The neutral trend in the most reliable long leading indicators in housing continues

 

 - by New Deal democrat

Let me start out this post with my updated overall conception of the long leading indicators. To begin with, most of them are financial — generally speaking, “the cost of money.” That applies to bond interest rates, the yield curve, real money supply, and bank lending. Recently, I’ve also indicated that fiscal policy should be added, as in, is there a major stimulus or austerity at work? But it is increasingly apparent that a recession does not occur until after the “real world,” non-financial long leading indicators also participate: corporate profits, real retail sales per capita, and - drum roll, please - housing permits and starts. 

Because although permits and starts are very much downstream of mortgage interest rates, they represent activity in the sector that constitutes the biggest consumer purchase of all, and through construction, landscaping, and furnishing, typically take around 2 years to fully filter through into the broader economy.

So let’s look at the numbers. To begin with, I no longer measure against the big 2022 interest rates hikes, because those have fully worked through the system. Thus almost all the graphs below are limited to the last three years. In the last several months I have written at length about how the entire housing market had reached an equilibrium, where almost all of the metrics were more or less flat.

This morning’s data on housing permits, starts, and units under construction for July continued that trend. Housing permits (blue) issued rose 69,000 to 1.443 million annualized, while the much more volatile number of starts (gold) declined -176,000 to 1.239 million annualized. Single family permits (red, right scale), which are the least volatile metric conveying the most signal, rose 22,000 to 894,000 annualized. The below graph shows each of them in comparison with their high water marks of the last several years, January and February of 2024:

Since that time, permits are off -8.1%, single family permits off -13.7%, and the three month average of starts off -17.8%. But perhaps more importantly, the stabilization of permits and starts slightly lower than their 2023-24 levels in the past 12 months is apparent.

In that vein, for me to consider housing recessionary, I would expect to see all three off at least -10% from recent highs, but also as the below historical graph shows, down -10% or more YoY:

Note that single family permits has turned positive YoY no later than 5 months after the end of any recession in the past 50+ years. 

But as the below graph of the last three years shows, not only are these metrics no longer down more than -10% YoY, but two of the three are positive, as permits are up +3.1% YoY, single family permits up to+1/1%, and only starts down -13.8% (-6.4% for the three month moving average):

This is no longer recessionary.

Last year I was calling housing units under construction the “last shoe to drop;” and after a plateau in 2022-23, it had been dropping like a rock, as shown in the graph below. But then it too started to stabilize, as this series has been virtually unchanged for the past seven months:

And the below historical graph is what really grabbed my attention, because in the past it has only flattened out, and started to improve on a YoY basis, only at the end of recessions and beginnings of expansions. And that is exactly what the YoY comparison shows now:

Units under construction are still down -6.0% YoY, but this is the best YoY comparison in 24 months. This is most consistent with a rebound after a recession.

To sum up, residential construction metrics in July continued the sideways trend, of a subpar housing equilibrium — but an equilibrium that nevertheless means this long leading indicators for the economy is no longer recessionary, but neutral.

Of inflation and corporate bond spreads

 

 - by New Deal democrat

Over the weekend, in response to my “Weekly Indicators” post, a commenter wanted to know why I was concerned about inflation. After all, hadn’t core CPI just tied its post pandemic low?

And it is true: core inflation (blue), at 2.5% YoY, is equal to its post-pandemic low. Further, CPI less energy (red), a measure I used to cite monthly before COVID stimulus-induced house price inflation dwarfed everything else, did make a new post-pandemic low YoY through July:

But in reply, I noted that not only was headline inflation still elevated, but pipeline price pressures were manifest in both the ISM monthly reports and the various Fed regional reports, one of which - the New York Fed’s Empire State manufacturing survey - was updated for this month this morning, and it showed continued widespread increases in both prices paid (blue) and received (orange). In the graph below, I also show the latest Philadelphia Fed readings (light blue and light orange) through last month:

Interestingly, although prices paid if anything have become more widespread in the last few months, relatively speaking producers are having a difficult time passing them all on. This is going to squeeze profit margins, and that tends to lead to cutbacks.

And gas prices through the first half of August have increased compared to July:

And of course the Strait of Hormuz is still closed, Wall Street futures traders notwithstanding. 

So yes I continue to believe that inflationary pressures have been building.

Which brings me to a second point that I made in my piece over the weekend: namely, that the increase in yields on bonds at longer maturities is a negative sign. That point was reinforced with another piece of information that was posted by Apollo Research via Carl Quintanilla, pointing out that yields on speculative corporate credits have - at least somewhat - blown out.

To wit: below is a graph of the last three years of bond yields for CCC grade speculative debt (blue), BBB debt (gold), BAA debt (red), and 30 year Treasurys (black):

Note that I’ve subtracted -5% to CCC yields better to show the comparison in trends. There’s no doubt that there has been some spreading.

Unfortunately, B of A only allows FRED to post the last three years of data. So here is a graph of CCC vs. BB corporate debt going back 30 years:

Whenever bond traders get worried, the speculative high yield CCC debt blows out first and worst. In comparison, the current relative increase in CCC debt is not that significant - at least not yet. Also, when the economy weakens, banks become more wary about extending loans for speculative debt, as shown in the relevant metric of the Senior Loan Officers Survey, which was just updated two weeks ago:

Through Q2 of this year, there was no such wariness yet.

As I’ve written a number of times recently, the current situation is best described as an inflationary expansion. Before it ends, among other things I would expect banks to tighten credit, high yield CCC debt to blow out considerably more against more creditworthy B-grade corporate debt, and I would expect even better rated corporate debt to increase in yield as well.

 

Admiral of the Fleet T—-p, meet Admiral of the Fleet Wilhelm II

 

 - by New Deal democrat

You probably recall that a number of months ago T—-p decided that, Pearl Harbor notwithstanding, it was time to bring back battleships in a new class not coincidentally to be named after himself.

Now this weekend we have been treated to the news that T—-p had decided he is an expert in the design of aircraft carriers. In case you’ve missed it, on Friday we learned that he has ordered the navy to ditch its modern, magnet-based aircraft launching system and return to old-fashioned steam catapults.

 

 Then, this morning, we found out that he also wants new aircraft carriers to look like their WW2 ancestors. Notwithstanding all the, you know, technological and efficiency improvements in the past 75 years.

Which brought to mind a post I wrote here eight years ago, in which I went in to great detail about how T—-p’s personality is almost exactly the same of another infamous narcissistic idiot of a ruler, Kaiser Wilhelm II of Germany 100+ years ago, calling T—-p the “doppelgänger” of the latter.

By the way, I am not the only one who arrived at the same conclusion during T—-p’s first term. Here, for example, is David E. Banks in The Independent, making a similar piont:

“Like Trump, the Kaiser was an insecure and aggressive narcissist who allowed his mood to dictate many of his policy decisions, and while the domestic effects of these traits could be limited by the quasi-democratic institutions of the German Reich, in the foreign policy arena his personality wreaked havoc.”

What brought the comparison back to mind this weekend is the following passage from the book "George, Nicholas, and Wilhelm," by Miranda Carter, that I highlighted in that post back eight years ago:

"Wilhelm considered himself an expert on many things and was not shy about saying so…..

"[In 1889 in an attempt to smooth over some family difficulties, Queen Victoria had awarded Wilhelm an honorary admiralty of the Royal Navy. Afterward,] Wilhelm fell upon his new title as if nothing had ever given him so much pleasure in his whole life..... Even Phillip zu Eulenburg noted disappointedly that he was "like a child over it [the British naval uniform]." Wilhelm told Herbert von Bismarck that his British naval title meant that "he would have the right as admiral of the Fleet, to have a say in English naval affairs and to give the Queen his expert advice... [He] was perfectly serious in what he said."

"...[Later that year,] Wilhelm put on his admiral's uniform, flew the pennant of a British navy admiral, and invited himself -- as a real admiral would -- to inspect the British squadron anchored [off the Greek coast].... In December, he sent [Victoria] a plan for the reorganization of the Royal Navy.... In 1891 he sent more "humble suggestions...." 

He really is Wilhelm’s doppelgänger. Let’s hope the war he has blundered us into (so far) does not turn out to be the generational catastrophe that World War I was. 

Weekly Indicator for August 10 - 14 at Seeking Alpha

 

 - by New Deal democrat

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

It is surprising how large a majority of the high frequency data is positive, despite all of the chaos that has been thrown at the economy. Still, the bond market in particular has not been fooled by the consequences of the Big Bad Bust-out Budget Bill, as 10 and 30 year yields are at or close to 20 year highs. As a result, the US will have to devote more and more of its budget to interest payments on its debt. It is - or at least may be - the beginning of the dreaded “hockey stick.”

In any event, clicking over and reading will bring you up to the virtual moment as to the state of the economy, and reward me with a little bit of pocket change.

July retail sales lay en egg; a hangover after a spring Budget Bust-out Bill binge?

 

 - by New Deal democrat

Let me start out this post with two comments: (1) real retail sales is one of my favorite economic indicators, because it tells us so much about consumer spending, which is about 70% of the economy, and also because, with a lot of noise, consumption leads employment; and (2) about once a year, it lays an egg. That’s basically what happened in July.

To the numbers: nominally, total retail sales declined a sharp -0.6% in July. Since consumer prices barely rose in July, the real inflation adjusted number rounded to -0.6% as well. Here’s the post-pandemic look at the absolute numbers:

Since gas prices have been a major driver of inflation - and deflation - in the past few months, here’s a look at the monthly % changes in nominal retail sales excluding gas stations (orange) vs. total retail sales (blue). And like all the other metrics this month, real retail sales excluding-gas also declined -0.6%:

So gasoline sales weren’t the main culprit. Rather, the weakness was widespread. Like I said, about once a year retail sales lay an egg, so maybe it was July’s turn this year.

On a YoY basis, nominally retail sales were up 5.0%, but since consumer inflation is up 3.3%, real retail sales rounded to 1.7% higher YoY. For comparison purposes, I also show the YoY% change in real personal consumption (orange) which won’t get reported until the end of this month: 

This is in accord with the weekly YoY% change in retail spending as measured by Redbook, which also backed off substantially YoY in the past five weeks:

There has been some speculation that the surge in consumer spending in the last few months was driven in part by larger tax refunds to upper income recipients due to the last year’s Budget Bust-out Bill. If so, that such added spending might be tailing off would also be a likely explanation for the July downturn. But per my opening comment, unless there is further erosion next month, I am treating this as a one-off downdraft.

Finally, since consumption leads employment, here is the update of YoY real retail sales and real personal consumption (/2 for scale) together with employment (red):

Two months ago, I wrote that “this suggests that on a YoY basis the rebound we have seen in the last three jobs reports is likely to continue in the next several months.”  I continued that position after the relatively poor June jobs report. 

After July’s even worse jobs report, showing an actual decline, maybe not so much. But as I wrote above, the leading/lagging relationship is a somewhat noisy one; but the fact remains that with the increase in consumer spending this year, employment should still follow suit. 

Producer prices indicate continued inflationary expansion, but how long can it last?

 

 - by New Deal democrat

I usually do not pay much attention to producer vs. consumer prices. Partly that is because in the past few decades, the PPI has tended to be coincident with the CPI rather than leading it, and partly because so long as the YoY PPI is less than CPI, producers do not feel under pressure to make cost (i.e., payroll) cuts. Unfortunately, that means that since the start of the Iran war I’ve had to pay more attention to the PPI release.

Let’s start with the raw numbers. Producer prices for final demand (red) decreased -0.1% in July, while raw commodity prices (including oil)(gold) declined -0.8%. This compares with consumer inflation (blue), which increased 0.1%:

 On a YoY basis, PPI final demand was up 4.7%, and 8.3% for raw commodities, vs. 3.4% for consumer prices:

Significantly, the PPI increase for final demand services, which had been 4.6% or higher YoY for the past few months, decelerated to 3.9% in July - which unfortunately is still higher than any such reading aside from the immediately inflationary post pandemic period, and briefly in the summer of 2024:

Of more concern, as per my lede above, is that the YoY measure of final demand producer prices remained higher than the that for consumer prices. And although it hasn’t been a uniform rule, as this historical graph shows, when the YoY% increase in PPI exceeds CPI, more often than not that spells trouble:

This indicates that the underlying inflationary pulse has continued to go well beyond energy related prices. And, just like last month, it also confirms what we have seen for a number of months now in the regional Fed indexes: widespread price increases in inputs, which are only incompletely being based on to buyers downstream. To summarize, the regional Fed indexes for the past few months have indicated rampant input price hikes, with much - but not nearly complete - pass-throughs to consumers. 

So my conclusion this month is the same as it was last month: “If producers stay squeezed, they are going to begin to make cost cuts where they can.” Which may include a freeze in new hiring, a cut in hours, or possibly even worse. In other words, this inflationary expansion is likely to either stop being inflationary, or stop being an expansion, sometime in the not too distant future.

Jobless claims continue to forecast a very positive economy in the near term

 

 - by New Deal democrat

Let’s take our regular weekly look at one of the most positive recent signs for the economy, initial and continuing jobless claims.

And they continued to be very positive. Initial claims rose 9,000 for the week to a still very tame 209,000, while the four week moving average was unchanged at 199,000. As a reminder, aside from several weeks in 2022, this average has not been below 200,000 for over half a century, when the US population was only about 1/2 of what it is now. Continuing claims, with the typical one week delay, declined -22,000 to 1.777 million:

As per usual, for forecasting purposes what we want to look at is the YoY comparison, and there initial claims were lower by 6.7%, the four week average by -10.3%, and continuing claims by -8.5%:

As I said above, this continues to be a very positive short leading indicator for the economy.

Finally, since it’s early in the month I won’t update the implications for the unemployment rate going forward this week. Instead, here is an update of the “quick and dirty” forecast model that includes the inverse of the YoY change in the four week average, plus the YoY change in stock prices:

Combined, these are the most positive they have been since the immediate post-pandemic Boom.

July consumer inflation: the second gift horse in a row, with gas prices down again and shelter subdued

 

 - by New Deal democrat

As I wrote yesterday, July’s CPI was likely to be subdued because on average the price of gas went down further in July. And it was, rising only 0.1% for the month and 3.4% YoY (blue). Perhaps more important, core CPI excluding food and energy (red) rose 0.2%, and was only 2.5% higher YoY, tied for its lowest advance since the pandemic was raging five years ago. And shelter, which is 1/3rd of the entire index, continued its deceleration, up only 0.1% for the month for the second month in a row, and 3.2% YoY (gold):

Ex-shelter, prices declined -0.1% for the month, and were up 3.5% YoY:

This is a complete change of dynamic from a few years ago, when energy prices were somnolent and shelter was driving inflation. Now shelter is helping keep headline inflation from re-accelerating.

Given its importance, let’s parse shelter further. As noted above, shelter prices increased only 0.1%.  Both of its two components, rent of primary residence (gold) and “owners’ equivalent rent” (red) each rose 0.1% for the month. The former was up only 2.9% YoY, while the latter was still up 3.2%. Recalling that the shelter computation had to be kludged during the government shutdown last fall, I suggest ignoring the small bump afterward and focusing on the last few months vs. before the shoutdown. And doing so, it is likely that the slow disinflation there is persisting:

But for the second month in a row, the big reason for the YoY deceleration in headline prices was energy costs (including gasoline), which declined another -2.9% in July alone, reducing the YoY gains to 14.7%:

Now let’s turn to the current and former “problem children,” which I define as significant components which have risen more than 4% YoY. The headline here is also good news, as, although I won’t bother with graphs, new vehicle costs rose only 0.1% for the month and are only up 0.5% YoY, while used vehicles increased 0.4% monthly, but have gone down in price by an average of -1.9% YoY. This is a market which has been worked to a new equilibrium after a sharp 20% increase in prices immediately after the pandemic.

Another former “problem child” was tansportation services (including car insurance and repairs). Here the former has also digested the big post-pandemic increase and is following the flatness in vehicle prices. Insurance declined -0.3% monthly and on a YoY basis they are down -4.5%; while repair prices continue to be an issue, up 0.6% monthly and 6.6% YoY::

But a new problem child may be groceries. These increased only 0.1% for the month, but are up 3.0% YoY, with several items like fruits and vegetables up 5.1%, breakfast cereal up 4.1%, bread up 4.0%, meats up 4.5%, seafood up 7.0%, milk up 5.1%, coffee up 10.3%, and sugar up 7.4%:

The complaints people have been making about the price of groceries are showing up in the data. Some of this may be a result from the product recalls we have heard so much about in the past month, and some of it may be downstream of the increase in prices of things like fertilizer secondary to the closure of the Strait of Hormuz.

Finally,  the AI data center related categories of electricity and utility services rose 0.3% monthly and up 4.3% YoY%. The electricity component was up 0.1% monthly and 0.7% YoY, while gas and oil utility services rose 4.2% for the month and is up 4.3% YoY. Additionally, computer software and accessories (not shown) rose 0.5% for the month and are up 21.2% (!) YoY:

Before I conclude, here’s a look at what the sleepy increase in headline inflation did for real nonsupervisory hourly wages (blue), which rose less than 0.1%  for the month but remain down -0.1% YoY; and real aggregate nonsupervisory payrolls (red), which were unchanged for the month and are up 0.8% YoY, although both remain about -0.5% and -0.2% below their February and January peaks respectively:

Recall that real aggregate nonsupervisory wages are an excellent short leading indicators for recession. The added information for July is a double-edged sword. On the one hand, it is very rare for this metric to stay below peak for more than half a year without a recession occurring shortly thereafter. On the other hand, a good coincident marker for the onset of recession is when they turn negative YoY - and right now there is no evidence that that is about to happen. But with the Strait of Hormuz still closed, and US emergency reserves almost all depleted, just don’t expect gas prices to cooperate for a third month in a row.

A “quick and dirty” look at anticipated consumer inflation in July

 

 - by New Deal democrat

Tomorrow we’ll get the CPI report for July. At first I thought this might resume the upward spike of April and May — but maybe not.

My “quick and dirty” way to create a back of the envelope estimate of consumer inflation is to divide the change in gas prices (conservatively) by 16, and then add 0.15% for underlying upward pressure in non-energy areas. What is somewhat surprising is that, *on average,* gas prices declined -2.9% in July, from $4.05 to $3.93/gallon. Dividing by 16 gives us a decline of 0.2%, so if we add 0.15% to that, we get a change in CPI of between 0 and -0.1% (red in the graph below), compared with actual inflation through June (blue):

The Cleveland Fed, which has an inflation nowcast, is also expecting somewhat subdued inflation, at a 0.2% monthly increase:

This translates into a 3.5% YoY increase:

Which, following up my post yesterday, would at least be less bad for real nonsupervisory payrolls, which would decline -0.1% for the month, but remain higher by 4.1% YoY, and so even if contracting from their peak at the beginning of this year would not be signaling any imminent recession.

In July, the existing home market remained in its suboptimal equilibrium

 

 - by New Deal democrat

I wrote last month in my summary of that existing home sales report: “The housing market has reached a new, suboptimal equilibrium in sales, construction, prices, and inventory. Until some new positive or negative shock occurs (like a surprise new Fed hiking regimen), expect little change in this important leading sector of the economy, which is needless to say neutral for forecasting purposes.” 

While existing home sales are much less important in terms of economic impact, they are about 90% of the market, and generally trend in accord with new home sales. And, like new home sales, they are very much downstream of mortgage rates, which have been in a range of 6% to 7% for almost all of the past four years:

Although with the Iran war they have risen from 5.99% in February to 6.69% last week, they are still well within that range.

So, unsurprisingly, while existing home sales in July declined a seasonally adjusted -1.7% monthly to 4.06 million on an annualized basis, this is almost exactly in the middle of its range of between 3.85 - 4.30 annualized for the past three+ years:

If sales follow mortgage rates, prices follow sales, and unsurprisingly with rangebound sales, prices on a YoY basis have been relatively calm as well. These are not seasonally adjusted, so we look at them YoY. And since February of last year, there has been no YoY comparison higher than 3.0%. in July the YoY comparison was +2.0%. (For the record, on a monthly basis they declined -2.0%, but this is the typical seasonal pattern):

Again, this is similar to both Case Shiller (blue) and FHFA (red) repeat home sales indexes and the median price of new homes (gold), which are up only 1.1%, 2.2%, and down -3.0% YoY respectively:

This year the most lagging metric, inventory, has also fallen in line. In July, the YoY% change in existing home inventories was -0.6%. By contrast, as recently as last December it was up 7.9% YoY, and in March was up 4.5% YoY:

Again, we see similar flatness in YoY new home inventories (blue), down -2.4%, the active listing count of homes for sale nationwide (red), up 1.9%, and the new listing count (gold), up 2.4%:

So my conclusion this month is the same as last month. While there may be some slightly upward pressure on prices, with the background financial fundamentals the same, the existing home market has reached a suboptimal equilibrium, with something like a -500,000 decline in housing inventory available compared with ten years ago; and rangebound sales as well.

Scenes, both positive and negative, from the July employment report

 

 - by New Deal democrat

As per usual, there’s no economic news today, the first Monday after the employment report. So let’s dig into some detail about what was naughty and what was nice from Friday’s anemic report.

Let me start with the naughty, and in particular the -50,000 job losses (seasonally adjusted) in local education. While this is in large part an issue with difficult seasonal adjustments in the summer when many staff are temporarily laid off, Ben Casselman highlighted that it isn’t the only reason; school employment has been swan diving for a few months:

To which Joshua Goodman makes an excellent point:

I looked up these funds, and sure enough, they were paid out to school districts over a three year period that ended in September 2024. Funds allocated had to be spent by March of this year. So it looks like Joshua Goodman is correct.

But of course losses in education jobs weren’t the only negative point. After stabilizing in 2024 and 2025, the YoY% change in average hourly wages (blue) have also been decelerating sharply this year, even as inflation (red) has accelerated:

Historically wage growth decelerates during sharp slowdowns and recessions; and having inflation pick up even more has never been a good sign:

Additionally, aggregate nonsupervisory payrolls (blue) increased less than 0.1% in July:

Should consumer prices increase more than 0.1% in July, this will mark another downturn in real payrolls, which peaked in January. This would be an important yellow flag for recession. On the other hand, the real number has historically tended to turn negative YoY within a month or two before or after a recession begins, and almost certainly that will not happen this month unless there is a very sharp increase in consumer inflation on the order of 0.8% or more, which is unlikely:

And of course total employment has grown only 373,000 in the past 15 months, for an average of 25,000 per month. As shown in the graph below, employment (red) has increased only 0.3% since the end of 2024. Of the other three noteworthy monthly series tracked by the NBER for recession dating, real personal income less transfers (orange) has actually declined since then, having peaked in summer 2025:

Although there may have been a “mini-recession” last summer and autumn, while the consumer-side metrics as per above have stalled or declined, the economy has been kept out of recession by the producer side, via industrial production (blue) and real sales (green).

In addition to the bad data, there was some mixed data in the form of aggregate hours worked, which declined -0.1% for the month. Historically, hours decline more intensely than jobs, and turn negative YoY before jobs do as well:

But here’s what the last several years look like:

Despite the monthly decline, on a YoY basis hours have improved compared with the total number of jobs, something that has typically happened during recoveries from slowdowns or recessions.

There was also some positive data. First, as forecast by the declines in jobless claims (Blue, right scale), the unemployment rate (red, left scale) declined to an 18 month low of 4.1%:

Additionally, the leading sectors of manufacturing employment (red), construction (gold) and goods production as a whole (blue) all saw increases in the month:

And the average workweek in manufacturing increased to a new post-pandemic high:

Of course, much of this is tied to the AI data center construction Boom, so cross your fingers that it does not prove to be a bubble. I do think that this positive trend will have to reverse before any recession might begin.