This is the first of several articles on how to interpret today’s macroeconomic environment.


I began tracking employment data beyond the headlines a decade ago for my teaching. At the time, I expected to find a lot of early retirements, under the impact of the Great Recession. To my surprise, that is not what the data showed: the “hit” to employment fell most heavily on workers under age 30. Of course, that was a sobering message for my students, and the inability of young workers to launch their careers, or even gain job experience, has been bad for the long-run health of the US economy.

So, where are we today? I look at two sets of indicators. Both convey the same message: employment is as good as ever and is unlikely to improve further. First, the data, then the implications.

Part I: Demographics-adjusted Total Jobs

One danger of looking at the “headline” employment numbers relative to previous eras is that they are an apples-to-oranges comparison. That’s because the age structure of the US population shifted dramatically over the past 2 decades. On one end, the baby boomers began retiring and have now largely left the labor force. The cohort behind them is much smaller. On the other end, birth rates have been lower for several decades, so there are fewer young people entering the labor force as a share of the population.

By my calculations, using Census data, in 2007, the working age population was growing at about 105,000 jobs a month. (This is already well below the rule of thumb SA readers may have learned in their Principles of Economics courses some decades back of 150,000 – I know because, to my embarrassment, I long used a much higher textbook rule-of-thumb in my teaching.)

What of today? – in my spreadsheet, the high-end estimate is that to tread water, the US economy only needs to add 65,000 jobs each month.

This past month thus represented a net addition of 70,000 jobs relative to that base requirement. Such net additions began on a consistent basis in August 2011. In the interim, we have regained the ground lost in the Great Recession, as per the following:

Graph 1: Employment (red line) and jobs lost/added (red and green bars) versus estimated normal employment levels corrected for changing demographics

Source: Author using Bureau of Labor Statistics and Population Census data. See an October 2016 Autos and Economics blog post, “Methodology”, for details.

Now, no single number should be used naively, both because the underlying data are noisy and because this estimate still overlooks lots of other changes (such as changes in the gendered nature of employment, and regional and sectoral shifts). One minor variation is to adjust for workers on involuntary short-hours. That number rose dramatically during the Great Recession, as employers tried to retain their core workers, putting them on short hours rather than firing them. That trend has now been more than reversed, so correcting for short hours shows an even larger improvement in the labor market:

Graph 2: Employment (red line) and jobs lost/added (red and green bars) versus estimated normal employment levels corrected for changing demographics and for the level of involuntary part-time employment

Source: Author using Bureau of Labor Statistics and Population Census data

Part II: Participation Rates

Details and Assumptions

Now, the above exercise tries to sum up the labor market with a single number. Generating it requires multiple assumptions, using (my own) projections of population and (my own) estimates of what was normal prior to the Great Recession. To check for robustness, I thus put together a second set of indicators, using the age-specific participation rates calculated by the Bureau of Labor Statistics from their monthly employment survey. Participation is not employment: being in the labor force includes those working part-time and full-time, and those looking for work. The latter is the group included in the “headline” unemployment number. And the size of each “cell” (age x participation x month) is smaller than that behind the aggregates that make the headlines, so the monthly numbers jump around.

First, the raw numbers show heterogeneity across the labor force. Since consistent data collection commenced in 1994, we see a sharp increase in participation by older Americans, particularly those aged 60 and above. Second, we see a sharp decline in the share of younger workers in the labor force. High schoolers apparently don’t get a part-time job when they turn 16 (as I and many of my friends did a half-century ago). More young people are in full-time education. (See the Atlanta Fed Labor Force Participation Dynamics page for details.) Aggregate numbers hide such trends.

Graph 3: Age-specific participation relative to the relatively steady level exhibited by prime-age workers during 2000-2006.

Source: Author using Bureau of Labor Statistics data

Graph 4: Shifts in participation at young and old ends of the US population

Source: Author using Bureau of Labor Statistics data

Now, the participation levels are smaller at the old and young end of the age spectrum. For example, in September 2019, only 19.7% of those age 70-74 participated or 1 in 5. But that’s up from 10.7% when age-specific data reporting commenced in January 1994. At the other end, 309% of those age 16-19 participated, down from 48.7% from January 1948 when data reporting began for that bracket. In contrast, there’s little shift in the age 55-59 bracket, despite my expectation that I’d observe a big drop during 2007-2011. Indeed, the share of those working age 65 and above rose steadily, before, during and after the recession.

Graph 5: Prime age participation since the start of the Great Recession in 2007

Source: Author using Bureau of Labor Statistics and Population Census data


I thus focus on prime-age workers, those age 20-59. First, observe the volatility at the monthly level in all the data. Hence, it makes more sense to focus on a moving average, the 2 black lines. Participation by younger workers is still below the pre-recession level, but some of that reflects a rise in those in community colleges and other active schooling/training. In any case, the absolute level is lower, at 67.9%.

Prime-age workers – those age 25-54 – show much more uniform behavior. First, the absolute levels are similar and much higher than those of older and younger cohorts, around 80% (the moving average was 80.5% in September 2019). Second, participation levels moved closer to each other. Finally, and most importantly, the labor force participation of prime-age workers has recovered, indeed now slightly exceeds, the level prevailing prior to the onset of the Great Recession in early 2007. There is still some room for growth, but over a 12-month period, participation rates have never reached as much as 83%.

Now, these are still aggregate numbers. In the background are declining male participation rates and rising female rates at younger ages.

Graph 6: Reasons for Prime-age Male Non-participation

Source: Atlanta Fed, Labor Force Participation Dynamics

Graph 7: Gaps between Female and Male Participation

Source: Author using Bureau of Labor Statistics data

One more nuance: female labor force participation has converged on that of men at lower ages – the blue and red lines in the graph – but remains much lower at higher ages. Why not full convergence? – the US provides very little in the way of support for families, from day care to parental leave. In Canada, it’s possible for women to have children and continue working, and work they do. In the US, it’s very hard to both work and to care for young children; in general, hourly day-care rates are higher than the minimum wage. We could have a larger labor force, but not without major changes in policy that the Administration and the Republican Party oppose. Even if the political landscape shifts, the impact would lie a decade in the future, beyond the time horizon relevant for most investment decisions.


1. The US labor force has very little room to expand. We should not expect large jobs numbers. Today, a “disappointing” number would be 65,000 jobs added. Investors should not be taken in by gloom-and-doom analysis rooted in the structure of the economy that prevailed in the 1990s!

2. There are other margins of adjustment – wages, hours worked. But outside of manufacturing, hours worked are higher than in 2007, so my judgement is that future growth must entail better wages. Pay continues to lag but is now rising; for that see the Atlanta Fed Wage Tracker.

3. That means investors should pare their expectations of future cuts in interest rates by the Federal Reserve.

3. Labor is key to the supply side, so slow growth there carries over to slow GDP growth. Slow growth will be the norm. Don’t panic at a low number!

4. That includes manufacturing, which has declined as a source of employment monotonically since the end of WWII, from 1 in 3 jobs (32+%) to 1 in 12 jobs (8.5%). Manufacturing is a tail that can no longer wag the dog.

Sum: We no longer live in the 1990s and need to be careful to update our rules of thumb for what is good and bad news.

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.

Source link