Full Employment: Is Labor-Saving Technology Making It Harder to Get There?

May 10th, 2013 at 7:03 pm

Got lots of feedback from this oped in the NYT on the need to consider full employment policies:

While the high jobless numbers are partly a legacy of the Great Recession, the fact is that our economy has generated too few jobs for most of the last 30 years and is likely to continue to do so. The only viable response is a return to an idea that once animated domestic policy making: full employment, the notion that everyone who wants to work should be able to find a job, and if the market isn’t up to the task, then the government must fill the gap.

Since I consider this issue to be so critical to the economic well-being of most American households, I’m going to start a new series that presents Q&A’s on some aspect of the problem.  Here’s the first entry.

Q: I know about the gap between median income and productivity—that’s inequality—but you allude to a gap between employment and productivity.  What’s behind that?  Automation?

A: The figure below tells the story. It plots productivity growth against employment (using a full-time equivalent measure of employment to control for changes in the number of part-time workers) indexed to 100 in 1959.


Sources: BEA, BLS; productivity is for nonfarm business sector; employment is for full-time equivalents in the private sector.

As the question notes, for decades analysts have tracked the divergence between median incomes or wages and productivity that began in the later 1970s, an important symptom of growing inequality.  That trend is by now well-known and considerably researched.  This trend—and the clear divergence around the beginning of the last decade—is less understood.  But it is just as important.

While “Luddites” and their descendants have worried for centuries that advances in productivity would render workers unnecessary, leaving them stuck in technological unemployment, economists generally recognized that the intervening variable of demand would absorb productivity growth.  In this way, productivity growth, while often disruptive, provided the opportunity for rising living standards and as demand grew, employment tracked productivity.

But since 2000, as the figure reveals, that has not been the case.  Much slower job growth, more involuntary part-time work, and a decline in labor force participation all contributed to a uniquely slack job market over that expansion.

Some analysts believe this divergence is caused by accelerating investment in labor-saving technology and while the evidence is largely anecdotal, my sense is that there’s something there.  Certainly, robotics are more prevalent and not just in factories by in other sectors as well.  EG, in services, a key contributor to productivity is throughput—“stack ‘em high and let ‘em fly,” as the retailers say.  I recently saw a documentary where robots were scurrying around an Amazon warehouse, getting products off of shelves.

More interesting and less well known (and surely less pervasive) are examples of software automating processes like legal research, medical diagnoses, and even writing an article about a sporting event.

Two caveats.  First, anecdote not equal to data.  Second, if bloggers existed back when looms and cotton gins were gearing up, they might have written similar pieces to this one.  We should always remember that the intervening variable of demand has made all the difference in terms of offsetting these technological advances in ways that lifted living standards for the broad public.  Arguably, this continues to be the case, as there are myriad ways in which the vast majority of us—rich and poor—benefit from technological advances.  But perhaps because income and wealth are some concentrated these days, demand growth is less broad-based, and labor-saving technology is hurting employment growth more so than in the past.

Third, as I alluded to at the end of the oped, if productivity growth is making us better off, we could choose to take those gains in less work and more leisure.  IE, by definition, higher levels of productivity mean we could enjoy the same amount of output with fewer hours of work.  But embedded in this solution–one seen more in Europe than here–is the assumption that the gains of productivity would be far more equitably distributed than they have been.  In fact, thery’re increasing less so.

At any rate, if the divergence shown in the figure above persists it will pose a challenge to achieving full employment labor markets, a challenge I argue must be met by job creation policies.  More on them in coming posts in this series.

Print Friendly, PDF & Email

18 comments in reply to "Full Employment: Is Labor-Saving Technology Making It Harder to Get There?"

  1. rjs says:

    what labor saving technological advance occurred between 2007 and 2010 that caused 14 million to be put out of work & caused the unemployment rate to double?

    • Jared Bernstein says:

      That’s largely cyclical–this other stuff is structural, a distinction I should have made in post.

    • Jared Bernstein says:

      That’s largely cyclical–this other stuff is structural, a distinction I should have made in the post.

  2. Kevin Rica says:


    Don’t worry about this. We have plenty of jobs. That’s why the Gang of Eight has created a “blue card” program to bring vital, low-skilled, low-wage workers that lazy, welfare-loving Americans won’t do.

    If you really believed that the problem is that we don’t have enough jobs, how could you support “Comprehensive Immigration Reform?”

    • Jared Bernstein says:

      I find a lot of other reasons to support CIR but also recall this post http://jaredbernsteinblog.com/a-few-thoughts-on-immigration-reform/ and particularly the parts on:

      –enhanced border security
      –implementing e-verify
      –the new ILMRB

      I do, as I said, worry about the guest worker piece. What in Keynes’ name is a “guest worker??!!”

      Also, dude, you need to recognize the positive labor market impacts of legalizing undocumented workers ALREADY HERE. Giving them RPI status (registered provisional immigrant (!)) does not add to labor supply and can only help in terms of reducing their exploitation, taking a low form of competition out of the market.

      • Kevin Rica says:

        “Also, dude, you need to recognize the positive labor market impacts of legalizing undocumented workers ALREADY HERE. Giving them RPI status (registered provisional immigrant (!)) does not add to labor supply and can only help in terms of reducing their exploitation, taking a low form of competition out of the market.”

        RPI status will allow them to walk into any business and offer to take anyone else’s job for the minimum wage and no benefits. (Unless, of course, the job is not covered by FLSA, and then they can offer to do it for less then the minimum wage.) And since RPI requires them to stay employed, they have every incentive to do exactly that.

        Alternatively, we can send out the no match letters and require every employee to go downtown and straighten out the problem. If there is a false positive (authorized worker got the no match), they can straighten out the SS file and get their benefits. Otherwise, after 90 days there should be a $1000/quarter fine on both the employee and employer until it is straightened out.

        That will free up millions of scarce jobs for American workers. Wages will go up!

        I’m a Truman Democrat and I approved this message.

      • purple says:

        The immigration bill will not just take into account people already here. It will increase the flow into this country as well, by a lot.

        I realize ensconced meritocrats think this is a swell idea, because it keeps their service help cheap and increasingly the professional help cheap as well.

        It’s also a nice way to get pre-packaged educated people into the US. without having to do anything about our own education system. If you can have India, etc. send you their PhD candidates why on earth would you invest properly in the U.S. education system ? Well, the proof is in the pudding on that one.

  3. Perplexed says:

    My first reaction to this was “really, we’re going to do the Luddite thing again?” Some zombie ideas will just never go away I guess, like the flagellants, anarchists (or more well known in their current, re-named form as libertarians or tea partiers), and austerians, they will just continue to surface as they have for centuries in a never ending game of whackamole.

    Obviously there’s something going on. The first question I’d have is “are we measuring what we think we’re measuring?” So we’re dealing with some ratio of “output” to “hours” but what actually goes into the numerator and denominator of these ratios? Given the time frame, I think the main suspect would be imported labor hours, the value of which made it into the numerator, but the hours themselves were excluded from the denominator. How comfortable are we that the ratio we’re using was properly adjusted to eliminate this possibility?

    The next most likely culprit is likely to be rents. If the gains in productivity are largely captured by monopoly profits in the form of software and hardware patents and licensing fees, how is the “intervening variable of demand” going to work its magic? It would only take a small reduction in “production” costs to completely eliminate the use of labor in production and replace it with technology. If prices don’t fall to reflect the actual costs of production and are kept artificially high, it seems pretty likely the “demand” solution would fail at the same time income & wealth concentration would increase doesn’t it?

    Maybe there’s an opportunity here for the “science of economics” to redeem itself it here by measuring, disclosing, and eliminating rents in the economy. Maybe they could even improve “welfare” enough to offset the tremendous costs their “political” solutions have helped to impose on the victims of the last 30 years of policies that support rents and redistribution of income and wealth to the top 5%. It would be great to see them be part of the solution instead of the other way around! As Larry the Cable Guy puts it: “Let’s gedder done!” http://pubs.aeaweb.org/doi/pdfplus/10.1257/jep.27.1.3

  4. Roger Chittum says:

    I’ve read this post twice and still cannot find the discussion of offshoring. China became a member of WTO in 2001, the very year the productivity growth and employment growth histories diverged. If the cause of the divergence is mostly robotics and other technology deployment, the divergence should also be showing up in other developed nations and the BRICs, no?

    Susan Houseman and others have found that, because of inadequacies in data collection, reported productivity has been overstated since offshoring of manufacturing has become significant. Louis Uchitelle reports on that here: http://www.nytimes.com/2009/11/09/business/economy/09econ.html?_r=5&ref=business& Also, as the layoffs were occurring in the US, remaining employees took on added worked and thus became more “productive”–lots of anecdotal evidence on that. Further, any employer worth a damn fires its least productive employees first, which has the effect of raising the average productivity of those who are retained.

  5. Robert Lee says:

    The trend for the increase in income to be retained by the upper few percent is long standing, not just since 2001. Income for the vast majority of the work force has stagnated for a generation. If these folks had enjoyed a living wage, increasing like the top percent’s income did during this period of increasing automation, they would have been buying all the new stuff that automation was helping produce, boosting the economy and generating plenty of new jobs. The cliche that Henry Ford new that people needed to have a living wage to be able to afford to buy his cars is still apt.

  6. purple says:

    The entire history of human existence is about labor saving technology. That’s what an axe is, as well. It’s what makes us human – tools.

    I look around and there is plenty of things to do, but of course “no money”. Right. So the excuse is we are in some new era, except we aren’t.

    I suppose this what happens when a society guts its knowledge of history and the liberal arts in favor of math & science (TM). Ahistorical thinking.

  7. Will says:

    The robots scurrying around are bringing shelves of inventory to human workers for picking. The company making the robots is called Kiva Logistics (now owned by Amazon). I recently toured Kiva’s factory. It’s a very cool place. Demand is so high that they need lots of humans to build the robots and write better software. As always: it’s the demand, stupid!

  8. Tom Cantlon says:

    Jared, I think what needs to be added here is the extra dimension of time. Yes, productivity eventually helps everyone but with a lag, and at some rate. When the lag and rate are slow enough the new generation can learn the new technology and not too many current workers are displaced, or other means of adapting to the change, then the pain isn’t too bad. When a whole generation is displaced and displaced again before it can even adapt, and other factors against enough jobs and good wages aggravate it, then what was a disruptive change that could be managed becomes a devastating disruption stretching across the working life time of a lot of people.

  9. Smith says:

    The notion that robotics is causing higher unemployment is laughable. Since the industrial revolution and later productivity gains of the 20th century didn’t lead to higher levels of unemployment perhaps, look at who actually controls employment and what really changed in the past 20 years. Think about what it means to live in a service economy. http://upload.wikimedia.org/wikipedia/commons/3/3d/Gdp-and-labour-force-by-sector.png For the entire period 1959 – 2013 agricultural employment was less than 5% in the U.S. Manufacturing declined well before the year 2000. So what you’ve really see measuring output is executive A charging executive B 1% more per year for services and this shows up as increased productivity. He used to hire more people when this happened, he thought more employees were needed to grow, and it also helped justify his higher salary. But repeated recessions and downsizing showed he could make more money operating in permanent downsizing mode. The gap on your graph has more to do with inequality and questionable measures of GDP. Financial services would be a prime example, with or without automated high-speed trading.

  10. tyler healey says:

    I’d love to read your thoughts on MMT’s Job Guarantee proposal. Here’s an explanation of it: http://www.nakedcapitalism.com/2012/02/pavlina-r-tcherneva-why-the-job-guarantee-is-superior-wonkish.html

  11. Alex says:

    If this robots thing was real, we’d be able to observe a massive spike in private capital investment. We do not observe such. Therefore the robots thing is nonsense.

  12. William says:

    I think the software automation needs a deeper look. Sometimes it accompanies hardware but in any case often a few dozen or hundreds of programmers and accompanying corporate support replace thousands of people.

    There used to be tens of thousands, if not hundreds of thousands of travel agents whose jobs are gone. Every town had a few agencies and big cities thousands of them. Mostly all gone now that we do it on our own on our phones etc.

    Smart electrical meters have eliminated tens of thousands if mostly union jobs going from house to house every month (and cutting carbon emissions by the side).

  13. France Childsky says:

    This year the major technology companies are coming out with 800 CPU (CPUs are computer brains) servers based on the ARM chip (what is in your smartphones and tablets, which are very fast and very cheap). According to the leaders in this field we will shortly be (measuring in months and quarters) producing 800 CPU robots (which will be very fast, cheap and smart). It will not be long before we reach 10,000 CPU robots with super-Watson levels of intelligence. Reaching that point, I would not be surprised if the Nobel in Economics would then be given to a robot (remembering that Watson tore up two of the greatest Jeopardy players in history), with humans never receiving a Nobel in anything every again. A little hyperbolic, but only a little. Food for thought.

    One additional thought on the positive side. We can teach a robot to be smart, fast, even creative, but not human. There is work that people can do, that robots can’t:

    A robot cannot teach young children or care about them.
    A robot cannot care for or about seniors.
    A robot cannot help people overcome poverty or disease or hunger
    A robot cannot cleanup the environment or save the planet
    A robot cannot build buildings, homes, road, treatment plants or energy systems
    A robot cannot really produce music or art (it will be interesting to see if robots will like robot music and art)
    A robot cannot produce fine craft or enjoy it
    A robot cannot be a doctor or nurse (but they can help a lot)
    A robot cannot product comedy or tragedy
    A robot cannot sell or market
    A robot cannot design or code computer systems (even though we have a glut)
    A robot cannot comprehend or appreciate justice, liberty or truth (sadly neither can our politicians)
    A robot cannot win the Nobel Peace Prize

    It should be noted that most of the work above is “social” – much like health care – and is very difficult to monetize (and maybe should not be monetized) given our current economic thinking. I do think we have a great opportunity, if we can expand our thinking (which I think we can).