residential construction employment start to decline?


 - by New Deal democrat

Because the long leading indicators turned down in 2018, over the last few months, I have repeatedly looked at the leading employment sectors of temporary jobs, manufacturing, and construction, to check for that weakness feeding through. Today I am taking a closer look at construction jobs.
To begin with, while construction jobs as a whole do lead, residential construction jobs have been the most leading sector of construction jobs, as shown in the graph below:  

Although, interestingly, construction employment is down slightly from January, while residential employment has continued to increase.
Next, here’s a comparison of the YoY% change in single family permits (blue), residential construction spending (red), and residential construction employment (green), averaged quarterly to cut down on noise:

To cut to the chase, the peak in YoY construction spending follows permits with usually a one quarter lag. Residential construction employment, in turn, follows spending with one more quarter lag. Both permits and spending are down YoY as of the most recent data. Employment is merely decelerating.
So when might we expect residential construction employment to turn down meaningfully? I next broke out permits vs. housing completed, and compared both of those to residential construction employment. Here’s what I got.
Measured by single family units:
Measured by total units:

In both cases, residential construction employment coincided most closely with housing completions.

 It appears that, usually, employers keep employees on the books even as units under construction begin to decline, until the decline goes all the way to completions:

Note that in the 1980s, this most closely tracked single family homebuilding. It  be tracking total units more closely now.
In any event, here is a close-up on total housing completions through March:

This may explain why residential construction employment has not turned down meaningfully yet.
Finally, I looked at historically how long it took after units under construction peaked for completions to peak:


In the past 50 years, with the exception of one outlier - 1978, in which completions actually led by 4 months - the peak in units under construction has led the peak in completions by between 1 and 5 months.

Note that units under construction, for now, last peaked in January. Completions may have peaked in February.That means, if January was the peak for units under construction, we should expect completions, and residential construction employment, to turn down meaningfully no later than June.




Sunday, April 21, 2019

Nailed it!


 - by New Deal democrat

Three weeks ago I wrote No, the Meuller report ***DID NOT*** “find no collusion!” in which I lambasted and parsed Barr’s conclusory snippet of the Mueller report, to wit, that “[T]he investigation did not establish that members of the Trump Campaign conspired or coordinated with the Russian government in its election interference activities.”

I pointed out that: 
 ... the bracketed [T] in Barr’s quote of Mueller is doing a lot of work. Because it means that there was a first part of the sentence that was omitted. Put that together with the fact the Mueller’s quote then specifically references that “the investigation did not establish ...” and there is compelling evidence that the first part of the actual sentence was a qualifier. .... Almost certainly the first part of the sentence is something like “Although...’” “Since ...’” or “Despite ...” followed by “the investigation...”,  or a formulation like “The grand jury’s work is incomplete, and so the investigation ...”
(Emphasis added)
Now that we have (most of) the actual Mueller report, we know that the complete sentence reads:

Although the investigation established that the Russian government perceived it would benefit from a Trump presidency and worked to secure that outcome, and that the Campaign expected it would benefit electorally from information stolen and released through Russian efforts, the investigation did not establish that members of the Trump Campaign conspired or coordinated with the Russian government in its election interference activities.”
(Emphasis added)
Exactly as I thought, and said. The first part of the sentence Barr quoted severely qualified the portion he chose to highlight. 

I also wrote:
While the “no finding” formulation is consistent with a “finding of no collusion,” it is also consistent with other readings: 
 1. The investigation isn’t complete yet (which is almost certainly a correct statement). 
2. The evidence is inconsistent, weak, or contradictory. 
3. There are too many unknowns to come to a conclusion. 
4. While the evidence of collusion is strong, it is not strong enough to support a jury verdict beyond reasonable doubt.
 Mueller’s report makes clear that, first of all, he made “no finding” as to the narrower question of criminal conspiracy, which requires an actual or tacit agreement, rather than encouragement and coordination, I.e., “collusion.” Further, he explicitly qualifies his “no finding” by noting gaps in the evidence, in the form of witnesses who refused to testify under oath, and/or deletions of crucial  communications  Mueller’s report leaves open the possibility that the conclusion could change if the missing evidence were provided.