Due to my work with the BaR Analysis Grid©, I've noticed that prior to recessions there tends to be a group of economic indicators that weakens first and signals a pending recession. This has prompted me to look at the last three recessions to see if this cohort consists of the same indicators. As will be shown, although there are some common indicators, the fact that there is a leading cohort may be more important than which measures are included.
I will be using 19 economic indicators that cover key aspects of the economy. These indicators are the same that are plotted on the BaR Analysis Grid©. Chart 1 shows the indicators, their abbreviations on the BaR, and their data sources.
One problem with using historical data is that fewer indicators were reported. At the start of the 2007–2009 recession, one of the 19 indicators, ISM nonmanufacturing index, was not available. Going back to the start of the 2001 recession, only 17 of the economic indicators plotted on the BaR were being reported, and going back to the 1990 - 1991 recession, only 10 indicators were being reported. But, even with the limited data, the BaR functions well, showing how the economy moved into a recession. This is evident when you view the BaR grids show the business cycle for the last three business cycles (see here).
(If you are not familiar with the BaR, here is one of the earlier articles on Seeking Alpha about it (the methodology has been refined since this article). You can also read about the BaR here on the Econ P.I. website.)
Economic indicators that led the decline prior to 2007-2009 recession
After the business cycle peaks and the economy begins to weaken, not all economic indicators decline at the same rate (the business cycle peak is a key period in a business cycle, which I wrote about here). Initially, there tends to be a small group of indicators that is a leading signal of a pending recession. But, over time, more and more measures falter until it becomes obvious a recession is looming.
On the BaR, an important signal of decline seems to be when indicators fall below 10% on the vertical axis, which means the measures are within 10% of the baseline (recession threshold). This is true even when a negative trending indicator swings into one of the right-side quadrants, which signal a positive rate of change. Such swings are due to regular ups and downs that all indicators experience. What matters is that the measure has steadily declined for a period of months and is nearing the baseline.
Grid 1 shows the economy six months prior to the start of the 2007–2009 recession. The red dotted line marks 10% on the vertical axis. With a couple of exceptions that I'll explain in a moment, the area below the red line captures what were the leading indicators of the pending recession at this point of time. This is why I have chosen the six-month point, as this is when there is a small group of early indicators signalling a recession.
Table 1 shows the five indicators shown in Grid 1 that are within the recession range. Not surprisingly, existing home sales are underwater as the housing bubble had burst nearly two years earlier.
The yield curve spread (measured by 10-year Treasury notes minus 3-month Treasury bills) inverted briefly in January 2006 and then again in February of 2006, but an extended inversion didn’t start until July 2006, 17 months before the December 2007 recession start.
On the BaR the yield curve spread is pushed forward 12 months as this better captures the nature of the yield curve as a forward-looking indicator (the Cleveland Fed also uses 12 months for its predictive model - read here). But, the 12 months is an approximation. Yield curve inversions rarely occur exactly 12 months ahead of a recession. The range for the last three recessions was between 10 and 17 months.
I have also included the MoC in Table 1, which is 14.9% above the baseline at this point. It has declined from a high of 27%, which was reached during the first quarter of 2006.
Other measures that should be considered are ones that are not below the 10% mark, but have seen significant decrease over the past three months. Two indicators are worth noting: building permits and consumer sentiment. Table 2 shows the percent above baseline at nine and six months prior to the recession. Over the three-month period, building permits decreased 49% and consumer sentiment decreased 40%.
Like existing home sales, building permits would be expected to decline because of the housing market collapse. The majority of the indicators in Tables 1 and 2 are consumer-related indicators. Consequently, the decline in consumer sentiment fits with this theme.
Grid 2 shows the BaR three months prior to the 2007–2009 recession. At this point, 10 indicators that are either negative or less than 10% from the baseline, a majority of the 18 plotted measures.
Table 3 shows the 10 indicators that are in the recession range in Grid 2. As can be seen, the increase in indicators within this range has also brought the MoC below 10%. This is a sure indication that the economy is nearing a serious slowdown.
For indicators that are not below the 10% mark, there were three that had significant decreases from over the six-month to three-month pre-recession period. As shown in Table 4, the St. Louis Fed Financial Stress Index decreased 62%, the Credit Manager’s Index decreased 28%, and ISM Manufacturing decreased 21%.
As can be seen, at the three-month point, nearly all of the indicators have turned south, showing that the economy will soon be in a recession. The leading indicators from the six-month point have been joined by indicators that cover all aspects of the economy: employment, consumer goods and services, manufacturing, and financial.
Economic indicators that led the decline previous to 2001 recession
For the remaining analysis, I will only use the tables to show the indicators that are below the 10% mark on the vertical axis. However, if you want to see the grids prior to the 2001 recession, use this link to Econ P.I. (scroll down to the 2001 grid).
Table 5 shows that six months prior to the 2001 recession, six of 17 measures were within the recession range.
In addition, although it was not below 10% on the vertical axis of the BaR, the yield curve spread had inverted. The yield curve first inverted briefly in September 1998, then again briefly in April 2000. It began an extended inversion in July 2000, 10 months before the March 2001 recession start. Because the BaR pushes the yield curve spread forward 12 months, it’s not perfectly capturing the inversion. However, once the yield curve spread has inverted, it has reached its recession threshold and should be identified as such, which is why it is included on Table 5.
In Table 6, I’m showing the indicators that are above 10% on the vertical axis of the BaR, but showed significant declines from nine to six months prior to the recession.
Table 7 shows the indicators that are within less than 10% range three months before the 2001 recession. There are 10 indicators in this range, plus the yield curve spread. Not surprisingly, with a majority of measures below 10%, the MoC has dropped to 4.1%.
There are two indicators that had not reached the recession range of less than 10%, but had notable declines: temporary employment and consumer sentiment. Over the three months, temporary employment had declined 38% and consumer sentiment had declined 32%, as shown in Table 8.
Once again, at the three-month point, the economy is pretty much off the rails.
Economic indicators that led the decline previous to 1990-1991 recession
Table 9 shows the economic indicators that are less than 10% on the vertical axis of the BaR six months prior to the 1990–1991 recession. As I noted earlier, there are only 10 indicators plotted on the BaR for this period of time, of which six indicators are leading recession indicators. In addition, the yield curve first inverted in May 1989, 14 months prior to the recession start of July 1990. So, again I am including the yield curve spread as it is signaling a recession.
The MoC in Table 9 is also below the 10% mark. This may be due to the economy being weaker at this point than in Tables 1 and 5, or it could be due to the fact that fewer indicators are being tracked and the MoC is more volatile.
The remaining indicators that were above the 10% on the vertical axis did not show a significant decrease from the nine-month to six-month time frame.
As shown in Table 10, at the three-month point, eight of the ten indicators are below 10% on the vertical axis. With the MoC dropping to 4.5% above the baseline, this too indicates the end is near for the economy.
Common indicators six months prior to the past three recessions
Taking the data from Tables 1, 2, 5, 6, and 9, Table 11 shows the leading recession indicators, both those below 10% on the vertical axis of the BaR and those that showed significant three-month declines between the nine-month and six-month periods prior to the start of each recession.
As is well established, the yield curve spread preceded all three recessions. Of the remaining six indicators associated with the 2007-2009 recession, five preceded at least one other recession. Excluding the yield curve, of the eight indicators that led the 2001 recession, only two were connected to another recession. And finally, excluding the yield curve, of the six indicators that occurred before the 1990-1991 recession, four were associated with another recession.
In all likelihood, if nonfinancial corporate profits were reported monthly instead of quarterly, it would have also made the six-month list prior to the 2007–2009 recession, making it a leading indicator across all three recessions. As measured on the BaR, nonfinancial profits decreased from 21.1% at six months to prior to the recession to 1.3% at three months prior. (The BaR uses nonfinancial corporate profits as reported by the Bureau of Economic Analysis, after tax with IVA and CCadj).
The problem with using adjusted corporate profits is that they are generally released two months after the end of a quarter. This doesn't matter when we look at historic data, but it is less than timely for the present. My way around this is to follow corporate earnings reports until the Bureau of Economic Analysis data is available. However, it isn't uncommon for reported corporate earnings (paper profits) to differ substantially from the BEA's adjusted profits (production profits).
It is interesting that consumer sentiment lagged prior to the 2001 recession. However, despite the dot.com crash, that was a relatively mild recession and the slowdown in the economy probably wasn’t recognizable to most consumers until the recession nearly began. On the BaR, from three months prior to the recession to one month prior, consumer sentiment dropped from 24.3% to 4.3% above the baseline.
If we take Table 11 as it is, there were seven indicators that led two of the three recessions: real retail sales, existing home sales, building permits, nonfinancial corporate profits, ISM manufacturing, total vehicles sales, and consumer sentiment. Of these seven, five are measures of consumer behavior.
Focusing only on the six-month period previous to the last three recessions, based on my criteria, there was a distinct group of indicators that had neared the baseline of the BaR, foreshadowing recession. Prior to the 2007–2009 recession, there were seven of 18 plotted indicators (39%) that were either negative or below 10% on the vertical axis, or were above 10% of the baseline but had seen significant weakening. Prior to the 2001 recession, there were nine of 17 indicators (53%) that were within these categories, and for the 1990–1991 recession, seven of 10 indicators were signaling recession.
After compiling data across all three recessions for the six-month time period (Table 11), only the yield curve preceded all recessions. Nonfinancial corporate profits, consumer sentiment, and a mix of consumer-related indicators were early warning signals for two of the three recessions.
It is also worth noting that there were six indicators that were leading for only one recession (bottom six measures on Table 11).
Regarding the yield curve spread, nothing new was revealed here, but as pointed out, the inversion period for all three recessions occurred between 10 and 17 months before the recession started. The yield curve may signal a recession, but there is a fairly large window as to when it will occur.
This analysis suggests that other than the yield curve spread, keeping an eye on nonfinancial corporate profits, consumer sentiment, and a mix of consumer-related indicators should provide the earliest signals of an economic slowdown.
However, the analysis also demonstrates that there is some inconsistency among the indicators that signal a recession at the six-months-prior-to-recession point.
To me this indicates that it is better to keep an eye on all of the eggs in the basket, rather than just a few. Headlines, pundits, and click-bait generally focus on one or two eggs, all of which should be taken with a grain of salt. After all, there really is no guarantee that the yield curve will invert before the next recession (but it is likely to flatten).
My suggestion is to follow a comprehensive array of indicators, whether it is the BaR or something else. When a number of key measures begin to fall to recession levels, this is a clear warning, regardless of which indicators are included.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.