Weekly Indicators for June 16 - 20 at Seeking Alpha

 

 - by New Deal democrat

My “Weekly Indicators” post is up at Seeking Alpha.

The trends that have been in place ever since the start of Tariff-palooza!, aided and abetted by the GOP budget-busting bill and the burgeoning Israel-Iran war, are continuing.

The US$ is declining, oil prices are spiking, rail traffic has turned negative, and consumer spending seems to be slowly decelerating.

WINNING! One of the things I may elaborate on as early as tomorrow is that not a single US economic expansion in the past 50+ years has died a natural death. Every single one has been murdered in whole or in part by a domestic political or geopolitical shock. As noted above, in the past 5 months T—-p has managed to unleash at least 3 of them.

As usual, clicking over and reading will bring you up to the virtual moment as to the state of the economy, and reward me with a penny or two for organizing and reporting on it.

Why hasn’t the housing downturn caused a recession yet? A detailed look

 

 - by New Deal democrat

As promised yesterday, today let me take an extended look at the important leading sector of housing. I want to walk through each of the important series generally in the order in which they have typically peaked, and how in at least one important respect this time is - somewhat - different.

First, here is another variation on a graph I have run many times, showing how mortgage rates (red) typically lead both total (dark blue) and single family (gold) permits and the more noisy starts (light blue). Here is 1972-98:

And here is 1999-present:

The magnitude of the response is not always proportionate (see the 1970s) and in at least one case (the housing bubble and bust of the 2000s) overwhelmed by other factors, but the relationship almost always holds.

Here is a close-up on the post-pandemic period:

As usual, not perfect, but the major leading/lagging relationship holds. Most importantly for today’s purposes, note that mortgage rates have not changed significantly in the past 20 months, and since then both permits and starts have become more rangebound YoY as well.

Next, let me show the relationship between new home sales (red) and permits, first from 1962 to 1994:

And 1995-present:

The point here is that sales lead even permits - but they are much more noisy (and also heavily revised), which is why I pay more attention to permits.

Here is the post-pandemic close-up, showing that as usual sales both peaked and troughed first, before permits, and have been similarly rangebound since mid-2023:

For completeness’ sake, here are purchase mortgage applications as well:

Unfortunately FRED is not allowed to publish these, but note that they peaked right after new homes sold and right before permits. They also troughed in late 2023, and were rangebound until the end of 2024, before rising in the early part of this year. That would normally be a good sign, but applications have declined again in the wake of the recent increase in mortgage rates due to Tariff-palooza!

Next, let’s delete the more noisy starts, and compare total and single family permits with housing units under construction (gold, right scale), all normed to their 2022 peaks:

The important points here are that units under construction - which as I almost always say, is the real total of the economic activity in this sector - follow permits (and starts) with a lag. And secondly that when there have been recessions, the number of units under construction continued to decline right into the period of actual economic contraction.

Once again, let’s focus on the post-pandemic period:

As usual, the number of units under construction followed permits by a little under a year. Note also that as per the above discussion of mortgage rates, the absolute number of permits has been rangebound for almost two years - until possibly yesterday’s report, which may have been the first one reflecting the impact of tariffs (on lumber, for example). 

But more importantly, the number of units under construction has declined almost every month in the past 18, and is well within the range of declines where previously recessions have begun. Indeed, if units under construction stabilized for a significant period of time, that would suggest a recession has been avoided - but that has not happened.

Before I get to the final series in chronological order, let me note that construction spending, whether nominal or adjusted for inflation, peaked almost simultaneously with permits in 2006:

Post-pandemic, as measured by construction costs, construction spending peaked at the time of the initial peak in permits in early 2001. Nominally or deflected by the CPI, it peaked several months before permits (teal) did in 2022:

Note that in the past several months residential construction spending has also declined further, just as permits have done.

Now let’s take a look at what the final shoes to drop typically have been before recessions have started, houses for sale (blue) and residential construction employment (red), in comparison with units under construction:

Although I won’t show a graph this time, remember that homes for sale lag homes sold in the housing sales report, and it has only been once homes for sale peak and then turn down that recessions have begun. There is no clear pattern as to whether homes for sale or residential construction employment peak first, but both lag units under construction.

And now, our final post-pandemic look:

The number of houses for sale declined slightly in April, but it is too soon to know whether that marks the peak or is just noise. But in the past 6 months, the number homes for sale have only increased by 1.2%. Residential construction employment is still increasing! 

Let me make one final point about these last two graphs. At the outset of this post I wrote that at least one thing was somewhat different about this cycle, which has a lot to do with why the number of units under construction seemed to levitate for so many months after permits and starts peaked, and why employment in this sector has not turned down yet. 

Notice, beginning in about 2012, how much construction employment lagged the increase in units under construction. Between then and the outset of the pandemic, units under construction nearly tripled, while the number of workers building those units only increased 50%. That has only been exacerbated since the pandemic, as units under construction peaked at more than *quadruple* their number in 2011, while employment in the sector is only up 70%. 

With such a big decline in the number of laborers per unit being built, it is no wonder that the lag time has been so great, and the need to lay off workers so little. But if there is to be a recession, almost certainly both of the final shoes in this chronological panorama will drop first. And Tariff-palooza! might just do the trick.

[P.S.: Since there is no significant data tomorrow, and given the length of this post, don’t be surprised if I play hooky.]

Housing construction looks even more recessionary

 

 - by New Deal democrat

Because no data will be released tomorrow due to the Federal holiday, I am going to defer a deeper look at the very important housing sector until then. Today I’ll just note the top line indications from this morning’s housing construction data.

As per usual, permits are much less noisy than starts, and slightly more leading. Single family permits are the least noisy of all the data, so I pay the most attention to that number. And the numbers for May were not good.

Permits (gold) declined -29,000 annualized to 1.393 million, the lowest number since June 2020. Single family permits (red) also declined, by 25,000 annualized, to 898,000, the lowest in 2 years. Starts (light blue) declined -136,000 annualized to 1.256 million, the lowest since May 2020:

Recall that mortgage rates generally lead housing permits and starts, so to some extent this is a function of those rates (blue, left scale in the graph below), which for the last seven months have hovered near the higher end of their range since late 2022:

Some of this may also be a function of Tariff-palooza!, which has caused the price of construction materials to jump (more on that tomorrow). But it is probably also part of a feedback loop, because the number that represents where the proverbial rubber meets the road for actual economic activity - housing units under construction - declined another -29,000 annualized to 1.375 million, the lowest number since June 2021, and -19.8% below their October 2022 high:

More often than not in the past, by the time 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%.

I will explore more why this time around no recession has begun yet with my more in-depth look at housing tomorrow.

Jobless claims show a weakening, but not (yet) recessionary economy

 

 - by New Deal democrat

Because tomorrow is the Federal Juneteenth Holiday, initial and continuing claims were released today. They continue to show a slowly weakening, but not (yet) recessionary, economy.

Initial claims declined -5,000 to 245,000 for the week, while the four week moving average rose another 4,750 to 245,500. This four week average is the highest since August of 2023. Meanwhile, with the usual one week lag, continuing claims declined -6,000 to 1.945 million, after last week the highest since November 2021 (not shown):

While this continues the weakening trend we have seen for the past few weeks, on the YoY% basis more useful for recession forecasting, it was “more of the same.” Initial claims were up 3.4%, the four week average up 5.8%, and continuing claims up 6.2%, right in the middle of the +5% +/-5% range we have seen for the past 9 months:

Finally, let’s take our first look at what this likely means for the unemployment rate in the next several months:

In this case I think it is obvious that the increasing number of initial and continuing claims suggests at least preliminarily that the unemployment rate is likely to move higher in the next few months.

In short, a jobs economy that is slowly weakening, with a slight increase in people being laid off, and finding it increasingly difficult to find new employment, but not enough of a weakeneing at this point to mean recession in the next few months.

Industrial and manufacturing production mixed, but also consistent with the end of front-running

 

 - by New Deal democrat

Today’s second report was for industrial production, including its important manufacturing component. 

The data for May was mixed, as total production (blue) declined -0.2%, while manufacturing production (red) increased 0.1%

In March I wrote that “I suspect the big increases in February and March in manufacturing, like this morning’s retail sales numbers, were about front-running T—-p’s tariffs. Which means that like retail sales, production might have been pulled forward from the next few months, which may lead to whipsaw declines.”

Revisions to the data released this morning and going all the way back to last year showed a more complex picture. The net effect of the revisions was a decline of -0.1% for each measure through April. But the main point stands, as total production peaked in February, and manufacturing production peaked in March:

On a YoY basis, both total and manufacturing production are up 0.6%, a sharp deceleration from the previous four months:

Both of this morning’s reports were important. Industrial production has historically been one of the two main coincident indicators for recessions, while real retail sales are both an important short leading indicator, and to the extent they presage real personal consumption, which is another important coincident behavior. Both of them declined, probably due to the end of front-running tariffs by both consumers and producers. At the same time, both are within the range of noise, and for both to actually signal a downturn I would need confirmation from other short leading indicators, as I discussed yesterday.

May real retail sales: the front-running of tariffs is over

 

 - by New Deal democrat

We can safely say that the front-running of tariffs is over. As usual, real retail sales is one of my favorite indicators, because it tells us so much about consumers, and since consumption leads employment, it gives us information about the trend in that as well. And this morning’s report for May was our earliest indication in the monthly data of the post-Tariff-palooza! trend.

Nominally retail sales declined -0.9% in May, and were up 3.3% YoY. But because consumer prices rose 0.1%, real retail sales declined -1.0% for the month (blue in the graphs below). In recent months I have also been calculated real sales excluding shelter, because that has been distorting the CPI. This month real retail sales ex-shelter were down -1.1% (gold). In the below graph I also show real personal consumption expenditures for goods (red), which tends to track real retail sales well, but won’t be reported for several more weeks (Note: graphs normed to 100 as of just before the pandemic):

With rare exceptions - one of which was in 2023-24 - when real retail sales are negative YoY, a recession has followed shortly. Even after this month’s downturn, in the past 12 months, real retail sales YoY are up 0.9%, and excluding shelter, up 1.8%:

Probably the best way to look at this is to average last month and this month, which would mean that the YoY trend has been about equal with that of the end of last year, when real sales were in the range of 2.0%-2.5% higher YoY. If next month’s report does not show a further decline, the YoY measure will still be positive. But if it is further decline like this month, likely the measure will be negative again.

Finally, let’s compare the YoY% changes with their potential effects on employment (red):

To reiterate, consumption leads employment. Changes in the strength of sales show up with some number of months’ delay in changes in the strength of employment. Because real sales are still positive, and the average of the last few months has not deteriorated, that suggests that the jobs report should stay positive for the next few months, with YoY gains on the order of 1%, with two caveats: (1) the YoY comparisons in employment in the next few months will be with gains of less than 100,000 last summer, meaning that similar numbers might be expected this summer as well; and (2) the QCEW, which is the “gold standard” for employment, has been indicating gains of only about 0.8%-0.9% YoY for the latter parts of last year, meaning that for the YoY gains to remain steady, even weaker job reports in the next few months might be expected.

But to reiterate: the main takeaway from this month’s retail sales report is that the consumer front-running of tariffs has ended, and payback (of unknown duration and strength) has begun.

Updating the nonfinancial long leading indicators, plus several important short leading ones

 

 - by New Deal democrat

Since a couple of years ago, I temporarily suspended my updating of the long leading indicators. That is because their negative slant in 2022 was completely overcome by the hurricane force tailwind of the unspooling supply chain kinks. By that time the Fed had already raised rates more steeply than at any point since 1981. So looking forward, should the financial indicators in particular still be normed to their best post-pandemic readings, or re-normed to after the supply chain unkinked? There was no way to know.

But at some point they must resume their salience. At this point I think it is fair to restart examining the non-financial long leading indicators; that is, those not directly under the control of the Fed or at least partially so, like interest rates, the yield curve, and real money supply.

The non-financial long leading indicators are housing permits, corporate profits, and real sales per capita. So let me take a K.I.S.S. look at each of those, plus several other salient short leading indicators; namely, real spending on goods, real aggregate nonsupervisory payrolls, and initial jobless claims. All of the graphs below are rendered YoY for ease of comparison.

First, here are corporate profits after tax, both before (light gray) and after (dark blue) accounting for inventories, divided into three time periods, again for ease of viewing. Here is 1948-1966:

Note that the measure adjusting for inventories appears to be a slightly better indicator.

Next, here is 1967-1996, and 1997-2019, also including the quarterly average of initial jobless claims, which began to be reported in 1966 (inverted):

Note again that adjusting for inventories gives us a better indicator. And when we include initial jobless claims as well, i.e., both indicators must be negative, the record is perfect except for one quarter in 1998.

Now here is the post-pandemic record:

Again, at no time since the pandemic have both profits after adjusting for inventories and initial jobless claims been negative simultaneously. 

Here is the update, as of last week, of corporate profits as reported to investors through Q1, together with Wall Street estimates going forward:

Note that profits are anticipated to decline for the second quarter in a row this quarter, but not to be negative YoY.

Next, let’s look at housing permits (blue) together with housing units under construction (red), which is the more “real” measure of ongoing economic activity in this sector:

With the exception of Q4 2023, permits have been negative since the summer of 2022. Units under construction went negative in late 2023, and turned ever more negative through the end of 2024. They are still down more YoY than at any time entering a recession, except for 2007.

Next let’s look at real retail sales (blue) for the 60 years up until the pandemic. The below graph is not per capita in order to K.I.S.S., but population growth has tended to average about 0.75% to 1% a year, so it is easy to adjust. I’ve additionally included real personal consumption of goods (red), which has a similar trend:

Except for a few isolated months, when real retail sales went negative YoY, a recession was imminent or had just started. Real spending on goods avoided those false positives, but several times did not turn negative YoY until late in recessions or not at all.

Now here is the post-pandemic look:

Real retail sales again gave a false positive for an extended period of time, but has turned up over the past year. Real spending on goods avoided most of that, and has similarly trended higher beginning in 2023. Much of the positivity in the last few months has probably involved front-running tariffs, so it will be interesting to see what happens in the May reports, which for retail sales will be issued tomorrow.

Finally, here is an update on real aggregate nonsupervisory payrolls, based on the CPI reported last week:

Again, this almost always has turned negative YoY, or at least decelerated sharply, just before a recession has begun. By contrast, this measure has been trending higher since the beginning of 2024. It currently sits at higher by 2.8% YoY. I would expect that to decline below 1% by the time a recession begins.

Let’s sum up the nonfinancial long leading indicators. The housing sector is giving recessionary readings. Corporate profits adjusted for inventories are weakening, but are still positive YoY. But real sales are not just positive, but they have been improving.

Additionally, neither real aggregate payrolls nor initial jobless claims are signaling a downturn at this point. In fact, while the latter is weak, the former measure is strongly positive through last month.

I would expect all of the above measures to falter, or falter further, before a recession were to begin, even with the complete chaos coming out of Washington. That is, I would expect corporate profits to decline further, real sales to turn down, and real aggregate payrolls to decelerate sharply by then.