- by New Deal democrat
Several weeks ago, I wrote that “after 20 years, I think it’s time to examine whether the broad range of long leading indicators hold up.” This was primarily because, while when they have been positive the economy has followed suit 12 to 24 months later, several times they have been negative with no endogenous recession occurring thereafter: once in 2018-19 (COVID being the decisive external factor), and once in 2022-23. Historically they also had a false positive in 1966.
In that post I examined the 4 identified by Prof. Geoffrey Moore in the 1980s (and subequently used by ECRI), which were corporate bonds, corporate profits deflated by labor costs, real money supply, and housing permits; plus common measures of the yield curve, and also real retail sales per capita.
At the end of that post, I made mention of the impact of fiscal, i.e., government spending. In this post I want to take a more detailed look at the two exemplar misses: 1966 and 2022.
Here is what Prof. Moore’s four long leading indicators, plus the yield spread between the 10 year Treasury and the Fed Funds rate looked like in the 1960s, normed to 100 as of December 1965 (I’ve also inverted corporate bond yields, so that an increase shows as a negative, and recalibrated the scale of the yield curve so that an inverted curve shows below value “100”):
As you can see, in 1966 every long leading indicator turned negative with the exception of real money supply, which was flat. This was a very strong recessionary signal. And yet, among other things, neither real GDP nor employment turned down:
Additionally, while real retail sales turned negative YoY, real income less government transfers did not:
Similarly, in 2022, every indicator declined for almost the entire year. Thereafter, several turned neutral, while corporate profits rebounded beginning in 2023:
Again, this was a strong recessionary signal. But real GDP only declined slightly for one quarter at the beginning of 2022, was flat the next, and then recovered, while employment never declined at all:
In 2022-23, both real retail sales and real income less government transfers did briefly decline YoY, but not in sync:
So, what overcame these recessionary signals? It appears that two even more powerful forces were in play.
The first was a positive price shock in the form of declining producer prices.
To put this in context, here is the long term historical look at the YoY% change in PPI for finished goods (red) vs. CPI (blue):
With the exception of 1970, the onset of every other recession since the end of World War II has featured producer prices increasing faster than consumer prices. How this affects the economy is fairly straightforward: if producer input prices cannot be passed through to consumers, corporate profits will suffer, and cutbacks in hours and employment will begin.
But especially since 2000, there have been times where PPI inflation has equalled or exceeded CPI inflation without any recession. The simple dynamic going on here has been the decline in the labor share of producer income generated:
Now let’s look at 1966. There was a surge in PPI vs. CPI during 1965-66, but thereafter the PPI index abated:
In fact, beginning in September 1966, producer prices declined -1.1% through April 1967, relieving the pressure on employers.
A similar dynamic played out in 2022-23. From July 2022 through December 2023, producer prices declined -4.7%:
Both of these were “positive” price shocks, enabling corporate profits to increase without cutbacks to employment or an ensuing recession.
The second commonality to both episodes was huge government stimulus. Once again, here is the long term historical look, in log scale:
It’s easy to see that the mid-1960s, plus the stimulus payments during the Great Recession and COVID have been the three biggest expansions in government expenditure during this entire 75 year period.
In the 1960s, from the beginning of 1965 through the end of 1966, even accounting for inflation, there was a 25% increase in government spending per capita:
This was LBJ’s “guns and butter” spending on both the VIetnam War and Great Society domestic programs.
The COVID stimulus was even bigger, with the 2020 stimulus increasing real per capita government spending by over 80% and the 2021 stimulus by almost 70% compared with just before the pandemic:
As a result, in 2021 real spending was 15% higher than before the pandemic, while real income even without the stimulus payments was up about 4%. Although each of these declined at differing periods during 2022 and 2023, employment (gold) never caught up even to its pre-pandemic level until the middle of 2022:
In other words, there was still a huge shortfall in the number of employees needed to fulfill all the consumer spending that had been unleashed by the stimulus.
Finally, it’s worth noting that we do have the contrary example, of a contraction in federal spending leading to a recession, in the -10% reduction in New Deal spending that took place in 1937 through to be primarily responsible for the deep recession of 1938:
Let’s put this all together: the long leading indicators have been reliable in forecasting continued expansion, but several times have suggested a recession was likely ahead, but none materialized. In each of those cases, however, the endogenous progression of the economic cycle has been overcome by both (1) a positive supply shock in the form of disinflating or even deflating commodity prices; and (2) huge government stimulus programs.
Because we don’t have enough examples, it’s impossible to know which of these two factors was more important, or whether both were necessary. Additionally, because the former is an exogenous event and the latter is often geopolitical, it is very unlikely that they could be forecast. On the other hand, in both 1966 and 2022 the government stimulus programs were already in place, meaning they can be taken into account in long leading forecasting.
I still plan on doing some further re-examination, so stay tuned.
