The advance-decline line and the (maybe) AI-fueled consumer spending bubble

 

 - by New Deal democrat

I rarely comment on the financial markets directly, since my focus in on the economy and how it impacts ordinary working and middle class Americans, especially in the near future. But in some cases, the financial markets themselves play an important role in that picture. And this is one of those times.

Specifically, in terms of what is called the “wealth effect.” It means that when people’s wealth increases, even if it only on paper, they tend to spend some of that gain. According to Ned Davis research, it averages about 40% of the amount of the gain.

Well, since the post-“Liberation Day” April bottom, the stock market as measured by the S&P 500 has increased almost 35%. Meaning, for example, that a household that held $100,000 in that index in April has seen it grow by $35,000. A wealthier household that held $1 million, has gained $350,000. And so on.

As I have pointed out a number of times in the past several months, it is strong consumer spending that is keeping the economy going forward, despite recessionary signs elsewhere.

But how robust, or fragile, are those gains? One of the best historical measures I have found to be the “advance-decline line.” This subtracts the number of stocks which have lost value on any given day from the number which have gained. For example, if there are 1000 stocks in a bucket, and 550 have advanced and 450 declined, then the advance-decline line increases by 100. If the advance-decline line shows that the broad mass of stocks are participating in an advance, that is a good sign, because it suggests that many sectors are benefitting. But if it is narrow, or worse even declining, while the market index increases, that means that only a few stocks in a narrow sector or group of sectors are participating, suggesting trouble for the markets, and the economy, ahead.

There are two very good examples of the advance-decline line giving such a signal.

First, here is the late 2006-2007 time frame just before the onset of the Great Recession. The top graphic is the NYSE, the second is the advance-decline line for the S&P, and the bottom is for the S&P 500:

Note that even as stocks broke out to new highs at mid year, and again to their early autumn peak, the advance-decline line retreated in spring, and was even lower by autumn. This told us that the economy was resting on a narrow slice of sectors.

An even more breathtaking example is that of the Dotcom bubble of 1999-2000:

Here the advance-decline lines for the NYSE and S&P 500 declined sharply throughout the latter part of 1999 into 2000, even as the index raced ahead by almost 50%! This was a telltale sign, indicating that outside of the bubble economic prospects were not good at all.

So how has the advance-decline line been behaving this year? Here’s the answer for the S&P 500:

As indicated above, since the bottom in early April, stocks have increased by almost 35%. The advance-decline line participated fully in the springtime advance to mid-year.

But since early July, while the S&P 500 has increased almost another 10%, although it has not declined the advance-decline line has only increased by about 1%.

This is “yellow flag” territory. Slightly more stocks than not have been participating in the advance. But compared with the amount of the advance, it is quite narrow.

I would need the advance-decline line to actually turn down before it would signal a “red flag” for me.

But here is the important thing to keep in mind. If the AI-focused stock market is in a bubble - which is almost impossible to know while you are experiencing it - then whenever it pops, stock valuations are likely to plunge all the way back to their pre-bubble levels (and maybe overcompensate to the downside).

Which in turn brings us to the opposite of the “wealth effect.” Psychologists have estimated that this “negative” wealth effect is about twice as potent as the “positive” one. In layperson’s terms, people *really* hate to lose money. When that happens, they pull in their horns much more dramatically than when they spend some of their paper gains.

The conclusion here is that it doesn’t look like the turn is imminent. But if and when the turn comes, if other economic circumstances are close to what they are now, the contraction could be rather sudden and intense.

Consumer spending has continued to increase

 

 - by New Deal democrat

I concluded my post yesterday with the conclusion that “while housing and trucking are plainly recessionary, the broader manufacturing orders outlook and consumer purchases of vehicles are not.”  Today I want to expand on that by taking a broader look at the main coincident measures of expansion vs. recession, and focus on some up-to-date measures of consumer spending.

Let’s start with two coincident measures that either have peaked or look like they are peaking in the present: real personal income excluding government transfers (blue) and jobs (red), both normed to 100 as of April:

Real income peaked in April, and nonfarm payrolls as of the last official report jobs have only increased by less than 0.1% since then. If one believes ADP and several other private measures for September, there were either slight increases or an outright decline for that month, meaning the current measure is somewhere between unchanged and up 0.1% in the 5 months since April.

Further, looking at my favorite measure of real aggregate payrolls, nonsupervisory payrolls only increased 0.1% through July and remain lower than March, while total private payrolls actually declined -0.2% since April.:

In short, both real income and jobs look like they are on the cusp of or have actually started to decline.

But the story is different when we look at the other coincident measures of real sales (blue), production (red), and real spending (gold):

All three of these continued to increase after April, and two are at new highs. While industrial production is below June’s level, its manufacturing component (not shown) made a new post-pandemic high in the most recent report, for August.

Since as I have pointed out previously, real spending on services has historically continued to rise, albeit at a more subdued rate, right through all but the worst recessions, for forecasting purposes I rely on real personal spending on goods (gold) and real retail sales (blue):

Real retail sales made a new 2.5 year high in August, while real spending on goods made a new all-time high. Which supports the point that manufacturing production and sales/purchases have been continuing to increase.

And we do have information suggesting that at very least, there was no decline in September.

First of all, the Chicago Fed publishes a twice-monthly “Advance Retail Trade Summary,” which uses weekly private data to forecast monthly nominal and real retail sales ex-motor vehicles. Here is its graph of nominal sales through September:

The Chicago Fed estimates that retail sales rose 0.3% nominally in September, and were unchanged after accounting for inflation. 

Officially the US government has only reported vehicle sales through August, but the private source Omida (via Bill McBride) reported a 2% increase in the volume of sales in September vs. August:

Motor vehicle prices have been almost exactly unchanged for several years, so it is likely that this is a real $ increase in sales as well.

Finally, we also have the weekly Redbook same store sales update in YoY terms:

This continuus the strong nominal increases in excess of 5% and even over 6% that we have seen for the past several months.

In short, while several important coincident measures of the economy may have peaked, in an economy that is 70% consumer spending, until I see evidence that it has stalled as well it is hard to conclude that a recession is imminent.

Auto and truck purchases give conflicting signals on expansion vs. recession

 

 - by New Deal democrat

The typical post-jobs report lull in the data is amplified this month (of course) by the fact that there was no jobs report this month! If there is a tiny silver lining beginning to appear, it is that the Administration is making noises about reinstating the health care subsidies that has been the key “ask” by Congressional Democrats. We’ll see.

One important data point that came out last week that I didn’t report on was vehicle sales. To recap, after housing, vehicle sales are typically the next sector to roll over before a recession begins. And within those sales, heavy truck sales typically roll over first and most decisively vs. car and light truck sales, which fade later and are much noisier.

The update last week, for August, indicated a very sharp -5.2% decline in heavy truck sales for the month, bringing the total decline from their April 2023 peak to -27.4%. Here’s what the entire historical trend looks like:

Typically any sustained decline of -10% from peak has been enough to signal a near term recession is more likely than not, so this is a very serious number.

When on a YoY% basis, both housing under construction (blue in the graph below) as well as heavy truck sales (red) are off more than -10%, with no false positives and only one (non-pandemic) false negative, in 2000 (note graph adds 10% to both values so that is shows at the 0 line):

The above recession indicator has been triggered for the last two months.

But no recession indicator is perfect, and two other components are missing from this picture.

First, as per above, car and light weight truck sales have also always rolled over before the onset of recessions, although they are much noisier. Thus the below pre-pandemic YoY graph shows monthly sales in light blue, and the quarterly average in dark blue:

There are plenty of false positives here, but no false negatives. Which is important when we look at the post-pandemic graph:

The latest three month average of car and light truck sales is higher by 4% YoY. This is simply not recessionary.

Also, the “third leg of the stool” in leading durable goods is the wider manufacturers’ new orders component, which I reported on last week. To reiterate, here is the pre-pandemic historical record:

While these did roll over well in advance of the 2001 recession, they were still weakly positive going in to the 2008 recession.

Here is the post-pandemic look:

As per my report last week, these are up roughly 5% YoY through August.

In short, while housing and trucking are plainly recessionary, the broader manufacturing orders outlook and consumer purchases of vehicles are not.

Using tabulated State data, estimated initial and continuing claims last week continued in neutral range

 

 - by New Deal democrat

As we all know, initial and continuing jobless claims were not reported last Thursday. Which, by the way, is interesting, because they were reported during the lengthy 2013 shutdown and for at least part of the 2018-19 shutdown.

A large part of the reason they likely continued to be reported is that all the Federal government does is collect the information from the States, add it up, and then supply a seasonal adjustment. Which in turn means that we ought to be able to reconstruct the data by going to the States directly.

And on Friday, FRED helpfully posted all of the 50 States’, plus DC, Puerto Rico, and the Virgin Islands’ data. While it didn’t add them up, your trusty correspondent is capable of addition, which means that I can now present you with the YoY% changes in initial and continuing claims for last week, as well as what should be an extremely close estimate of the seasonally adjusted data.

Let me begin with the raw data. For the week ending September 27, unadjusted initial claims from all of the jurisdictions totaled 178,763 vs. 181,017 one year ago, or -1.3% less. Once we have this data, by adding it to the three previous weeks we arrive at a four week moving average of 4.2% higher. Continuing claims with the typical one week delay totaled 1,696,961 vs. 1,616,527 last year, or 5.2% higher. 

Since the YoY% changes are the most important, here are the FRED graphs through September 20 of that metric:

Note that both initial and continuing claims have been running in the vicinity of 5% higher YoY all year long, with the exception of most of July and August for initial claims (likely due to changes in the layoff and rehiring schedules for school districts). In other words, this past reporting week both the four week average of initial claims and continuing claims are right in line with their recent averages, while weekly initial claims were lower.

Now let’s turn to the seasonal adjustments. First, here is the comparison of seasonally adjusted weekly initial claims (dark blue) vs. non-seasonally adjusted claims (thin gray line):

Now here are SA (gold) and NSA (thin, grayish) continuing claims:

Both of these metrics are in the time of year where there is a hefty higher seasonal adjustment. Last year in the reference week, for initial claims, the adjustment was *1.2569; for continuing claims it was *1.1310. 

This gives us estimated weekly initial claims of 224,000, +6,000 higher week over week. Adding this to the previous three weeks of seasonally adjusted data already available gives us the four week moving average of 234,500, down -3,000 from the prior week. And estimated seasonally adjusted continuing claims were 1.919 million, -7,000 from the previous week. For graphic comparison purposes, here is are the seasonally adjusted numbers through the last federally reported week:

So long as the State by State data continues to get picked up, I will update this each week for the duration of the shutdown. In the meantime, the takeaway this week was a continuation of the recent “neutral” slightly higher YoY readings in the four week average of initial claims and also continuing claims.

Weekly Indicators for September 29 - October 3 at Seeking Alpha

 

 - by New Deal democrat

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

All of the high frequency data that I gather in those posts comes from the Fed or its regional banks, the States, or private firms. That means it is almost completely unaffected by the federal government shutdown, and can be continually updated for as long as the shutdown lasts.

While the majority of the short leading and coincident data in the set continues to indicate expansion, there are underlying signs of increasing weakness, masked by a declining US$ and increased spending by the top income brackets.

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