A “Big Picture” summary of why a recession still looks likely, even if it hasn’t occurred yet

 

 - by New Deal democrat

The below started out as a comment somewhere else, but it is too good a “big picture” summary of where the economy is today not to post it here. I still intend also to take a more detailed look at the housing market, but since the below lays the groundwork for that, I’ll post it later (maybe later today, maybe not).

What follows is hopefully an as dispassionate as possible very, very summarized version of why there is still ample reason to see a recession ahead, while explaining why it hasn’t occurred yet.

Let me start by saying that while a few people called for a recession last year, most did not. That’s because the downturn in GDP was largely inventory driven (so very transitory) and also by the gas price spike, which was geopolitical and speculative. When the West figured ways around Russian energy, that situation abated.

The situation this year is different. The easiest way to see it is for me to point you to the graphs of the Index of Leading Indicators. The index is the best K.I.S.S. method to forecast the economy, having the very inconvenient habit of being correct much more often than the average pundit:

 

[source: Conference Board via Advisor Perspectives]

The LEI is now down almost -10% - as much as it has been at the very bottom of all but the deepest recessions.

And yet we’re not in a recession, at least not yet. So something *is* different this time.

What’s different is that we are still dealing with the reverberations from the pandemic, in two ways.

The first is the resolving disconnect between sales and employment. When sales increase, you need more employees to handle them. Typically in expansions sales have grown faster than employment by 0-5% YoY. In the first year after the recession, sales grew by over 6%, and real retail sales a whopping 20(!) (thank you, stimulus) while employment initially declined by about 15% and was still down 6% one year later, the biggest disconnect ever. Below I divide real total sales by employment, to show how even with labor-cutting efficiencies in the production and sales processes over time, the 2020-21 surge was the most extreme divergence on record:

Payrolls have been closing the gap ever since - by growing sharply - but by my best estimate are still about 1.5 million below where they would need to be to catch up to trend sales growth (this was the essence of my post Friday). My best guess is that will take another 6-12 months.

Because of this, aggregate wage growth has outpaced inflation. 60 years of history says no recession occurs until inflation (red) outpaces aggregate payrolls growth (blue), which as a “fundamentals” manner which causes households in the aggregate to start cutting back spending:

As shown above, for the past year, payrolls growth has decelerated sharply, but inflation has decelerated even more sharply. Hence the aggregate “real” buying power of the American middle/working class has increased.

Going forward, with the tailwind of declining gas prices behind us, as trend job growth catches up with trend sales growth, my best guess is that inflation will trend more sideways, while aggregate payroll growth will continue to decelerate, meaning inflation may very well outpace payrolls very soon.

The second difference is also pandemic related. If you go back to your Econ 101 supply and demand curves, stimulus spending pushed the demand line to the right, increasing the amount demanded and also increasing prices. On the flip side, there were very significant supply bottlenecks that most especially affected vehicle production and housing construction. This created shortages, which going back to your Econ 101 supply and demand curves, pushed the supply line to the left. This decreased the amount demanded, but also drove up prices. 

As these supply chain kinks are being resolved, prices are becoming more stable, but both industries are taking a long time compared with previous expansions to make up the backlog, as shown by light vehicle sales (blue, left scale) and housing under construction (red, right scale):

Even though manufacturing as a whole has turned down (not shown), the vehicle production sector continues to show expansion. Similarly, even though housing permits and starts (not shown) turned down well over a year ago, it is taking an extremely long time for that to work its way through construction.

The above analysis is far from complete. It’s barely a summary of a summary. But while no method is perfect, my preference is to mechanically be on the K.I.S.S. side of the Index of Leading Indicators. Even if this time really is different in some respects, I would prefer *not* to reverse-engineer from a conclusion backwards of why they are wrong. 

Weekly Indicators for July 17 - 21 at Seeking Alpha

 

 - by New Deal democrat

My Weekly Indicators post is up at Seeking Alpha. 

At least when it comes to weekly measures of consumer spending, the “waiting for godot” recession seems to have finally arrived. Meanwhile other metrics have been picking up steam as to the near future. This suggests a period of wobbling ahead.

As usual, clicking over and reading will bring you up to the virtual moment, and reward me a little bit for my efforts.

How long until the historically tight jobs market reverts to trend?

 

 - by New Deal democrat

There are some very unusual cross-currents going on in the housing sector, revealed by yesterday’s existing home sales report. But it will take some time-intensive organization to present it to you, so I’m saving it for (hopefully) Monday.

In the meantime, let’s take another look at the job market and how it compares with consumer spending.

One of my favorite long-time graphs has been that the YoY% change in real retail sales, /2, forecasts the YoY% change in jobs growth, with a delay of some months. Here’s the latest update of that:

Works like a charm, until the pandemic. Then there’s a disconnect, first on the downside (2020) for jobs, as they declined over 5% vs. over 2% growth in sales), and then on the upside (2022-23), with an actual decline in sales vs. over 2% growth in jobs.

When will the disconnect resolve itself? For that, let’s take a different look at the same data. The above implies that sales grow faster than jobs,  vs. their long-term trend in the first part of an expansion, and sales grow slower than jobs in the last part. To show that, I’ve normed both real sales and job growth to 100 as of the mid-point for jobs growth in the last expansion in mid-2014. Because sales historically have tended to grow almost twice as fast as jobs, I account for that mathematically in the sales data.

Here’s what the last 30 years look like:

This shows that sales exploded in 2020 and 2021 with the pandemic stimulus, while jobs initially plummeted and then grew back strongly.

Now let’s zoom in on the post-pandemic era:

Real sales have been flat to slightly declining over the past 2 years. Meanwhile jobs have almost entirely caught up, and are now only 1% (about 1.5 million jobs) under trend compared with their pre-pandemic norm. This can be resolved by a further decline in sales, a further increase in jobs, or both.

Which brings me to the job openings data from the latest JOLTS report:

Over the past 16 months, job openings have declined about 50% towards their pre-pandemic level. If that trend continues, in about another 16 months the anomaly will have disappeared. 

The sales vs. jobs data highlighted previously above suggests that it will be quicker than that, on the order of 6 to 12 months. At that point the historically tight labor market will revert to mean, and an actual downturn in jobs in response to flagging sales would become much more likely.

Jobless claims: close but no cigar for the red flag

 

 - by New Deal democrat

Initial claims declined -9,000 to 228,000 last week, and the four week average declined -9,250 to 237,500. Continuing claims, with a one week delay, rose 33,000 to 1.754 million:

More importantly for forecasting purposes, initial claims are up 7.0% YoY, the four week average up 10.6%, and continuing claims up 30.8%:

Just as importantly, the average for July so far is about 233,000, only about 8.4% above last year’s average for the month.

My discipline requires 2 straight months, or 8 weeks in a row, of comparisons higher by 12.5% or more YoY. Otherwise, the spike could just be transitory noise. Thus, with this week’s number, the chain has been broken. Unless there are sharp increases in new claims in the remaining weeks of July, there is no red flag recession signal.

Despite that, since initial claims lead the unemployment rate, a slight increase in the unemployment rate during the next few months is still forecast:

Keep in mind that the unemployment rate in the above graph is rendered as a “percent of a percent,” so even a 10% YoY increase from 3.5% would sill only be about 3.9%, not enough to trigger the Sahm Rule.

June housing report: a tale a two diametrically opposed sectors

 

 - by New Deal democrat

Yesterday I wrote that housing under construction, along with new vehicle sales, were two important reasons that no economic downturn had occurred yet. Today’s report on housing construction for June showed two almost diametrically opposed trends: single family houses had a sharp increase in permits and starts, while units under construction made a 12 month low. Conversely, multi-family dwellings had sharp declines in permits and starts, while units under construction were at multi-decade highs.

So let’s break this month’s report down into those two categories: single-family vs. multi-family dwellings.

Permits (red in the graph below) are the most leading of the metrics, and single family permits have the least noise and most signal of any metric. These increased 20,000 annualized to a new 12 month high. Starts (blue) declined -70,000 annualized, but were nevertheless close to the 12 month high set last month. But units under construction (gold, right scale) declined -6,000 to a new 12 month low:

The story was completely reversed for multi-family dwellings. Permits (red) declined -73,000 to a new 12 month low, and starts (blue) declined -63,000 to the 3rd lowest level in 12 months. But units under construction (gold, right scale) increased 7,000 to not just a 12 month, but another all-time high:

It’s easy to see in the above how the data progresses from permits to starts to units under construction. Since mortgage rates lead permits, below are mortgage rates averaged monthly (inverted) vs. single family permits (red, right scale):

You can see that last year’s big increase in mortgage rates led to a big decrease in permits. The decline in mortgage rates earlier this year has led to an interim increase in permits. I suspect permits will decline slightly again as the recent increase in rates back to 7% filters through the system.

But the actual economic impact is via units under construction. Below I show total units under construction (blue, right scale)), as well as the single-family (red) and multi-family (gold) units components:

In total, units under construction have fallen slightly from their peak (meaning a very minor effect on the economy), while single-family units have declilned at a recessionary pace. But that has been almost totally offset by the record increase in multi-family units under construction. Undoubtedly the dominant reason for that is the big surge in housing prices after the pandemic, which priced single family dwellings out of the price range of many younger potential buyers. Since multi-unit apartments and condos tend to be less expensive, and thus somewhat of a replacement good, this is where the growth has been.

For forecasting purposes, most importantly, in the booming multi-family sector, permits have fallen to their lowest level since October 2020. Starts are close to their lowest level since the end of 2021. Units under construction have not peaked yet, and are probably several months away from peaking.  The exact reverse is true about single family units. Since multi-unit construction is over 60% of total activity, and more importantly because there is a much greater downside to multi-family construction if they follow permits, vs. limited upside for single family units, I anticipate downward pressure on the economy overall as a result.

Industrial and manufacturing production continue to falter

 

 - by New Deal democrat

I frequently call industrial production the King of Coincident Indicators, because so often the turning point in this metric has been at the peaks and troughs of the economy as a whole. That has not been the case since last September, when this indicator last peaked.

And it continued its declining trend in June. Total production declined -0.5%, and manufacturing production declined -0.3%:

On a YoY basis, total production is down -0.4%, and manufacturing production is down -0.3%:

As you can see, up until the recent past, such declines had almost always been recessionary. But since the “China shock” that began in 1999, there have been similar production declines that have not spread out into the wider economy.

Finally, here’s a look at the sub-sector of motor vehicle production:

This series is noisy, so while the big decline played a role in the declines in both total and manufacturing production in June, there is simply no way to know if this is simply one bad month, or the beginning of a downward trend.

The bottom line is that this important indicator continues to be negative. Recession as been avoided - at least so far - because of resolving supply bottlenecks in vehicle production and housing construction, and robust spending on services. As we saw above, June was a poor month for vehicle production. We’ll find out about housing construction later this week.

June retail sales continue to falter, with the important exception of motor vehicles

 

 - by New Deal democrat

As usual, retail sales is one of my favorite indicators, because it tells us so much about the 70% of the US economy that is consumption, as well as being a short leading indicator for employment. It has been faltering for the past year, and June was no different.

Last month retail sales increased 0.2% nominally, but because consumer prices also increased 0.2%, real retail sales were unchanged:

In real terms, retail sales remain -3.1% below their 2021 peak.

The YoY comparisons, which have been very negative, continued to be negative, although slightly less so. The below graph also shows real personal consumption on goods, which tends to closely track real retail sales, although with a different deflator which tends to make it more positive:

Although I won’t show the long term graph, the simple fact is that, going back 75 years, with rare exception a YoY decline of 2% in real retail sales has been recessionary.

Since real retail sales tend to lead the trend in employment by several months, nonfarm payrolls are also shown above in gold. The indication is that the deceleration in YoY employment gains is going to continue in the coming months.

Finally, because a unique aspect of the current economic environment is the outsized role played by motor vehicle sales, which had been severely constrained by supply shortages - leading to outsized price increases as well - here is the comparison among nominal total, motor vehicle, and sales ex-motor vehicles for the past several years, all normed to 100 as of just before the pandemic recession of 2020:

Note again that the above graph is nominal, not real, and is shown for comparison among the sectors. Real sales of motor vehicles increased 0.1% in June. This sector continues to be a boon for the economy, and - along with robust real spending on services, which aren’t part of retail sales - an important reason why no recession has occurred yet.

Weekly Indicators for July 10 - 14 at Seeking Alpha

 

 - by New Deal democrat

My Weekly Indicators post is up at Seeking Alpha.

Several indicators that had been stubbornly positive throughout the decline in leading metrics as of this past week finally turned either neutral or negative. Much as the dominant punditry at the moment is that the economy will actually stick a “soft landing,” these turns argue that instead consumers may finally be reining in purchases, and holdout metrics in the important construction sector may finally be turning down.

As usual, clicking over and reading will bring you thoroughly up to date, and bring me a little pocket change.

The American working class is doing better, thank you very much

 

 - by New Deal democrat

With the release of the CPI report earlier this week, I can update several measures of average middle class American income.

Real average hourly wages increased 0.2% in June, and are up 1.6% from one year ago:

Real aggregate payrolls for the entire spectrum of nonsupervisory American workers increased  0.3% in June and are up 3.1% from one year ago:

This is an excellent coincident marker of an expanding economy vs. recession. It tells us that, mainly thanks to declining gas price since June 2022, average American workers and their households have had a significant increase in their ability to buy things. A very good positive sign.

Finally, one last note about various measures of inflation. A few commentators highlighted the still-high +5.9% YoY increase in “sticking price” inflation. But as shown below, the 1 month and 3 month average changes in even that metric have declined sharply, to close to the Fed’s target range:

By any reasonable measure, inflation is no longer a major problem.

Initial claims move closer to red flag recession warning

 

 - by New Deal democrat

Initial jobless claims declined -12,000 last week to 237,000. The four week average declined -6,750 to 246,750. With a one week delay, continuing claims increased 11,000 to 1,729,000:

More importantly for forecasting purposes, the YoY% increases were 7.2% for the weekly number, 14.5% for the 4 week average, and 31.6% for continuing claims respectively:

This is the 4th week in a row that the 4 week average has been higher YoY by more than 12.5%. If this continues for several more weeks, that would be sufficient to trigger a red flag recession warning from this series.

Finally, since initial claims have a long-established history of leading the unemployment rate, here is the comparison of YoY claims averaged monthly (blue) as well as the 4 week average, compared with the YoY% change in the unemployment rate (gold):

Note importantly that the graph above measures the change in the unemployment rate as a “percent of a percent,” so for example a 10% increase from 3.6% would be just below 4.0%. To trigger the Sahm rule, we would need a 0.5% increase in the actual rate from the 3 month average of the lowest unemployment rate in the past year. To put it plainly: for forecasting purposes, we’re not there yet. But since the Sahm rule tells us retroactively when we have started a recession, it doesn’t affect my own forecasting analysis.

June inflation almost non-existent except for the fictitious measures of shelter

 - by New Deal democrat

 The message of this morning’s consumer inflation report was the same for almost everything except for the fictitious measures of shelter: sharp deceleration everywhere.

Let’s take a look:

Headline CPI up 0.2% m/m and 3.1% YoY (lowest since March 2021)

Core CPI up 0.2% m/m and 4.9% YoY (lowest since October 2021):

CPI less shelter up +0.2% and 0.7% YoY (lowest since February 2021):

New and Used vehicles: 0.0% and down -0.5% respectively m/m, and up +4.1% and down -5.2% YoY respectively:

Food up 0.1% m/m and 5.7% YoY:

But food is only up 0.3% in the 4 months since February:

Transportation services (replacement parts, repairs etc.) has also been a hot spot, and has also decelerated, up 0.4% m/m and up 8.2% YoY (but down from a peak of 15.2% YoY last October:

Finally, Owners Equivalent Rent up 0.4% m/m:

and 7.8% YoY (down from all time YoY high of 8.1% YoY in April):

Here’s what it looks like in comparison with house prices as measured by the Case Shiller national index YoY (/2.5 for scale):

Since the beginning of this year, monthly increases in OER have declined from 0.8% to 0.45%. YoY OER is probably going to be below 4.0% and maybe below 3.0% by the end of next winter.

To sum up: except for the fictitious measure of shelter, the only other remaining “hot spots” for inflation are new vehicles (but resolving as the supply chain issues have finally resolved) and transportation services. Food inflation has basically stopped in the past 4 months. 

And if actual new rent increase and house prices were substituted for the fictitious OER measure and the 12 month average used for leases, headline inflatioin would only be up about 0.8% YoY, and core inflation up 3.0%.

But I’m sure there’s some sticky price blah blah blah somewhere that will justify the Fed’s continued hawkishness.

Scenes from the employment report: important leading and coincident indicators of recession

 

 - by New Deal democrat

Here’s another detailed look at some significant data from last Friday’s employment report.  In this post I’ll look at some leading and coincident indicators.

First, about the unemployment rate. This is a lagging indicator coming out of recessions, but a leading one on the way in. While it is interesting that it declined -0.1% for the month, for forecasting purposes the YoY change is more important.

Here’s another look at the “jobless claims lead the unemployment rate” metric. The below graphs measure the YoY% change in jobless claims, averaged monthly, /10 for scale (blue). The red line is the change in the unemployment rate YoY, from which I subtract -0.1% for reasons I’ll explain after these first two graphs:

What these show is that, going back 50+ years, initial claims always turn higher YoY first, while the unemployment rate turns higher by +0.2% or more within 6 months before to 2 months after the the onset of recessions. There are only two exceptions: one month in 1985, and the near “double-dip” of 2002.

Here is the same graph for the past 24 months:

The unemployment rate was +0.1% higher YoY in May, and unchanged YoY in June. This is close to but not yet signalling the onset of a recession. But if jobless claims remain as elevated YoY in July and August as they were in June, that would almost certainly indicate that the unemployment rate trigger is shortly to follow.

Next, let’s take a look at two leading job sectors that haven’t quite turned yet, because of supply logjams in the Fed’s interest rate transmission mechanism.

Motor vehicle production had severe supply constraints all the way into 2022. As a result (not shown), only in the first half of this year did sales routinely meet or exceed 15 million on an annualized basis.

As a result, employment in vehicle manufacturing (blue in the graph below) has continued to increase sharply. Normally manufacturing employment (red) turns down well before a recession begins. But while it has flattened, it has not turned down:

As shown above, vehicle manufacturing employment is about 5-8% of all manufacturing employment.

This next graph below compares total manufacturing employment, with manufacturing employment excluding motor vehicles (blue):

Ex-motor vehicles, manufacturing employment has indeed turned down since the beginning of this year.

There’s a similar situation in housing construction. The first graph below compares housing units under construction (red) with residential construction employment (blue). Unsurprisingly the former leads the latter:

Since housing under construction is only down slightly from peak, residential construction employment has barely turned down at all.

Finally, let’s update one of my big coincident indicators of recession, the YoY% change in real aggregate payrolls. Below I’ve divided it into nominal payrolls (blue) vs. CPI (red), which hasn’t been reported yet for June:

You can see that, going back 50+ years, whenever YoY inflation exceeds YoY nominal payroll growth, a recession is just beginning.

Here’s the same graph since the onset of the pandemic:

Nominal payroll growth, while decelerating, has done slightly more slowly than inflation, meaning that real aggregate payrolls have continued to increase. Meaning more buying power for average Americans. 

An important issue is whether, with YoY comparisons now including gas prices that were declining from $5 (meaning they are less positive), the serendipitous comparison between payrolls and inflation will continue. And a big part of that issue is also, now that the supply bottlenecks in vehicle production and housing have abated, those two sectors of employment will finally roll over.

Scenes from the June employment report: consumption leads employment, goods vs. services edition

 

 - by New Deal democrat

No big new economic news today, so let’s take a more in-depth look at some of the information from Friday’s employment report. Today I’m going to focus on the division between goods and services.

As I’ve written many times in the past, consumption leads employment. Typically I have shown that via real retail sales. The variation I am going to use today is employing real personal spending on goods vs. services, and how consumption leads employment for each.

As a refresher, here is real personal spending on goods YoY (blue) vs. services (red), up until the pandemic:

Note that spending on goods is much more volatile than spending on services (in fact I’ve divided the result for goods by 1.5 so that services spending doesn’t just show up as squiggles). Most importantly, it tuns down YoY generally coincidently with the onset of recessions, whereas growth in services spending usually just decelerates. Also, while the two moved coincidently from 1960-90, since then spending on goods has usually led spending on services somewhat.

Here’s the same information since the pandemic (omitting the year of huge distortions):

Goods spending did fall below zero during much of 2022, and is only slightly above zero YoY now, while spending on services is much stronger.

Now let’s compare real spending on goods (blue) with employment in the goods sector (red) YoY, pre-pandemic:

With two exceptions (the mid 1980’s and late 1990’s) goods consumption leads goods employment. Note that this holds true even though due to globalization and offshoring, goods employment never rose nearly as much as goods consumption beginning with the 1980’s.

The same leading/lagging relationship holds true for consumption of services (blue) vs. employment in services (red):

Now let’s look at each post-pandemic. First, here is goods consumption vs. employment:

Goods spending recently peaked YoY in summer 2021, while as we should expect based on past history, goods employment did not peak until spring 2022. Goods spending rebounded somewhat as spending power increased with the decline in gas prices from $5 to $3/gallon in late 2022. Goods employment is still decelerating, and is only up 0.3% since February, or roughly at a 1% annual rate.

Here is services consumption vs. employment:

Both consumption and employment in the services sector have been much stronger than in the goods sector, and while both have been decelerating since the spring 2021 stimulus spree, consumption has decelerated faster compared with late 2021.

Since gas prices have been pretty stable this year, the tailwind for goods spending is subsiding. I expect goods spending to decelerate further YoY, and probably turn negative again, with goods employment following. My best guess is this will occur by the end of this year, possibly earlier.

As per past history, the deceleration in both consumption and employment in the goods sector is likely to be slower, but will follow goods spending and employment with lower growth if not an outright decline.