Saturday, April 22, 2023

Weekly Indicators for April 17 - 21 at Seeking Alpha

 

 - by New Deal democrat

My Weekly Indicators post is up at Seeking Alpha.

There’s more *extremely* slow deterioration in some coincident indicators of recession, but at the same time, the downturn has been telegraphed for so long that some leading indicators are on the verge of turning bullish again.

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

Friday, April 21, 2023

The economic tailwind from last autumn’s declining gas prices is probably over

 

 - by New Deal democrat


On Wednesday I discussed how gas prices, with an assist from higher stock prices leading to stock options being cashed in, was the primary reason why the coincident indicators hadn’t rolled over yet.


I wanted to explore that a little more: was the boost from lower gas prices going to allow for a “soft landing,” or a “modified limited rolling recession,” if you will? Or was the boost ending, where we could expect the other factors driving the economy to take precedence? Let me look at this two ways.

First, gas at the pump has to be funded by wages. So one way to measure the impact of gas prices on consumers is to compare the two. In order to take the effects of Russia’s invasion of Ukraine out of the picture, in the below graph I have normed both gas prices and non-managerial wages to 100 as of April 2 years ago. Here is the long-term graph:



We can see the periods of really cheap gas prices at the end of the 1990s, and during both recent recessions, as well as the big downturn in 2014-15. More recently, we can see that the run-down in prices in the second half of last year took us about to the historical norm. Since the beginning of this year, prices have turned “relatively” expensive, without engaging an actual “choke collar” on the economy, as they did in the first part of last year.

Here’s another look at the same data, this time dividing gas prices by hourly wages:



The stress caused by the Russian invasion of Ukraine is over, but the tailwind of the downturn in prices thereafter has ended as well.

Second, it’s been suggested that whenever gas prices hit a certain level of GDP, that has been enough to trigger a recession. I show that below by dividing gas prices by nominal GDP, and norming the data to 100 as of Q4 2007, the last time an oil price spike helped spark a recession:



Note at the far left how the impact of the invasion of Kuwait by Iraq in 1990 helped trigger that recession as well.

Last year’s run-up didn’t quite hit the 100 threshold, peaking at 90. This was enough to crimp GDP without actually causing a recession. By the end of Q4 that had entirely receded. We’ll find out about Q1 next week.

Let’s look at this same data on a YoY basis. Below I show the YoY% change, inverted, of gas prices averaged quarterly (so that a decrease in price shows as an increase, e.g.), together with YoY nominal GDP (*5 for scale):



Basically this shows that “the remedy for high (low) prices is high (low) prices.” Big run-ups in prices create GDP slowdowns about 1 year later, and big run-downs in prices an increase in YoY GDP a year later (but of course it’s not monocausal, e.g., the tech boom of the late 1990s and the shallow industrial recession of 2015-16 were not particularly in tune with gas prices).

This graph suggests that the effects of the big run-up in gas prices early last year have not yet fully been felt, and that the effects of the run-down in prices thereafter will probably show up by next year.

The bottom line for now is that gas prices probably were part of the GDP slowdown in the first 2 Quarters of last year, and probably a part of the rebound thereafter, up into Q1 of this year. But the tailwind is probably over beginning this Quarter. In other words, I expect the effects of other aspects of the economy to increase in salience beginning with this Quarter’s data.


Thursday, April 20, 2023

Jobless claims continue to warrant yellow caution flag, while continuing claims shade closer to crimson

 

 - by New Deal democrat


Initial claims (blue in the graph below) continued their recent track into recession caution territory this week, as they rose 5,000 to 245,000, 12.9% higher YoY and the 5th time in the last 7 weeks that claims have been 240,000 or above. The last time they were at this level was in January 2022. 


The more important 4 week moving average (red) declined -250 to 239,750, 10.6% higher than 1 year ago. This is the 4th week in a row that the YoY% change has been above 10%, but it has not yet crossed the 12.5% threshold that historically has been a recession warning. On an absolute basis, except for 2 of the 4 previous weeks, the highest it had been at this level was also January 2022.

Finally, continuing claims (gold) rose 61,000 to 1,865,000, 22.1% above their level one year ago, and the highest since November 2021:



Here is the YoY% change, which is more important at the moment:



The increase in continuing claims appears especially significant. Historically, continuing claims have lagged, and have not been higher YoY by 20% or more until after a recession had already started (below graph subtracts 20% so that a YoY 20% increase shows at the zero line):



The only two exceptions prior to the pandemic were 2 weeks in November and December 1979, just before the January start of the 1980 recession, and 1 week in November 1989, 8 months before the onset of the July 1990 recession. 

Parenthetically, it is important to note that the massive seasonal revisions which were announced 2 weeks ago did not significantly affect the YoY comparisons. 

For forecasting purposes, this metric continues to warrant a yellow but not red flag. But if continuing claims are over 20% for even one more week, that yellow will shade closer to orange or even crimson.

Wednesday, April 19, 2023

Coincident indicators hold on, mainly due to improvement in gas prices YoY

 

 - by New Deal democrat


I’ve been paying particular attention lately to the coincident indicators, because the leading indicators have telegraphed a recession for about half a year - so why isn’t it here yet???

A good representation of coincident indicators remaining positive is the Weekly Economic Index of the NY Fed:



It looked on track to turn negative at the beginning of the year, but has not deteriorated any further since. Can we isolate where the strength has been coming from?

Yes we can. The NY Fed helpfully tells us that the index is an amalgamation of 10 data series, 8 of which are in the public realm, and all 8 of which I have been keeping track of for the past 10 years in my “Weekly Indicators” posts. The 8 are: initial jobless claims, continuing claims, the American Staffing Index, gas usage, Redbook consumer spending, Rail traffic, tax withholding payments, and steel production.

So, which of these are now or at least have recently been positive?

Most importantly, gas usage. As I’ve noted a number of times, gas prices declining from $5 last June to $3 in December can do a world of good to economic statistics. Unsurprisingly, gas usage was at its worst YoY last summer, and turned positive this winter:



With gas prices still roughly $0.50 less than they were last year at this time, usage has improved to about 5% better YoY.

A second series which has improved considerably, and more surprisingly, is tax withholding payments. Here’s a graph through the beginning of April provided by the CA Department of Taxation, which is similar to tax withholding payments for the nation as a whole:



Tax payments declined considerably YoY in the last few months of 2022, but then stabilized beginning in January. The CA Department of Taxation had attributed the decline to the failure of stock options to vest (and so be cashed in) as 2022 progressed, due to the stock market decline. Stocks bottomed in October and have been in a positive trend since, so likely stock options have been more attractively priced this year. So their explanation makes sense.

For the record, for the first 11 days of April, withholding tax payments are ahead by about 7% compared with last year, $149.4 Billion (11 days in) vs. $140.0 Billion last year.

One other series, which had been looking better YoY, has deteriorated in April: steel production. This had been down over 10% last year, before improving earlier this year and actually turning positive YoY in March. But now it is back down about -5% YoY:



A second deteriorating series is staffing. This was positive but increasingly less so last autumn, then turned neutral, and solidly negative beginning in February:




The index is now down -7% YoY. I consider it more of a leading than coincident indicator, since it correlates with temporary help in the payrolls report, which typically turns down well before jobs as a whole.

Finally, consumer spending as measured by Redbook, which had been up over 15% last summer, has been almost consistently deteriorating since then, and as of the last reading this week was only up 1.1% YoY:



This series is on track to turn negative YoY in the next month, if the trend holds.

As a whole, the coincident data continues to deteriorate. It has been helped considerably by lower gas prices, with a big assist from increasing stock prices. I do not think this will last long, but we’ll see.

Tuesday, April 18, 2023

New housing construction appears to have bottomed; but expect further declines in construction employment ahead


 - by New Deal democrat


For the past few months, I’ve noted that new home sales, which while very volatile frequently are the first metric to signal a change in trend, appeared to have bottomed by early last autumn. This morning’s report on housing permits and starts appears to have confirmed that signal. 

While total housing permits (gold in the graph below) declined -137,000 on a seasonally adjusted annual rate from last month, they remained higher than their November-January lows by about 75,000. Starts (blue), which are noisier and tend to lag a month or so, also declined -12,000, but remained 86,000 higher than their January low. Most importantly, single family permits (red, right scale) which are the least volatile measure of the three, rose 32,000, for the second straight monthly increase, and are now almost 100,000 above their January low:



It is very likely that the bottom for the housing sales market is in. Remember that sales follow interest rates, and in particular mortgage rates, which peaked last October and November. Below is a the latest update of the graph comparing the YoY change in mortgage rates (blue, inverted, *10 for scale)  with the YoY% change in both total and single family permits:



This is all good news, despite the monthly declines in total permits and starts.

The one important piece of bad news is that total housing units under construction declined again (blue in the graphs below), and are now -2.2% below their October peak. As I’ve noted monthly for awhile now, this is the metric that shows the actual total economic activity of the housing market, so it shows that housing is now detracting from GDP. Further, once construction turns down, shortly thereafter so does construction employment (red). Here is the historical view until the pandemic:



Now here is the last year, with both metrics normed to 100 as of their peak months:



Nonfarm payrolls has been the main coincident indicator holding up the economy, and construction employment is one of the leading sectors of the jobs market overall. This morning’s report tells us to expect further declines in that jobs sector.


Monday, April 17, 2023

Two “fundamental” indicators for the American middle/working class and the economy


 - by New Deal democrat


This week is a little light on data, except for housing permits and starts (Tuesday) and existing home sales (Thursday), so let me catch up on a few other indicators.

In particular, two of my favorite indicators are based on “fundamentals.” Basically, how much the average American is earning, and how much they are spending. Needless to say, we want both of them to be increasing. That’s because, as I have often said, consumption leads employment. If Americans are cutting back on spending, then cutbacks in employment will soon follow.

Because consumption leads, let’s start with spending, i.e., real retail sales YoY. This data series goes back 75 years. And it has been very reliable. Here’s the graph:



Leaving aside the pandemic, real retail sales has *always* turned negative YoY within about 6 months before the onset of a recession, except for two times in the 1950s where it turned negative YoY 4 and 5 months into the recession.

There have been some false positives (13 in total), where negative monthly YoY readings were not associated with a recession, but a majority of those were never worse than -1.0% YoY (the exceptions being 1951-52, 1956, 1966-67, 1987, and 2002). Further, a majority of the 13 times only lasted for 1 month, and only 3 have lasted for longer than 2 months in a row (1951-52, 1966-67, and 2002). 

In short, even a 1 month negative YoY reading is a yellow flag that a recession might be near, and if it goes on longer than 2 months with at least one negative reading of more than -1.0%, almost certainly a recession has either just started or will within the next few months.

Now let’s turn to employment, in the form of real aggregate payrolls for non-supervisory employees. In other words, in real terms the total pay that the American working/middle class is taking home. This has a 60 year track record, and has also been very reliable:



There have been *no* false positives, ie., where the indicator signaled but there was no recession. There have been 5 times when the indicator did not turn negative until several months into a recession: 1970 (4 months), 1974 (3 months), 1981 (3 months), 2001 (1 month), and 2008 (5 months). But with the exception of 1981, in the other 4 episodes this metric was in a clear and severe downtrend during those months.

Now let’s see what both look like together in the 50+ years both were in existence prior to the pandemic:



What this shows is that, if both of these two indicators are positive, you could be sure that you are not in a recession. With the sole exception of one month in late 2002, if both are negative you could be sure that you are either just before, during, or coming out of a recession. And frequently real retail sales had turned negative a few months before real aggregate payrolls.

Now let’s look at the past 18 months:




Real retail sales have been negative YoY for 7 of the past 13 months. I have discounted the negatives from last spring, because they were in contrast with.the spring stimulus spending spree of 2021. But 4 of the past 5 months have also been negative, and last month (March) by -1.9%.  Putting both indicators together, with the exception of 1966-67 and arguably the near double-dip of late 2002, there has never been a time in the past 60 years where a downturn this big for this duration has not meant a recession.

Most importantly, at the moment real aggregate payrolls are still positive, and they are not declining. Because, for reasons I discussed last week, I expect consumer inflation to only be about +3.2% YoY after June, for aggregate real payrolls to turn negative, there will have to be a pronounced slowdown in either hours, or jobs, or wage growth, or a combination of the three.