The non-mystery of declining employment rates for prime-age workers

September 7th, 2016 at 6:15 pm

Allow me to pile on alongside Dean Baker on this alleged mystery-of-the-missing-men, the theme of an NPR story that Dean critiques. There’s little mystery here: the driving factor is the loss of employment opportunities, or what economists call weak labor demand, particularly for non-college-educated, prime-age (25-54) workers. It’s not the only factor, but it’s the main one. There are various ways to show this, but the one I find most intuitive, as it’s kinda staring us in the face, is the cyclical pattern of the employment rates of prime-aged men, which I’ll get to below.

A caveat: this being economics, I want to be careful not to over-claim. There are, of course, other factors in play and no one has smoking-gun evidence that demand explains everything. But nothing ever explains everything in economics, and you should be suspicious of those who claim otherwise.

That said, I recently wrote something on this issue, out soon, critiquing new work by Nick Eberstadt, who’s cited in the NPR piece and has a book on this coming out soon (summarized here). Nick was kind enough to include a response by yours truly in the book. His evidence of the long-term decline in men’s employment rates, by the way, is far-reaching and well-constructed. But we disagree on the diagnosis.

This is a problem, because when you downplay the straightforward demand-side story, you end up emphasizing supply-side stories about disability insurance, the safety net, and other less important explanations which lead you to prescriptions that are ineffective at best and harmful at worst.

For those of us who’ve tracked this phenomenon for a while, the recent report by the Council of Economic Advisors is among the most thorough analyses out there. The CEA finds that “reductions in labor supply—in other words, prime-age men choosing not to work for a given set of labor market conditions—explain relatively little of the long-run trend…In contrast, reductions in the demand for labor, especially for lower-skilled men, appear to be an important component of the decline in prime-age male labor force participation.” (My bold.)

It is common, for example, to cite the generosity of the Disability Insurance program as a large contributor to the long-term decline in men’s work. But CEA shows that the magnitude of the increase of prime-age men on DI is too small to explain the lion’s share of the decline in work. Over a period where their participation rates fell 7.5 points (1967-2014), CEA finds disability rates go up from 1 to 3 percentage points. Of course, it would be extreme to conclude that none of that increase was warranted by actual disabilities (or conversely, that all of it was warranted). CEA’s analysis assigns less than half a percentage point (out of the 7.5 points just noted) to DI, suggesting it accounts for less than 10 percent of the decline in work.

Citing other safety net programs gets you no further, because they’ve become more, not less, conditioned on work.

But to me, just eyeballing the data returns a solid bit of demand-side evidence (“EPOP” in the chart’s title means employment-to-population ratio, or the employment rate). The figure below shows male employment rates with shaded recession bars and a trend line running through the series. The negative trend is obvious and suggests the depth of the problem. But the cyclicality around the trend is equally clear. Simply put, prime-age guys get whacked by recessions (and the less educated they are, the harder they get hit); then, as demand strengthens, they slowly start to climb back.

Source: BLS, linear trend

Source: BLS, linear trend

Most recently, these guys have climbed 2/3’s of the way back from their losses in the Great Recession. In recent downturns, they’ve made in most of the way back: in 1979, their EPOP was 91 percent; in 1989, 90 percent; in 2007, 88 percent. None of this denies the negative trend which is real and serious. But this series is a ratchet, not a straight line.

The reason that’s important is because it suggests they’re not lost to the labor market: these guys still respond to demand shocks in both directions. In other work, I’ve shown that less educated prime-age men are more responsive to stronger demand—their employment rates are more cyclical—than those with higher levels of education (see table 3 here). The coefficient on the demand variable for prime-age men with terminal high school degrees in this work was 1.12 (t-stat: 22.28). For those with advanced college degrees, the coefficient was 0.23 (t-stat: 5.37).

Here’s another figure that might surprise you. To estimate the extent of cyclicality in the employment rate series you see above, I regressed that series on a flexible trend and, to capture cyclical variation, the unemployment gap, or u – u*, where u is the unemployment rate and u* is CBO’s estimate of the natural rate. When u – u* is positive, there’s slack in the job market; when it’s negative, the job market is tight. Thus, we expect a negative coefficient (tighter job market, higher EPOP), which is what we get.

Source: See text.

Source: See text.

But if you start the regression with data from the late 1940s to the 1960s, add one observation at a time and rerun it (a rolling regression), you get the plot you see below. The fact that the coefficient falls as more observations are added (and also becomes more statistically significant) suggests an increasing cyclical response.

Here again, I don’t want to overplay the point—the coefficient on the unemployment gap grows by about 25 percent over the series. But if you read much of the commentary on “missing men,” you’d be quickly convinced that these guys are increasingly cut off from the job market, disconnected from work, and immune to stronger labor demand.

Final point: I fear some guys are. That is, shut out and disconnected. These are the guys with criminal records, who face, as Eberstadt discusses, extremely steep barriers getting back into the job market, and there are millions of them. They need targeted help finding their way back to work, alongside criminal justice reform that stops excessively punishing them for non-violent crimes.

Bottom line, no mystery, just the need for a lot more labor demand. And given the persistence of this trend and the extent to which the economy leaves these guys behind, I’m not at all sure we can count on the market to do this for us. At least for some of these guys, and as Dean says, some women too, we may well need to think in terms of direct job creation.

Print Friendly, PDF & Email

8 comments in reply to "The non-mystery of declining employment rates for prime-age workers"

  1. Smith says:

    I’m in agreement there is a problem as I’ve cited even on this blog repeatedly, most recently on August 17, 2016
    “Four years ago labor force participation still a full percentage point higher, representing 2.5 million more workers. Only half the loss can be attributed to demographics (aging workforce). Where’d the other 1.25 million go since 2012?”
    Things missing in Baker and Bernstein’s short take:
    Mostly it is not new programs that are needed, but rather a restoration of conditions that existed before the conservative and neo-liberal war on labor triumphed (a long war of attrition I might add). If more money was available to working men and women, in higher wages, they would spend it creating a virtuous cycle. Not only would more money seed a healthier growing economy, but higher wages would lessen dependence on unsecured debt. Unsecured debt (credit, student loans, even auto loans since vehicle depreciation far outstrips loan value) drains money to reward the financial sector and the haves. Slack is expected when overall government spending, especially per capita, is below previous levels, let alone previous government aided recovery levels. Krugman occasionally graphs this fighting austerity, but always on defense. Companies sit on hoards of cash due to lack of demand. Tax it away if not spent.
    This problem (less workers) appears to be a no brainer, less workers because of less jobs. The puzzle is why no eventual recovery even in the absence of Keynesian stimulus? Why the lost decade?
    I would propose that this blog and others may be getting an important point backwards. While obviously low unemployment boosts wages, it could be more emphasis actually needs to be placed on boosting wages. That natural experiment did occur twice recently, in the payroll tax holiday, and the drop in oil prices. Furthermore, unemployment was low in the 2000s before the recession, but wages went nowhere. The trend in higher minimums recognizes this.
    When Yellen says we’re nearing full employment and it’s time to raise interest rates to slow the economy, you are doomed before you start. Again, time to revolutionize how we fight inflation, stop using bludgeon of unemployment. Prevent higher prices with measures to combat higher prices and curtail corporate profits.
    Other ways wages are throttled:
    Women in workforce without access to free quality childcare and paid maternity leave.
    Exploited immigrant workers (all employer sponsored H visas) who have zero worker rights.
    No penalty for foreign profits parked overseas
    The glut of college educated workers who are forced to take low skills positions, creating much greater unemployment among the actual less educated (stuff rolls downhill).
    Replacement of older workers with higher salaries by younger workers with lower salaries.
    Exempt status that means OT without compensation (Obama did not even boost wage threshold to previous level, why?)
    The usurious interest rates that were legally prohibited before 1978 by state law, and high student loan debt caused by so called non-profits, also amounts to a tax on wages, effectively lowering rates for those who can least afford them.
    Yes unemployment depresses wages, the surplus army of labor is not a new idea, why is it only capitalist seem to grasp the benefits derived? But rising wages (which also propels productivity increases) may be whether chicken or egg like needing greater priority for a recovery.

    (An aside: Direct job creation is not indirect job creation, the New Deal was about direct job creation, a good idea, wage subsidies are corporate welfare that lowers everyone’s wage, and disadvantages non participant businesses, just as EITC is and does, and just $2.50 increase in minimum replaces dollars of EITC entirely, which seemed a pipe dream when I first wrote about on this blog, but now a reality being phased in, in NY and Ca., but not government funded or needing tax professionals to collect. I’d favor programs targeting hard to employ, but not in the form of subsidies)

  2. Denis Drew says:

    The forever missed-low skilled dropout factor — is low pay for low skilled jobs brought on by the uniquely American (not Canadian or European) collective bargaining dropout. 100,000 out of my guesstimate 200,000 Chicago gang-age, minority males are in drug dealing street gangs (and other nefarious lines of “work”) because jobs that could pay $800 a week pay $400 (1968 fed min was $440 — at half today’s per capita income).

    My “gang” also dropped out after our pay dropped from $800 a week to $400 — American raised taxi drivers. Let’s dwell on the latter for a moment. Between 1981 and 1997 Chicago allowed one 30 cent raise in the mileage rate — at which 1990 midpoint the city started adding subways to both airports, unlimited limos, free trolleys between all the downtown hot spots (gone but replaced by Uber) — and — 40% more cabs. Last I looked the meter was 50 cents below 1981.

    Ditto for Walmart killing good, middle class supermarket contract. Neither would happen in a labor run country. Walmart faced centralized bargaining (everybody doing the same job in the same locale, e.g., retail clerk, working under one common contract) in Germany and had to move back out.

    The Wage That Meant Middle Class, by Louis Uchitella, April 20, 2008

    Nothing that normal (Canadian, European) bargaining power couldn’t cure overnight — works everywhere bargaining power exists.

    New idea: NLRB finding of firing organizer can be answered with mandatory cert election. The core problem is not firing the worker, but muscling the labor market.

    • George Lenny says:

      There are no ‘workers’, everyone is a temporarily embarrassed capitalist on the verge of striking it rich, so let’s make sure when that day comes taxes on the rich are low…

  3. Rima Regas says:

    There is also age and race-based discrimination. The latter has been a problem since the start of the recovery, with (highly) educated males and females middle age on up having the toughest time getting back into the labor force in their former occupation, in anything that resembles their last pay scale. Some of your colleagues used to who used to refer to them as the “lost generation,” stopped reporting on them entirely two years ago. Ageism is a serious problem. Stats I’ve seen at EPI and elsewhere on minority college graduates and labor force participation also show that they are being hired at a lower rate.

    I would love to see a piece from a US economist that deals with US labor force participation and what British economist, Guy Standing of the London School of Economics calls the “precariat,” in the context of what Larry Summers has been saying is our new normal, secular stagnation.

    There are an awful lot of highly skilled people out there in their late 40s-mid 50’s whom we cannot – should not – give up on or expect to finish out the remaining work years of their lives surviving on the gig economy.

  4. Rex Nutting says:

    Good post, Jared.

    I’d like to see more research on the links between the increased incarceration and arrest rates since the 1980s and the decline of labor market participation. More than 70 million adults have been arrested or convicted, and for many of them that fact dooms them to underemployment or unemployment for the rest of their lives. (even an arrest record can reduce employment prospects).

    I wrote about this issue here:

  5. Longtooth says:

    The long term downtrend trend in EPOP for prime age males (and for the composite) has been given many different reasons, and there are undoubtedly many of them. Among them is US’s trade policies which have incentivized domestic goods producers to seek lower wage offshore labor sources.. This in turn reduced labor union bargaining power for most goods producing industries (along with gov’t approvals thereof beginning strongly in the early 1980’s).

    The long term trend of union employment participation has been in linear decline since the 1950’s (see this graph link

    just as is the same long term linear trend in EPOP since the 1950’s. Both trends support increasing shifts of higher priced union labor and thus also to goods producing labor in general to lower wage regions and thus reduction in demand for domestic labor in goods production. This places a surplus of domestic labor available for employment on the services sector which, due to the added supply in that sector, reduces real wages in services (nominal wages only increase at very near the rate of inflation, if that)..

    With a glut of services labor supply then services labor demand shifts to lower skilled (high-school or less) who are then competing with higher skilled labor from the laid off goods producers for jobs at McDonalds, Walmart, retail clerks, hospitality services (read restaurants), etc. This forces the lower skilled (high school or less) increasingly out of the job market, including mothers deciding to stay at home instead of working (child care costs eat up too much of their meager low skill jobs to make real net income worthwhile at all).

    Increasing use of automation in both goods and services sectors by computing methods since the late 1980’s and 1990s simply exacerbates this.

    If there’s a net greater return on capital ROI by shifting labor to lower wage rate regions, then that’s where capital will be used for profit. US tax policies have incentivized increasing shifts of US capital use to offshore facilities for goods producers. This was known to be an increasingly occurring condition when I took Econ 101 in collage in the late 1960’s, one of the inevitabilities of both tax policies and unrestricted capital flow, including the inevitable effects on US shifting to a services based economy. Interestingly my then Econ prof told the class and showed analytically why this was then occurring and would continue — it wasn’t that labor unions could prevent it … at best they could reduce the rate of occurrence, or better yet, effect gov’t tax policies to dis-incentivize domestic producer capital shifts to offshore production. He said however that it was unlikely that labor unions could politically change those tax policies — “money talks, everybody else walks” was his reason. That was in 1967 or 1968.

    So the longer term reductions in EPOP’s trend rate were as predictable in 1960’s as they are now and for most of the same reasons now as then. The only major changes since 1960’s have been China’s trade policies and an acceleration of automated methods due to the dramatic price reductions in compute power as rates of performance improvements went ballistic…. which in the late 1960’s was only beginning with the semi-conductor advances (I wen to college in San Jose… was there as it happened and in the high tech industry when I graduated).

    The major difference to Germany (often cited) is that labor and labor unions were constitutionally given power at the table — and I worked in Germany in the 1980’s for several years at a high tech company and saw the affects of the legal entity of “betriebsrat” directly and in person as more and more automation took affect there at that time..

  6. Alan Blair says:

    You note that it is the decline in demand for Male (especially unskilled Labor) that is source of declining labor participation. Yet during the same time we allowed over 1,000,000 legal immigrants to enter the country each year and also now have some 12M – 20M illegal immigrants. The demand appears to be there, it has just be satisfied by the legal and illegal immigrant populations. A possible solution is to lower the number of immigrants allowed to the US.

  7. Alan Blair says:

    The other issue is there was also a dramatic rise in female labor during this time period. In principle (I think), the males could have done these jobs. Manufacturing and outdoor work is being automated, but why did not the males just move into office work and nursing, etc.