Housing under construction continues to levitate

 

 - by New Deal democrat

Although they aren’t the most leading of housing metrics, because of supply-chain issues during the pandemic, housing units under construction has been the most important one, because they represent the actual economic activity of construction. With the exception of 2001, which was an investment-led downturn, they also have always turned down significantly before a recession has begun. Further, as I wrote there months ago and as shown in the below graph, “multi-family units under construction, which typically turn after single family units, have also usually (except for 2008 and the pandemic) turned down before recessions have begun:”

And this morning’s residential construction report for November showed that housing units under construction *still* have not turned down significantly. In fact, total units under construction (blue in the graph below) rose a seasonally adjusted 11,000, only 25,000 below their October 2022 peak. Almost all of that was due to an increase in single family construction (red), while multi-unit structures under construction (gold) declined 1,000:

This is not recessionary, especially keeping in mind that the manufacturing sector is not as important for forecasting purposes as it was before the China manufacturing shock that began in 2000, so construction has become more important.

The most leading of the construction metrics is permits (blue below), which declined 38,000 for the month, and the least noisy sub-sector is single family permits (red), which rose 7,000 and are 228,000 above their January 2023 low. On the other hand, multi family permits (gold) declined to a new three year low:

The likelihood is that single family permits will flatten or decline slightly in the immediate months ahead, as they typically follow the much more noisy single family home sales (blue below) by several months, and the trend there has turned down:

For completeness’ sake, below are total (blue), single family (red), and multi-unit (gold) housing starts, which are much noisier than permits and tend to lag a month or two behind them:

Starts rose sharply - by 191,000 and are 255,000 above their August lows, driven mainly by the upturn in single family permits. This simply confirms the increase in permits as shown above.

Finally, since housing follows interest rates, here is an update of the YoY change in Treasury interest rates (inverted, blue) vs. the YoY change in single family permits (/10 for scale):

The big increase in permits this year has not been supported by interest rates through November, which continued to be higher (hence suggesting lower activity) than one year ago. This is yet more reason to believe that the recent increase in permits will flatten out or decline in the near future.

To summarize, both interest rates and single family sales indicate that it is unlikely the recent increase in permits and starts will continue. Additionally, the big surge in the past several years in multi-family unit construction is likely to follow permits in turning down. But there is no recessionary import in the present trend.

Have wages “really” increased since before the pandemic?

 

 - by New Deal democrat

All of the remaining 2023 data - housing sales and construction, and personal income and spending, as well as the Index of Leading Indicators - will be reported this week. After Christmas, only initial claims will be reported next week.

Today there is no significant data. But over the weekend, a little tiff erupted on another site that I read, about whether real wages or earnings had increased since before the pandemic. This caused me to go back into the time capsule to when, about 10 years ago, I used to report quarterly on “four measures of real wages,” and take a look. Below is what I found.

There are a variety of economic data series to track both average and median wages:

• 1. The most commonly known measure is that of average hourly pay for nonsupervisory workers, which is part of the monthly jobs report.
• 2. The Bureau of Labor Statistics, which conducts the household employment survey, also reports "usual weekly earnings" for full-time workers each quarter.
• 3. The BLS also measures the Employment Cost Index quarterly.
• 4. The BLS also measures "business sector real compensation per hour" quarterly.

Below are all 4, normed to 100 as of Q4 2019, the quarter before the pandemic hit, updated through Q3 of this year:

To cut to the chase, all of them except for the Employment Cost Index for wages, show increases since before the pandemic. Real average hourly earnings through September were up 2.7%, real hourly compensation for workers was up 2.3%, and real usual weekly earnings were up 0.8%. The employment cost index for wages, however, decreased -1.2%.

Normally this would be of some concern, because the employment cost index is designed to measured changes in wages for the same job, i.e., it adjusts for differences in the number of people employed in various sectors. And indeed, when we take out inflation and measure the quarterly *nominal* percent change in each index, only the employment cost index rose during the 2020 shutdowns, indicating that once we take out compositional changes in the labor force, wages continued to rise slightly:

Just as significantly, though, note that the point where the employment cost index did not rise nearly as much as the others was during the 2021 jobs boom. Most likely what has happened is that the E.C.I., because of its composition, did not pick up the difference in wage gains between “job switchers” and “job stayers,” where the former have gotten much bigger wage increases than the latter: 

And with the extremely tight labor market of the last several years, there have been a record number of job switchers, as we know from the monthly voluntary “Quits” data in the JOLTS survey:

OK, but if the labor market is so tight, why haven’t wages grown even more?

To answer that, let’s take a look at the longer term trend in all 4 measures going back to the end of the Great Recession in 2009:

Real wages by all measures declined for several years into the recovery, even though the unemployment rate was decreasing, before picking up and continuing to increase for the rest of the expansion.

The answer to that, in turn, can be found by looking at the YoY% change in the inflation rate (blue in the graph below, right scale) and in particular picking out the change in gas prices (red, left scale):

At the bottom of the Great Recession, gas prices got as low as $1.60/gallon. As the economy began to expand, they rapidly recovered to $4/gallon, giving rise to what I called the “oil choke collar” restraining economic growth. Similarly, after the pandemic, and especially with Russia’s invasion of Ukraine, gas prices, as well as the big 2021 springtime stimulus spending spree, helped drive inflation sharply higher. It took a while for wages to catch up.

So the verdict is: yes, real wages have increased since before the recession. But a necessary part of that increase has been the massive switching of jobs from lower to higher paying positions by workers who, for once, had leverage.

Weekly Indicators for December 11 - 15 at Seeking Alpha

 

 - by New Deal democrat

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

The big decline in long term interest rates this week in the wake of the Fed’s announced “pivot” towards lowering rates created one of the biggest changes in the long leading indicators for several years. Meanwhile most of the coincident indicators continue to speak of a strong economy.

The intersection that is going to tell the tale going forward in the next 6 months or so is with the short leading indicators, including manufacturing, construction, and commodity prices.

As usual, clicking over and reading will bring you up to the virtual moment as to the economic data, and reward me a little bit for organizing it for you.

Industrial production remains below late 2022 peak even after end of UAW strike

 

 - by New Deal democrat

Industrial production historically has been the King of Coincident Indicators, turning up and down at the onset and end of recessions in the past. But as I wrote last month there are signs that has changed in the past 20+ years since China was admitted to normal trade relationships with the US. Because manufacturing is a much smaller share of domestic economic activity, and employment, downturns which before 2000 would always have meant recession probably do not do so now.

More specifically to the past several months, we had a very effective UAW strike in the motor vehicle industry which distorted both of the last two months. To wit, after revisions motor vehicle production in October declined -9.9%, and this morning’s report indicates a rebound in November of 7.1% which obviously is not complete.

More broadly, industrial production as a whole rebounded 0.2% in November. Manufacturing production increased 0.3%. But there were negative revisions of -0.3% to total production, and -0.2% to manufacturing, meaning that after rounding on net total production was unchanged, and manufacturing only improved 0.2%. The below graph show their levels compared with their post-pandemic peaks in September 2022:

Total production remains down -0.8% since then, and manufacturing down -1.4%.

On a YoY basis, total production is down -0.4%, and manufacturing production is down -0.8%. Going back over 100 years before 2016, there have only been 6 occasions when negative readings did not coincide with a recession within the next 6 months, and on only 2 of those occasions did the negative readings last more than one month:

But that changed in 2016, when production was down YoY for over a year, and yet no recession happened:

In other words, while industrial production remains recessionary, it has not been enough to produce a recession in the entire economy without the participation of other sectors, most notably construction.

Real retail sales mildly positive, but still suggest further deceleration in job gains

 

 - by New Deal democrat

Before proceeding further, I should mention - and should have mentioned as to jobless claims - that we are in that part of the year where seasonality often wreaks havoc, so outsized gains or losses should be taken with a grain of salt.

This is particularly true as to YoY comparisons of retail claims, because last year November and December stunk, and then there was a huge rebound in January. And this morning’s numbers, as well as the next two reports, will be compared YoY with a pretty big whipsaw.

With that out of the way, nominal retail sales increased 0.3% in November. After adjusting for inflation, they increased 0.2%. Meanwhile October was revised down by -0.1%, so the net real gain was only +0.1%. On an absolute scale, real retail sales are still slightly more than -2% below their April 2022 peak, although they have been improving pretty consistently if slowly all this year:

The relative improvement is best shown in the YoY comparisons. The first graph below shows the past 30 years of real retail sales YoY (dark blue), and also real personal consumption expenditures for goods (light blue), which tend to follow a similar trend. Additionally, the YoY gains are /2 for purposes of comparing with employment grains (red):

As I have pointed out many times, real retail sales /2, while noisy, is a very good short leading indicator for the trend in employment. Put another way, consumption leads employment.

Here is the post-pandemic record of the same:

One year ago was one of the very few times that a YoY negative number for real retail sales did not accurately forecast a recession (gas prices going from $5 to $3/gallon, and a general post-pandemic decline in supplier prices can work wonders for the economy!). 

Nevertheless real retail sales continue to forecast some further deceleration in jobs gains in the coming months.

Jobless claims: good news all around

 

 - by New Deal democrat

This was one of the best weeks as to jobless claims all year. Initial claims declined -17,000 to 202,000, a tie for the 2nd lowest number in 10 months. The four week average declined -7,750 to 213,250. With the usual one week delay, continuing claims rose 20,000 to 1.876 million:

Even more importantly for forecasting purposes, the YoY% changes both for the weekly number and the four week average were negative, i.e., lower than one year ago, at -1.9% and -0.1% respectively. Meanwhile continuing claims, while elevated at +17.2%, are nonetheless lower on a YoY basis than at any point in the past 8 months:

This makes initial claims an outright positive for the economy in the near future. And it cuts against the idea that continuing claims were forecasting a recession.

Since initial claims lead the unemployment rate, this also adds a level of comfort to the forecast that the elevated comparisons earlier this year will not trigger the “Sahm rule” as to the YoY change in the unemployment rate:

To reiterate, simply: good news all around.

Producer prices, “sticky” consumer prices - basically, everything except shelter show nearly complete abatement of inflation

 

 - by New Deal democrat

The producer price index released this morning for November is yet further confirmation that inflation ex-shelter is simply not in the pipeline. Both total and core PPI were unchanged for the month. Both commodities (blue) and finished goods (red) declined by another -0.5%, as shown in the below graph normed to 100 just before the pandemic, and including headline CPI as well (gray):

On a YoY basis, commodities are down -3.6%, and finished goods inputs down-0.9%, compared with the 3.1% rise in consumer prices:

Some of this difference probably remains producers refusing to pass on price decreases to consumers, and since consumers still have more income and savings in real terms than they had before the pandemic, consumers are able to pay those increased costs. 

And of course, a big part of the difference is shelter. The idea that the official measure of shelter lags reality seems to have gained increased currency. Steve Liesman of CNBC made a point of it this morning, saying it lagged by 3 Quarters. Meanwhile Scott Grannis (a/k/a Calafia Beach Pundit) put up the below graph with an 18 month lag:

After the official CPI came out yesterday, the Cleveland Fed posted its “sticky price” CPI, which also showed that, absent shelter, even sticky prices are up only 3.1%, vs. both total and core sticky price inflation including shelter, both up 4.7%:

One important question going forward is whether the supply chain has been fully “de-kinked.” The Goldman Sachs Supply Chain Index suggests that it has. Before the pandemic, the average typically varied between 0 and -1 on the index. After the huge pandemic-related increase, it went below -1 earlier this year. In November, for the first time since summer 2022, it went above 0 again:

The Bloomberg Industrial Metals Index, which I use because it does not include gas and oil, also peaked in early 2022, and has settled into a flat to slightly declining trend for the last 6 months:

In summary, pressures on producer prices have completely abated. The un-kinking of the pandemic related supply chain seems to have been completed. Consumer price increases ex-shelter also continue to be within a short distance to the Fed’s 2% target. What’s left is shelter, and ironically by constricting the supply of existing houses on the market, the Fed’s increases in that regard have become counterproductive.

Fortunately, as I pointed out yesterday, average and aggregate earnings have exceeded inflation for the past year. I do continue to be concerned that a substantial downturn in actual residential construction is going to materialize in the next few months, with all that it implies as a leading indicator.

Real aggregate payrolls rise to new high as CPI ex-shelter continues somnolent

 

 - by New Deal democrat

With few exceptions, the November CPI report once again demonstrated how important fictitious shelter is to its calculation, as well as how important the inflection point of $5 gas in June 2022 has been.

Headline inflation rose only 0.1% in November, and is up 3.1% YoY. Core inflation less food and energy increased 0.3%, and is up 4.0% YoY.

Shelter, which is 1/3rd of the headline index, and 40% of core, increased 0.4% for the month and was up 6.5% YoY. Perhaps more importantly, CPI ex-shelter was *down* -0.1% for the month, and up only 1.5% YoY.

Per the above, below I show headline (blue), core (red), and CPI less shelter (gold) all normed to 100 as of June 2022:

Ex-shelter, in the past 16 months prices have risen only 1.7%, while headline inflation has risen 4.5%, and core inflation has risen 6.5%.

In other words, take out shelter and inflation is a non-issue.

The former problem areas of new and used vehicles continue to cool, with new car prices declining -0.1% in November, while used car prices increased 1.0%. YoY new car prices are up only 1.3%, while used car prices are up 3.8%.:

But for several reasons, including that people are holding on to their older vehicles longer, which means they need more repairs, as well as the fact that insurance has had to keep pace with both the prices of new cars as well as the increased costs of repairs, the “transportation services” subset of CPI increased another 1.1% for the month, and is up 10.1% YoY:

The other current problem children are food away from home (restaurants), up 0.4% m/m and 5.5% YoY, and medical care commodities, up 0.5% m/m and 5.0% YoY. But these make up relatively small weights in the index, and are not nearly the problem that shelter continues to be.

As to shelter, here’s the update of Owners’ Equivalent Rent YoY compared with the Case Shiller index (recall that the former has a history of lagging the latter by 12 or more months):

As has been the ongoing case, YoY shelter rose more gradually than house prices, and is falling more gradually, but it is continuing to fall. At its current pace of decline, it will take another 12 months or more to be back into the Fed’s comfort zone.

Finally, we can also update real aggregate payrolls. These rose nominally by 0.9% in November, so after inflation they rose 0.8% to a new all-time high:

This means that in the aggregate average working and middle class Americans have more buying power now than ever before, and is a very potent positive for the economy over the next few months.

Scenes from the leading sectors of the November jobs report: why I sounded a note of caution

 

 - by New Deal democrat

I seem to have been something of a negative outlier with respect to last Friday’s jobs report. Not because I was downbeat - although I said there were “warning signs of weakness,” but almost all the other commentary I have seen was upbeat.

So today let’s take a look at the leading sectors in the jobs report, to show why I sounded a note of caution.

Let’s start with the manufacturing workweek, which is one of the 10 official components of the Index of Leading Indicators. It typically has turned down in the past even before manufacturing employment itself does. Here’s its record from the end of WW2 almost 80 years ago:

Almost always (exceptions 1966, 1985, and 1995) when manufacturing hours and overtime have declined more than by -.05 hours, a recession has followed. In the past 18 months, hours have declined by -0.9 hours, and overtime has declined by -1.1 hours.

Next, here is the long term look at manufacturing employment up until the “China shock” began in 2000:

With rare exception (twice in the 1950s plus 1981), manufacturing employment turned down before a recession began.

Here’s the recent record, also splitting out motor vehicle manufacturing, which has benefited from the unlinking of the supply chain:

In total, manufacturing employment has been close to unchanged for a year. Excluding motor vehicle employment, it has declined slightly.

But as I’ve written recently, manufacturing plays less of a role in the US economic cycle than it used to. So let’s look at other sectors as well.

Another jobs sector that has always turned at least flat in advance of a recessions is construction, and residential building construction has always declined:

Residential building construction employment has been flat for almost a year, while general construction employment (aided by the restoring of manufacturing capacity) has continued to increase:

Typically residential construction employment has coincided with the number of residential units under construction, and it has done so this year as well:

Another leading sector of employment has been truck transportation, which with the exception of 2000 has also turned down in advance of recessions:

It has turned down as well in the past half year:

All of the above are generally in the “goods” rather than “services” employment sector. Goods employment has generally flattened, and occasionally declined in advance of recessions:

It is still increasing, although at a more subdued rate this year:

Last week I discussed the importance of real spending on goods, noting that it led employment. On a YoY basis, real goods spending is up 2.1%, while goods employment is up 1.1%. The below graph norms both values to zero, to show how they compare historically:

Both real spending on goods and goods employment are actually weak on a historical basis, but both are consistent with a number of slowdowns that did not turn into recessions in the past. 

There is also at least one services segment which has also historically been leading: temporary help services:

These have declined sharply in the past year:

Finally, although they are not a leading sector, I noted in my report Friday that professional and business employment is down, and is virtually unchanged YoY. These are well paying jobs that are the backbone of the affluent portion of th middle class. They have never been at this level YoY without a recession occurring:

In summary, we have several sectors - the manufacturing work week, temporary help, professional and business jobs, and truck transportation - that are consistent with an oncoming recession right now. There are several others - manufacturing and residential construction - which are flat, but have not turned down as they frequently have in advance of recessions. And finally, there are several sectors - overall construction and goods production - which are in expansion mode.

As I wrote in the context of the personal income and spending report last week, with interest rates still elevated, I am paying particular attention to real spending on goods, as a short leading harbinger for whether weakness will spread further into the economy.