Normalized Unemployment Rates; Cyclical vs. Secular Forces
Many have heard of normalized P/E ratios based on 10-year earnings averages and the concept of reversion to the mean.
To the best of my knowledge no one has attempted to normalize unemployment rates based on demographic trends, although some have attempted unemployment normalizations based on changing definitions of the meaning of unemployment.
The following participation rate chart defines the problem.
Participation Rate by Age Group
click on any chart for sharper image
The above chart first appeared in my post Boomer Demographics and the Unemployment Rate. It is from a series by reader Tim Wallace. (click on link for rest of the series).
I added trendlines, a vertical black line, and circles to the chart.
Definitions and Notes
- The participation rate is the ratio of the civilian labor force to the total noninstitutionalized civilian population 16 years of age and over.
- The noninstitutionalized civilian population consists of civilians not in prison, mental facilities, wards of the state, etc.
- The labor force consists of those who have a job or are seeking a job, are at least 16 years old, are not serving in the military and are not institutionalized.
- There are strict requirements on what constitutes “seeking a job”. Reading want-ads or jobs on “Monster” does not count. One actually needs to apply for a job, go on an interview, or send in a resume.
- Please see Reader Question Regarding “Dropping Out of the Workforce” for an explanation of how the BLS determines someone is actively seeking a job.
The trends in purple (age 25-54), and green (age 20-24) up until 1988-1990 or so reflect the entry of women into the workforce.
Starting around 1988 there were abrupt changes in the participation rates of age groups 25-54, 20-24, and 55+. The shift in participation rate of 55+ was especially abrupt. Those shifts reflect baby boomer dynamics.
Structural vs. Cyclical Forces
A second major shift (marked by the black vertical line), was far more sudden, occurring exactly at the onset of the 2007 recession.
The question at hand is whether or not the trendlines at the onset of the recession are still valid. If so there is a pronounced impact on expected participation rates, and therefore unemployment rates (by implication).
It’s easy to confuse what’s cyclical vs. what’s secular because the direction of the 55+ trend did not change in 2007, only the slope changed. The direction did change for age group 55+ in or around 1992.
Participation Rate Discussion
I started work on these ideas much earlier this month with preceding posts as follows:
- December 07: Startling Look at Job Demographics by Age
- December 12: Questions of the Day: What if the BLS is Wrong About its Labor Force and Participation Rate Projections? How Would it Affect the Unemployment Rate?
- December 14: Boomer Demographics and the Unemployment Rate
In the meantime, others have been investigating cyclical vs. secular shifts as well.
For example, on December 13, Andrew Wilkinson, Chief Economic Strategist for Miller Tabak & Co. asked the question How Realistic is the FOMC’s Unemployment Forecast?
Wilkinson notes that if the participation rate rises a mere .5 percentage points employers would need to add 181,000 to prompt the Fed to take away the punchbowl by the end of 2015.
If the participation rate were to rise to 65.8 (the average reading since 2000), it would take a whopping 284,000 jobs per month. See my December 12 article for charts on the same lines of thought.
A table I posted in 2009 will show why 180,000+ jobs per month is not in the cards.
Monthly Job Growth 1999-2009
Chart courtesy of BLS. Annotations by me, numbers are in thousands.
The areas in deep blue mark recessions.
- At the height of the internet bubble with a nonsensical Y2K scare on top of that, the economy managed to gain 264,000 jobs a month.
- At the height of the housing bubble in 2005, the economy added 212,000 jobs a month.
- At the height of the commercial real estate bubble with massive store expansion, the economy added somewhere between 96,000 and 178,000 jobs per month depending on where you mark the peak.
Neither the housing boom, nor the commercial real estate boom is coming back. Nor is there going to be another internet revolution. If anything, outsourcing of jobs to Asia is likely to remain intense. Finally, consider all the financial engineering jobs, banking jobs etc, that are not coming back.
Returning to the present, Wilkinson states “As you can see there are plenty of moving pieces in projecting advances for the labor market. In plain and simple terms arriving at an unchanged unemployment reading through 2013 might require payroll gains of anywhere between 110,000
and 500,000 depending on how many people you want to involve in the labor market.”
Yes, I can clearly see that, noting that I lost a bet on it. Although I was certain the participation rate would drop, I never envisioned it would drop this far this fast (dragging the unemployment rate down with it).
For details, please see my October 7, 2012 post Lost Bet.
Where to From Here?
Wilkinson concludes with “In order for the desired unemployment rate to meet the falling projection laid down by the Fed, payroll growth is going to have to start picking up from its present pace. If it does not, the Fed is likely to be highlighting on an ongoing basis the lack of quality in the improving fortunes of the labor market.”
Assuming “lack of quality” is the same as “falling participation rate”, I am in agreement with Wilkinson.
Indeed, I will go one step further, suggesting the Fed is likely to miss badly even if payroll growth picks up (which I do not think will happen).
Will the Jobless Rate Drop Take a Break?
While pondering my thoughts regarding a bad miss by the Bernanke Fed on unemployment projections, please consider an economic research article by the San Francisco Fed, Will the Jobless Rate Drop Take a Break?
Looking at secular trends, participation grew rapidly from the mid-1960s through the 1990s as women entered the labor force in higher numbers. As women’s participation stabilized and baby boomers began to retire, labor force participation leveled off and in the 2000s began to decline. Young people spent more time in school during this period, further reducing participation rates. On the cyclical side, participation has tended to fall in recessions as workers withdrew from the labor force. This pattern was particularly pronounced during the recent financial crisis.
Trend vs. cycle: Is this time different?
Disentangling secular and cyclical fluctuations in labor force participation is never simple, but has been especially difficult lately. Figure 1 shows that labor force participation began to decline well before the Great Recession. This decline reflected, at least in part, the leading edge of the baby boom generation moving into retirement. The decline in labor force participation accelerated sharply during the recession and continued in the early years of the recovery.
The extent and timing of the recent drop in participation have been difficult to interpret conclusively. Many analysts agree that at least some of the recent drop has been cyclical. But they disagree on how much the decrease stems from the weak labor market during the recession and recovery. Estimates of the effect of these cyclical factors range from a third to half the total drop in participation (Davig, Maki, and Newland 2012; Aaronson, Davis, and Hu 2012; Van Zandweghe 2012). Some analysts argue that, when these workers return to the labor force, that will completely offset ongoing downward structural and demographic trends (Stehn 2012).
Will the unemployment rate stall in 2013?
Nearly 6.9 million people report being out of the labor force but wanting a job. As economic conditions improve, it is reasonable to expect that some of these workers will move back into the labor force or join for the first time. Based on historical averages, about 2.1 million of them could enter the job market. These potential entrants will either take jobs directly or join the labor force as unemployed workers actively searching for jobs.
The near-term path of unemployment will reflect both how quickly potential workers enter the labor force and the rate at which jobs are created. Assume that the average pace of job creation over the past two years continues. We can then project the path of the unemployment rate over the next year according to the rate at which the 2.1 million potential workers enter the labor force.
If these workers take a year and a half to join the labor force, which would be about a year faster than the entry rate from 1994 to 1999, the recent decline in the unemployment rate would stall at more than 8% by the end of next year. Suppose though that the number of workers who want a job but are not actively looking falls at a more moderate pace and it takes three-and-a-half years for this group to join the labor force. In that case, the unemployment rate would stay at 7.7% through the end of next year. For comparison, if none of the 2.1 million potential workers were to enter the labor market, the unemployment rate would fall to 7.4% by the end of 2013. Of course, the rate at which these workers join the labor force may reflect the labor market’s overall strength. A faster rate of job creation may offset a faster rate of labor force entry, allowing the unemployment rate to fall.
Wilkinson, the authors of the San Francisco Fed article above, and I are all having similar thoughts and concerns regarding the forces in play.
Cyclical Forces at Play
Please consider this chart from my December 7 article Startling Look at Job Demographics by Age.
Note that age groups 16-19 and 20-24 do not register at all in terms of importance.
We can guess whether or not the trend towards more school will continue or even reverse (increased schooling is what is driving the participation rate in those age groups), but what matters is the trend in age groups 55+ and even more so 25-54.
For convenience here is the chart I posted at the top of this now lengthy article.
Of the three green circles, the ones at the top and bottom are the ones of relevance.
Age 55 and Over Discussion
The current (since about 1992) trend in age group 55+ will not continue forever. At some point, perhaps age 75 or so, people will stop working whether they want to or not, simply because they will be physically unable. At what age that will happen is debatable, but it will happen at some point even though overall longevity is increasing.
Thus, some of that trend change since 2007 may be structural, but the rest is cyclical.
Based on the abrupt change at the start of the recession, I suspect most of the trend change since 2007 is indeed cyclical, even though some flattening of the trend should occur later based on physical ability as noted above.
The important point is a return to the trendline will increase the participation rate (even if at a flattening rate over time).
Age 25-54 Discussion
Unlike demographics for students (and the trend for more college education), I see no reason other than an extremely poor job market for the abrupt change in the 25-54 participation rate starting in 2007.
Thus, in this demographic, nearly all of the drop in participation rate is cyclical. And that drop is extremely important as the following table shows.
Population, Labor Force, Unemployment Rate, Participation Rate (Age Group 25-54)
|Year 25-54||Civilian Population||Labor Force||CP Employed||Unemployment Rate||PR|
The average participation rate for 2000-2007 is 83.3. Let’s normalize the labor force and unemployment rate for the average participation rate and check out the results.
Normalized Labor Force, Unemployment Rate(Age Group 25-54)
|Year 25-54||Civilian Population||Normalized LF||CP Employed||Normalized UR||Average PR (2000-2007)|
Reported vs. Normalized Numbers
- The current labor force for age 25-54 is 100,977,000
- The normalized (expected) labor force for age 25-54 is 103,498,000
- The current unemployment rate for age 25-54 is 6.39%
- The normalized unemployment rate for age 25-54 is 8.67%
- Current employment for age group 25-54 is 94,523,000
- Normalized (expected) employment for age group 25-54 at an unemployment rate of 6.39% would be 96,884,000
- The economy is currently short 2,361,000 jobs (96,884 – 94,523) in this single age group.
At present, the labor force is realistically understated by 2,521,000 minimum and the the economy is currently short 2,361,000 jobs, also minimum.
Making adjustments for this demographic group alone, the overall unemployment rate would be 9.2% instead of the reported 7.7%.
Moreover, and as noted above, I think reversion to the trendine in the second most important demographic (age 55+) is coming as well, except the 55+ trendline will flatten by natural causes (demographic as opposed to cyclical) over time.
Factor in a slight rise in participation rate for the 55+ demographic and the unemployment rate would be close to 10.
How Fast Can We Catch Up?
- On a immediate return to the participation rate trendline for age group 25-54 alone, the economy would need to pick up an immediate 2,361,000 jobs
- 2,361,000 jobs in a single year would be 197,000 jobs per month
- Stretching those jobs out for three years until the end of 2015, it would take an additional 65,000 jobs per month just to hold the unemployment rate flat
The key word above is “additional” and it ignores the possibility of a move back towards the trendline for age group 55 and over.
This “normalization” exercise (reversion to the mean demographic-adjusted trendline) shows how small upward changes in the participation rate in age group 25-54 alone will make it near-impossible for Bernanke to hit the Fed’s unemployment target by 2015.
The above projections do not even factor in the possibility (I think near-certainty) of a recession in the meantime.
Mike “Mish” Shedlock