Saturday, December 24, 2022

Weekly Indicators for December 19 - 23 at Seeking Alpha


 - by New Deal democrat

My Weekly Indicators post is up at Seeking Alpha.

Coincident indicators continue to ever so microscopically worsen - but not yet in recession territory; while there is an increasing suggestion from the long leading indicators that a recession could be relatively short. Provided, of course, that the Fed takes the hint.

As usual, clicking over and reading will bring you up to the virtual moment as to the economic situation, and reward me slightly for my efforts.

Also, this programming note: Merry Christmas to all who celebrate! There will be a few economic releases in the next week, but don’t be surprised if I take a few days off.

Friday, December 23, 2022

New home sales for November: at last, a bright spot! (relatively speaking)


 - by New Deal democrat

New home sales are very volatile, and heavily revised. But they frequently are the first housing metrics to turn. And November’s new home sales report suggests that they may indeed have made their low.

Last month new home sales increased to 640,000 annualized, from a downwardly revised 607,000 (vs. the original 632,000) in October. Here’s what the past year looks like (FRED hasn’t updated, so here’s the Census Bureau’s graph):

The preliminary read for November is the highest number since March, with the exception of August’s 661,000.

Now, a word of caution, but also a word of caution *about* that caution: we know that cancellation rates for new home contracts have increased sharply since springtime. So, that has made the *actual* sales numbers worse than the reported numbers. BUT, even taking them into account, as of October, the low point was July. I don’t have this month’s number for cancellations, so that might change. But also, there is no reason to think that there weren’t similar levels of cancellations during any of the other historical housing downturns brought about by increased mortgage rates. In other words, new home sales this year should have a comparable pattern to previous downturns.

Finally, YoY prices were up 9.5% (this data is not seasonally adjusted) (again, FRED hasn’t updated, so here is the YoY% change through October:

Since this is less than 1/2 the highest % growth in the past 12 months, per my heuristic this indicates that house prices, if we could seasonally adjust, have actually started to decline.

As I’ve mentioned a number of times recently, I am on the lookout for long leading indicators that might suggest how long (or short) a recession we might be in for. At the moment, new home sales is suggesting the downturn may not be that long (Fed willing, of course).

Real personal income and spending hold up (thank you, lower gas prices!) but still consistent with onset of recession


 - by New Deal democrat

This morning’s report on personal income and spending for November shows why I pay more attention to real retail sales as a forecasting tool.

First, to the data: personal income increased nominally by 0.3% in November, while nominal spending increased only 0.1%. Since the deflator for the month was 0.1%, that means real income increased 0.3% and real spending was unchanged. Since the end of stimulus spending in May 2021, real spending is up 4.1%, while real income has declined -1.7%:

The personal saving rate increased 0.2% to 2.4%, which is just above its all time lows, as shown in the below graph which subtracts -2.4% so that the current reading shows as 0:

Real personal income less transfer receipts is one of the 4 monthly data series heavily relied upon by the NBER in dating recessions. This increased in November and is less than -0.1% below its all time high of exactly one year ago. The big decline in gas prices since June is a major driver of the recent improvement:

Which means that the YoY reading is just below 0. Why is this significant? Because in the past this metric has only declined to 0 or negative during - frequently late in - recessions:

This lag in the performance of real income and spending is why I pay more attention to real retail sales. Here is the 50+ year look at the YoY% changes in real personal spending (blue) vs. real retail sales (red):

Note that real retail sales have *always* turned negative YoY before recessions start, whereas real personal spending either turns late, and sometimes does not turn negative at all.

Here is what that looks like for the past 12 months:

Real retail sales have been flat to slightly negative ever since this past March with the exception of July and August, while real spending is still higher by 2.0% - although in the past such a low positive level has also been consistent with the onset of a recession.

At the moment, the labor market is the only segment of the economy that does not appear to be actively rolling over into recession.

Durable goods orders appear to have peaked

 [Note: I’ll post about personal income and spending, as well as new home sales, later.]

 - by New Deal democrat

I normally don’t pay much attention to the monthly durable goods report, but this morning’s report for November appears significant.

That’s because durable goods spending has been one of the few short leading indicators to have continued to improve - until now. Here’s the long term view:

New factory orders for durable goods declined -2.1% in November, while “core” durable goods orders excluding aircraft and defense increased 0.2%. Here’s what the last 12 months look like:

Durable goods orders have been essentially flat since June, and are now below that level. “Core” orders last made a high in August. They appear to be in the process of rolling over.

That leaves consumer durable goods spending and initial jobless claims as the only remaining positive short leading indicators.

Thursday, December 22, 2022

Initial claims continue in range; why they will give us a lead on when the Sahm rule for recessions may be triggered


 - by New Deal democrat

Initial claims ticked up 2,000 last week to 216,000. The 4 week moving average declined 6,250 to 221,750. Continued claims, with a one week delay, declined 6,000 to 1.670 million:

To state the obvious continued good news, it remains the case that almost nobody is getting laid off. 

Also continued good news is that claims, and in particular the 4 week moving average, remain lower than their level one year ago:

So long as this remains the case, we can be confident that the economy remains in expansion. I’ll hoist a yellow cautionary flag if and when claims turn higher YoY, and a recessionary red flag if they turn higher by 10% YoY.

I’ve seen some commentary that no recession can start so long as initial claims remain very low.

Historically this is not true. There has been no “magic level” of initial claims correlating with increased or decreased employment or unemployment levels. Sometimes it has taken 400,000 or more (1980, 1981), sometimes as low as 250,000 or less (1970, 1974). In 2001, it took about 370,000; in 2007, it took 340,000. The key has been a sufficient increase from the expansionary lows.

Confirmation of the above can be found indirectly via the Sahm Rule, which holds that we can be confident that a recession has started if the 3 month average of the unemployment rate has risen 0.5% from its previous 12 month lows. (Note in some cases the actual start of recessions has not required this much of an increase. Rather, the rule is one of sufficiency rather than necessity).

With that rule in mind, it has also been the case for 60 years that initial claims lead the unemployment rate. Here’s the graph that plainly shows the leading/lagging relationship from 1966 through 2019:

And here is the continuation of the graph for the past 2 years:

So the fact that initial claims made their low last March and remain slightly higher continues to indicate that the unemployment rate would make a subsequent low (it did, in July and September), and has also risen slightly since.

If and when the 4 week average is 10% above its previous low YoY, we can be confident that the unemployment rate will similarly follow higher (keeping in mind that a 10% increase from 3.5% unemployment is 3.85%). So initial claims will give us a good heads up as to when the Sahm rule might be triggered in the near future.

Wednesday, December 21, 2022

November existing home sales: prices have unequivocally turned down


 - by New Deal democrat

Existing home sales do not have much actual economic impact, since the primary economic activity generated by housing is the construction. But they do help tell us a great deal about pricing.

For the record, sales continued their relentless decline this year, down to 4.09 million on an annualized basis, down almost 1/3rd from their recent February peak of 6.02 million:

This is in line with the 35% declines we saw yesterday in single family housing permits and the 30% decline in total permits. Only housing starts, off 20% from their peak, are “less bad.”

The longer term view (note: graph only goes through September) shows that November sales were the lowest since November 2010, with the exception of May 2020 and June 2012:

But the real importance of existing home sales is in their price signal, and here that signal was unmistakable. At $370,700 for the median existing home, prices are only up 3.5% YoY (the NAR does not seasonally adjust, so this is how we have to measure):

Here is a 5 year graph through September from Mortgage News Daily, showing that peak YoY appreciation within the past 12 months was 17.6% last January:

My rule of thumb is that data which can only be measured YoY has peaked when the YoY increase is less than 1/2 of its highest rate in the past 12 months. So any YoY increase of less than 8.8% would indicate a peak. Needless to say, 3.5% is well below that.

My mantra for the housing market is that sales lead prices. This year, sales turned in the Jan-Mar time frame for all of the various measures like permits, starts, and new home sales, as well as existing home sales. We now know to a virtual certainty that prices peaked at some point during the summer. 

The sales data is, as I said yesterday, recessionary. But the old saw is that “the remedy for high prices, is high prices.” Now that we are seeing prices come down, the groundwork is being laid for the economic turnaround to come - the timing of which is yet to be determined.

Tuesday, December 20, 2022

November housing permits and starts: the biggest news is not in the headlines


 - by New Deal democrat

The report on housing construction for November was very much a tale of two very different trends - and the most important one will almost certainly be under-reported.

Housing permits issued declined to 1.342 million annualized, the lowest number since June 2020, and before the pandemic the lowest since July 2019. The even more reliable single family permits declined to 781,000 annualized, the lowest since May 2020, and before that November 2016! Finally, the more volatile housing starts declined to 1.427 million annualized, the lowest since August 2020. Here’s the graph showing all three:

Those are all big declines, and definitely recessionary. But they’re not the biggest story.

For one thing, the backlog of housing units authorized but not yet started declined only slightly to 293,000, only slightly below its March peak:

This most likely is much affected by cancellations, which have soared in recent months.

But the big story, in my opinion, has to do with the metric that measures the actual economic activity in the housing sector; namely, housing units under construction. This remained at 1.709 million, tied with last month and at its peak. In the below graph, I also show the number of residential construction workers from the recent jobs report:

In other words, in terms of actual economic activity, housing isn’t yet contributing to a decline. Since units under construction and the number of construction workers typically move in tandem, note that employment has not declined yet either.

There was some slight movement, in that single family units under construction declined to 777,000, an 11 month low, while multi-unit construction increased to a new high:

This reflects buyers being priced out of the single family market.

Finally, note that mortgage rates have come down significantly in the past eight weeks:

If this persists, we may put in a bottom in housing permits in the next few months, which of course would be good news for 2024.

But, to emphasize again, despite the big declines in the headline numbers, actual economic activity in housing construction remains at its peak.

Monday, December 19, 2022

Job growth beginning in Q2 looks to be substantially revised downward


 - by New Deal democrat

Last week the Philadelphia Fed published a working paper suggesting that in the second quarter of this year only 10,500 jobs were actually added, rather than the 1,047,000 as indicated by the monthly Establishment survey. 

Here’s their graph:

Here’s what you need to know about the QCEW (Quarterly Census of Employment and Wages): 

The late Jeff Miller, a portfolio manager who was extremely popular at Seeking Alpha, and was previously a college professor who taught public policy courses and quantitative methodology at the University of Wisconsin-Madison and Lawrence University, had this to say:

 Each quarter the BLS reports data from state employment agencies. Since no one pays insurance premiums on phantom employees, we can expect conservative information. The only problem is that it takes about nine months to get these actual counts.

“Any honest observer of the market would circle the date of this release….” 

The QCEW is generated by more than 95% of all employers, essentially all who pay unemployment insurance; versus the monthly Establishment survey, which is generated by a sample of 650,000 employers.

In order to understand how the Philadelphia Fed arrived at its result, I have gone back and crunched the numbers for Q2 job growth (or losses) in the QCEW since the beginning of the database in 2001. In the below chart, the first line is the year, the 2nd is the non-seasonally adjusted number of jobs added in Q2 in the QCEW, in millions, and the 3rd is the non-seasonally adjusted Establishment survey number:

2001 2.0* 1.62
2002 2.5 1.79
2003 2.4 1.74
2004 2.9 2.76
2005 3.0 2.81
2006 2.9 2.58
2007 2.7 2.43
2008 1.9* 1.55
2009 0.7* 0.14
2010 3.1 2.61 
2011 2.7 2.71
2012 2.7 2.46
2013 2.8 2.61
2014 3.2 2.83
2015 3.2 2.75
2016 2.7 2.62
2017 2.9 2.58
2018 2.9 2.70
2019 2.6 2.44
2020 -12.0* -16.3
2021 3.6 3.25**
2022 2.1 2.91

*=lower than Q2 2022 QCEW
**=higher than Q2 2022 CES

I then went back and compared with the seasonally adjusted Q2 numbers in the CES. Every year that the NSA QCEW numbers for Q2 were below 2.0 showed actual job losses in the seasonally adjusted CES. The three years with the next higher numbers compared with Q2 2022 - 2007, 2012, and 2019 - generated seasonally adjusted CES job gains of 90,000/month, 85,000, and 165,000.

So at this point it very much looks like the Philadelphia Fed’s paper has a lot of merit: the NSA QCEW data shows a very serious slowdown in job growth in Q2. And since the QCEW isn’t a survey sample, but rather collects about 95% of the entire data, it needs to be taken seriously (with the caveat that it is preliminary data).

It’s also in accord with the Household survey, which shows only 12,000 jobs added since March:

and payroll tax collections by the US Treasury, which show a sharp slowdown that is first apparent in June’s +0.5% YoY growth. Here’s my recapitulation of YoY payroll tax payments to the US Treasury for each month of this year through November:

  • January 2022/21: up +21.3%

  • February 2022/21: up+11.6%

  • March 2022/21: up +6.2%
  • April 2022/21: up +11.5%
  • May 2022/21: up+17.2%
  • June 2022/21: up +0.5%
  • July 2022/21: up +5.4%
  • August 2022/21: up +10.2%
  • September 2022/21: up +1.7%
  • October 2022/21: up +12.2%
  • November 2022/21: down -2.5%