Leading indicators, weekly earnings and hours worked, GDI and productivity are all falling. That’s bad for Aussie as well as U.S. equities.
In Should we be in a recession already? (2 June), Roger Montgomery wrote: “given the slumping credit demand and tightening lending standards in the U.S., which typically precede or accompany a recession, one wonders why we aren’t in a recession yet. We should be. What is curious is the obviously (sic) bearish sentiment failing to correlate with the performance of the equity market. Are equity investors unreasonably optimistic, or are the pessimists missing something?” He concludes: “perhaps the stock market has already got it right, and the pessimists need to recalibrate.”
That’s possible. Yet unlike Montgomery, who like virtually everybody else views recession almost exclusively through the lens of Gross Domestic Product (GDP), I’ve examined an array of measures. Moreover, I conduct my own analyses.
My analysis concludes that since late last year the U.S. has been in recession. If so, markets in Australia as well as the U.S. (ironically, and thanks to its “mega-caps,” on 8 June the S&P 500 commenced a new bull market) will likely sink as America’s GDP regresses towards its Gross Domestic Income (GDI). In that event, optimists will – once again – be caught flat-footed and incur hefty losses. Have you prepared or will you be surprised?
The Crux of the Bulls’ Case
“More than a year after the Federal Reserve began rapidly raising interest rates to tame inflation,” reported The Wall Street Journal (“Why the U.S. Is not in a Recession,” 4 June), “the hallmarks of a widely expected recession remain elusive. Employers are hiring aggressively, consumers are spending freely, the stock market is rebounding and the housing market appears to be stabilising … The most recent evidence (is) that the Fed’s efforts have yet to significantly weaken the economy. Instead, the lingering effects of the pandemic have left consumers and employers still playing catch-up. That momentum could prove self-sustaining.”
Payrolls grew by a very large and unexpectedly big number – 339,000 – in May, and increases during the preceding two months were higher than initially estimated, the Labor Department reported on 2 June. Above all, the rate of unemployment continues to scrape historical lows. “You don’t have a recession when you have 500,000 jobs and the lowest unemployment rate in more than 50 years,” Janet Yellen, the Treasury Secretary, told Reuters on 7 February.
A “Full-Employment Recession”?
Yellen’s assertion doesn’t convince some people. It also raises a crucial question: can a recession occur without big job losses and a sharp rise of the rate of unemployment?
Answers to this question differ (for a readable overview, see Mike Walden, “Will We Have the First Full Employment Recession?” CALS News, 26 August 2022). A small number of economists who believe that a recession is imminent also reckon that it’ll be much like those in the past: many jobs will be lost and the rate of unemployment will soar. Excluding the recession induced by the COVID-19 pandemic and panic, in recessions since 1980 the jobless rate rose an average of four percentage points. Today, a few economists reckon that the rate rate may reach 7% – double its current level – in a recession induced by the Federal Reserve’s goal of abating consumer price inflation’s rate of increase to 2% annually.
Most others disagree. What’s indisputable is that the number of job openings is presently twice the number of jobless people. That’s unprecedented – especially during what some people including me reckon is already a recession. Specifically, almost 11 million openings, equivalent to 7% of filled jobs, now exist.
That’s more than enough to cut all unfilled jobs rather than sack many employed people, and thereby to generate the job losses that accompany the average recession. Indeed, that’s what some economists believe will happen.
If so, Mike Walden estimates, in the months to come the jobless rate will increase by no more than one percentage point. As a result, a reduction of the output of goods and services – which is one definition of a recession – could occur at the same time that the rate of unemployment rises very modestly. Does that sound implausible? It’d hardly be the only unusual event created by the COVID-19 pandemic and panic: it’s changed the workforce, altered how we buy both products and services, and encouraged remote working and living. Walden hides his conclusions within questions: “is our definition of a recession the next post-COVID change? Could we experience a recession without massive unemployment?”
Such an event should remind us of the adage that change is the only constant. “The U.S. economy just hasn’t looked like its old self lately,” reported The Washington Post (“Jobless Recoveries Are Here to Stay, Economists Say, but It’s a Mystery Why,” 19 September 2013), “especially when it comes to regaining the jobs lost during a recession. It looks a lot more like 1980s-era Europe — slow to rebound and hire after a downturn, leaving workers to flail for years in a weak job market. The U.S. is stuck in its third consecutive ‘jobless recovery,’ stretching back to the rebound from the 1990 recession. And Americans might need to get used to them: a major new study … suggests that vintage Europe is the new American normal for recessions and recoveries.”
A decade ago, experts and their major studies were flatly wrong: the “jobless recovery” hasn’t become the norm. Quite the contrary: with the exception of the COVID-19 panic and its immediate aftermath, since 2016 America’s rate of unemployment has remained well below 5% (its mean since 1948 is 5.7%). The quality of the recovery from the pandemic is debatable; unquestionably, however, it’s been anything but jobless.
Accordingly, it’s hardly a stretch to suppose that experts are again wrong, that is, to envisage the “full-employment recession” that many – including Janet Yellen – have failed to foresee and will struggle to explain.
“I don’t think there’s any chance we’re in a recession,” Justin Wolfers, a professor of public policy and economics at the University of Michigan, confidently informed The Wall Street Journal (4 June). The National Bureau of Economic Research (NBER), an academic research group and the official arbiter of U.S. recessions, analyses a slew of data to help determine whether the economy is in recession. “Most of those indicators look healthy,” Wolfers asserted.
Seriously? The NBER doesn’t use a constant set of measures; nor does it quantify and publish them. But the Organisation for Economic Co-operation and Development (OECD) does. Figure 1 plots its monthly Composite Leading Indicator (CLI) for the U.S. since January 1960. CLI incorporates measures of business outlook, consumer confidence and spending, construction activity, hours worked, orders booked and shipped, etc. According to the OECD, “CLI is designed to provide early signals of turning points in business cycles showing fluctuation of economic activity around its long term potential level.” Observations above 100 denote current conditions above the economy’s long-term potential; observations below 100 indicate conditions below its potential.
Figure 1: Composite Leading Indicator, U.S. Economy, Monthly Observations, January 1960-April 2023
Since January 1960, the U.S.’s CLI has averaged 99.95; its all-time high (103.4) occurred in December 1972 and January 1973; and its major lows (93-96) occurred towards the troughs of recessions in 1974-1975, 1981-1982, 1991, 2008-2009 and 2020. Its most recent reading (April 2023) is 98.6. Memo #1 to Justin Wolfers: that doesn’t look very healthy.
Figure 2 plots CLI’s 12-month percentage changes. On average during any 12-month period, CLI changes exactly 0.0%. The biggest falls have occurred towards the troughs of recessions and its biggest increases during recoveries from recessions. During each rolling 12-month period from July 2022 to April 2023 (that is, July 2021-July 2022, August 2021-August 2022, …, April 2022-April 2023), CLI has fallen between -1.5% and -2.0%. Memo #2 to Wolfers: that doesn’t look healthy, either.
Figure 2: Composite Leading Indicator, U.S. Economy, 12-Month Percentage Change, January 1961-April 2023
The Slump of Manufacturing
According to the Reserve Bank of Australia, “while there is no single definition of recession, it is generally agreed that a recession occurs when there is a period of reduced output and a significant increase in the unemployment rate. Views differ about how to best identify this.” Using data collected by the U.S. Census Bureau’s monthly “M3” survey and collated by the St Louis branch of the Federal Reserve System, Figure 2 plots the 12-month percentage change of the CPI-adjusted dollar value of U.S. manufacturers’ shipments, inventories and new orders.
During the average 12-month period since February 1992, the CPI-adjusted value of manufacturing output increased 0.8%. During 12-month periods from October 2000 to July 2002, however, it fell (by as much as 16% in the year to June2001). Output also decreased during the 12-month periods from August 2008 to December 2009 – and by as much as 29.5% in the year to April 2009.
Manufacturers’ output can fall for extended periods and by considerable amounts without a recession. Most notably, during 12-month periods from September 2014 until September 2016 (that is, from September 2013-September 2014, October 2013-October 2014, …, to September 2015-September 2016), output sagged by as much as 18% in the 12 months to July 2015. With this important exception, over the past 30 years a decrease of manufacturing output has coincided with a recession of the overall economy.
Figure 2: U.S. Manufacturer’s Shipments, Inventories, and Orders, CPI-Adjusted, 12-Month Percentage Change, February 1992-April 2023
That’s why the deceleration of growth from its all-time high of 33% in the 12 months to April 2021, and the shrinkage of output since November of last year, are highly relevant. During the year to April 2023, it’s sagged 4.5%.
Americans Are Working Fewer Hours per Week …
According to the U.S. Bureau of Labor Statistics (BLS), “average weekly hours relate to the average hours per worker for which pay was received and is different from standard or scheduled hours. Factors such as unpaid absenteeism, labor turnover, part-time work, and stoppages cause average weekly hours to be lower than scheduled hours of work for an establishment … Average weekly hours are the total weekly hours divided by the employees paid for those hours.”
Hours worked exhibit a high degree of seasonality: they vary according to public and school holidays, etc. BLS adjusts its data for these factors; it also converts weekly hours measures into hours worked across the whole month. According to the Australian Bureau of Statistics, “early labour market impacts from major disruption to the economy tend to be most evident in the hours people work. Hours worked can change more quickly than employment, given the variability in individual circumstances for people and businesses from one week to the next. During an economic downturn, reducing hours is often an early response taken by businesses, often with the view to avoiding people losing their jobs.”
Using monthly data compiled by the BLS since March 2006, Figure 3 plots private sector employees’ average hours worked per week (not necessarily at one job). Over the past 17 years, the average is 34.4. Before the GFC, the average was 34.35; during the crisis, it plunged below 34, but by 2011 rebounded to the pre-GFC and overall means. During the next eight years, the average workweek fluctuated without trend and mostly within the range 34.3-34.5 hours. During the COVID-19 panic (March 2020) the average fell to 34.1 hours – but by January 2021 reached its all-time high of 35.0 hours.
Figure 3: Average Weekly Hours, U.S. Private-Sector Workers, Monthly Observations, March 2006-May 2023
Since then, however, the average workweek has shortened considerably – to 34.3 hours in May 2023. Albeit from a very high base, the decrease since January 2021 is by far the longest (almost 30 months) and the most marked (total fall of 0.7 days) since the GFC (12 months and 0.7 days).
… and Earning Fewer Dollars per Week
“When the economy is in a recession,” says Forbes (“What Happens during a Recession?” 9 August 2022), “incomes stagnate or drop due to employers slashing hours or reducing their workforce.” Also using monthly data collated by the BLS, Figure 5 plots private sector workers’ average CPI-adjusted weekly earnings. They’ve risen from $1,041 in March 2006 to $1,147 in May 2023; that’s a compound annual growth rate (CAGR) of 0.57% per year.
Earnings fell marginally during the GFC but rose very steadily until just before the COVID-19 panic. They then jumped to their all-time high of $1,214 in January 2021. Since then, however, they’ve slumped 5.5% to their current level ($1,147 in May 2023).
Figure 5: Average CPI-Adjusted Weekly Earnings, Private Sector Employees, Monthly Observations, March 2006-May 2023
Figure 6 plots the data in Figure 5 as rolling 12-month percentage changes. Between March 2006 and March 2007, for example, average CPI-adjusted earnings increased from $1,041 to $1,053 (1.2%). Average weekly earnings fell 2.3% at the nadir of the GFC, but then recovered and fluctuated without trend (and mostly within the range -1.5% to + 3%. They then increased by 7.5% in the 12 months to May 2020.
Figure 6: Average CPI-Adjusted Weekly Earnings, Private Sector Employees, Rolling 12-Month Percentage Changes, March 2007-May 2023
Since the year to April 2021, however, average weekly earnings have plunged as much as 3.9% in the 12 months to June 2022. That’s the largest percentage decrease on record and twice as much as they fell during the GFC. More recently, their fall has abated (and is presently -1.1% in the year to May 2023. Accordingly, for almost two years Americans’ CPI-adjusted average weekly earnings have decreased.
The good news is that in aggregate more Americans than ever are working: since March 2006 the population of the U.S. has risen 12.4% (from approximately 298 million in March 2006 to 335 million today). Its total civilian workforce has also grown, albeit more slowly (by 10.6%, from 150.8 million in March 2006 to 166.8 million today).
The bad news is that, on average over the past couple of years, Americans have been working fewer hours per week; moreover, a combination of falling wages and rising consumer price inflation has slashed their weekly take-home pay almost 6%. If that’s not a recession, many Americans would nonetheless agree that it feels like one.
GDI versus GDP
It’s a commonplace: one person’s spending is another’s income; aggregated to the national level, it’s a tenet of macroeconomics. That, in essence, encapsulates the theoretical distinction between and the empirical equivalence of Gross Domestic Product (GDP) and Gross Domestic income (GDI). GDI measures the economy’s income (namely pre-tax wages, salaries, etc., and profits); GDP measures the value of the goods, services and technology produced. In essence:
GDI = Wages and Salaries + Profits + Interest and Rental Income + statistical adjustments;
GDP = personal Consumption + business Investment + Government spending + net Exports + statistical adjustments.
In theory, GDI = GDP. In reality, they usually track one another closely, but occasionally deviate to varying extents.
Why? According to the Bureau of Economic Analysis (BEA) of the U.S. Department of Commerce, the market value of goods and services consumed often differs from the amount of income earned to produce them due to sampling errors, coverage differences, and timing differences (for a readable overview, see the BEA’s factsheet “Why Do Gross Domestic Product and Gross Domestic Income Differ, and What Does That Imply?” 26 July 2018).
Despite their conceptual (and usually empirical) equivalence, at crucial points in time GDI can be a more accurate gauge of the economy’s health. At these junctures, initial estimates of GDI are generally closer to the final estimates of both calculations. Initial estimates of GDP, in contrast, are occasionally subject to subsequent significant – indeed, substantial – revision.
Most notably, Jeremy Nalewaik, an economist at the Federal Reserve, showed that GDI detected the unfolding GFC much sooner and more accurately than GDP. Estimates of GDI began to signal a sharp slowdown in the middle of 2007 – even as GDP kept climbing. Subsequent measures of GDI also better reflected the severity of the downturn (see in particular, “The Case for GDI: a Q&A with Jeremy Nalewaik,” The Financial Times, 27 September 2011, as well as its references).
“The point,” says FT, “is that at a critical moment in the recession … there was a canyon-sized gap between economic reality (which GDI captured) and the measure (GDP) that is generally accepted by markets, economists, and policymakers …”
The present may be one such critical moment. Using quarterly BEA data compiled by the St Louis branch of the Federal Reserve System, Figure 7 plots 12-month percentage changes of GDI since January 1990. On average over this interval, GDI has grown 2.0% per 12-month period. But on five occasions it has decreased at least two quarters in succession:
- during four quarters in the early-1990s (October 1990-July 1991) it fell at an average annualised rate of 1.75%;
- during two successive quarters early this century (July and October 2001) it fell at an average annualised rate of 0.85%;
- during seven consecutive quarters (beginning in July 2008 and ending in September 2009) it fell at an average annualised rate of 3.03% (and by as much as -5.5% in the 12 months to December 2008);
- during two consecutive quarters (April-June and July-September 2020) it fell at an average annualised rate of 5.4% (and by as much as 8.4% in April-June 2020).
Figure 7: U.S. Gross Domestic Income, CPI-Adjusted 12-month Percentage Changes, January 1990-January 2023
GDP and GDI Regress Towards One Another
The fifth occasion, the quarters since April-June 2021, is thus of great interest. During the 12 months to April 2021, GDI rose at an annualised rate of 13.0%. During each quarter thereafter, its 12-month rate of growth has decelerated, and during the two most recent ones (October-December 2022 and January-March 2023) GDI has decreased at an average annualised rate of 1.85%.
As measured by GDI, in other words, and using the same rule of thumb as applied to GDP, since October of last year the U.S. has been in recession. Gauged by GDI’s rate of decrease, the present recession is harsher than the Dot Com recession of the early-2000s, less severe than the GFC and COVID-19 panic, and on a par with the recession of the early-1990s.
It’s possible, of course, that the two most recent quarters’ estimates of GDI are overstated, and that these negative readings will be revised into positive ones – and the present “GDI recession” will thereby disappear.
Equally, however, it’s possible that the most recent estimates of GDP overstate the U.S. economy’s strength, that is, that over the next few months these weakly positive readings will become negative ones – and a “GDP recession” will belatedly appear.
Figure 8, which plots 12-month percentage changes of GDI and GDP, and Figure 9, which plots these two series’ difference, suggest that whichever revision occurs, it’ll be a big one. The two series in Figure 8 are strongly correlated (r = 0.93): that is, the higher the one lifts, the higher the other tends to rise; and the lower the one falls, the lower the other tends to sag.
Figure 8: CPI-Adjusted GDI and GDP, 12-month Percentage Changes, January 1990-January 2023