The Missing Middle Class: Jobs in the Second Machine Age

Xconomy National — 

Technological change is not a tide that lifts all boats in our economy. The truth is that it’s more like a tsunami. It threatens to overturn all the boats and drown their occupants, sparing only the lucky few who have already reached safety in the hills.

That’s the kind of admission you won’t often see here in the pages of Xconomy. The X in our name, after all, stands for exponential, a reference to the stunning pace of technological progress and economic growth over the past 75 years—growth attributable largely to advances in computer hardware and software and the organizational changes companies have made to exploit them.

But the reality is that technology’s great bounty isn’t reaching everyone. This is the flip side of the phenomenon I described in my column last week—the Medici effect. Thanks to the ongoing technology boom, there’s a growing supply of super-rich companies and individuals in Silicon Valley. To their credit, they’re making philanthropic investments on a scale so large that you have to look to places like Renaissance Florence for parallels.

But in between the rich and the poor, the skilled and the unskilled, there’s a big hole opening up, as millions of middle-class workers lose their jobs and can’t get them back. It’s fair to blame computer technology for widening this gap. It’s a phenomenon MIT economist David Autor has called labor market polarization. In his words, this is a pattern of “expanding job opportunities in both high-skill, high-wage occupations and low-skill, low-wage occupations, coupled with contracting opportunities in middle-wage, middle-skill white-collar and blue-collar jobs.”

Because we here at Xconomy cover the business of innovation, we tend to focus mostly on the mechanisms of economic growth, and not so much on the question of who’s benefiting from this growth and who isn’t. But I have looked at the issue of technology-induced unemployment a couple of times before, in a 2011 column and in the run-up to our 2013 robotics forum. And this week, after spending a couple of days at an interdisciplinary conference on technology and employment, it’s on my mind again.

Experts at the Innovation for Jobs Summit, organized by Swedish-born innovation journalism advocate (and Xconomist) David Nordfors and Internet pioneer Vint Cerf, offered a range of solutions for labor market polarization. I’ll describe a few of these in a moment. Some of them are being tested in places like Sweden and Finland, whence many of the conference attendees hailed; none will be simple to implement. But neither are they out of reach, if we can start an honest conversation now about what we value as a society.

First, let’s get clear about the scope of the problem. While the jobless rate in the United States has dropped since the height of the Great Recession in October 2009, the bigger unemployment picture is far from encouraging. Looking at the official rate of 6.7 percent masks the real extent of the crisis, because it leaves out people who have given up searching for work. The total unemployment rate, which counts discouraged workers as well as those who have part-time jobs but would prefer full-time work, is stubbornly high, at 12.7 percent. The youth unemployment rate is even higher, at 16.3 percent for all 16-to-24-year-olds and 28.2 percent for black youth. (These are all Labor Department statistics, and they’re up to date except for the youth unemployment data, which is from mid-2013.)

And while the economy is slowly getting stronger, you’d be hard pressed to find an economist who thinks unemployment will ever return to the 4 percent range, where it stood before the recession. If you want to understand why, go read two recent popular-economics books: Tyler Cowen’s Average is Over: Powering America Beyond the Age of the Great Stagnation, and Erik Brynjolfsson and Andrew McAfee’s The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Machines. The basic problem, these three authors agree, is that the recession gave employers the excuse they’d been waiting for to eliminate millions of workers who’d become too expensive. They were overpaid, in the sense that they’d been doing jobs that could now be done more cheaply by people in other countries or—just as frequently—by software or machines.

The result has been a thinning out of the middle class, with many moderately skilled, medium-wage jobs disappearing for good. When Steve Jobs told President Obama at a 2011 dinner that “those jobs aren’t coming back,” he was referring to manufacturing jobs that Apple had moved from the U.S. to contractors in Asia, but he could just as well have been talking about jobs lost to improved automation. Apple partner Flextronics boasted that it was creating 1,700 new jobs when it upgraded a factory in Austin, TX, to make Apple’s new Mac Pro. But that was just a fraction of the 8,800 jobs eliminated in Austin when Dell closed a desktop manufacturing plant there during the recession.

The new factories always need fewer workers than the old factories. That’s great for productivity, but it’s a bit of a problem for people adapting to Brynjolfsson and McAfee’s “second machine age,” in which we’re starting to feel the full force of advances in computer technology.

To repeat, these big structural changes in the job market aren’t a product of the downturn—they were merely laid bare by it. The labor force participation rate, the percentage of people who have jobs, peaked in 1999 and has been on a steady decline ever since. (1999 was also the year when the real income of the median American household reached its height of $55,000; it has since fallen by 10 percent.) What’s happening is that organizations are shedding every single one of the mid-wage, routine jobs that can be handed over to machines—think telephone receptionists or toll collectors at bridges. At the same time, they’re creating a lot of new, low-wage jobs, but the folks who lost their medium-wage jobs haven’t yet lowered their expectations far enough to take them. So they’re staying out of the workforce, by the millions. “However unrealistically, most of these individuals are holding out for a better offer than what the American economy is serving up,” as Cowen puts it.

If you are highly educated, or you have skills that can’t easily be translated into software algorithms, then congratulations: you’ve made it to the hill. You’re part of the high-skilled elite; you fared far better than your fellow Americans during the downturn, and now you are once again enjoying an economic windfall. In 1983, the top 20 percent of U.S. families held 81.3 percent of household wealth, according to an Economic Policy Institute briefing paper. By 2009, they held 87.2 percent. In essence, 100 percent of the new wealth created in that 26-year period went to the top fifth of U.S. households. (More than 100 percent, in fact; the top 20 percent of households also captured some of the wealth that had formerly belonged to the bottom 80 percent, as Brynjolfsson and McAfee point out.)

To a large degree, what differentiates winners from losers in this economy is whether they have the right education, training, skills, experience, and talents to benefit from “skill-biased technical change,” the economists’ term for cases where technological progress isn’t benefiting everyone equally. The right skills today are those that can be used to augment the things computers already do well (finance, consulting, software engineering), or those computers won’t be able to mimic until far in the future (gardening, marriage counseling, advertising, management). If you’re looking for job security these days, Cowen writes, “The key questions will be: Are you good at working with intelligent machines or not? Are your skills a complement to the skills of the computer, or is the computer doing better without you? Worst of all, are you competing against the computer?”

But even if you’re in the first or second quintile of wealth, education, and skill, you may not be as safe as you think. There are still a lot of inefficient, essentially guild-dominated professions where huge productivity gains will be extracted once technology starts to disintermediate things, just as it has in manufacturing, the media, and so many other fields.

I’m talking about areas like government, healthcare, education, and law. In each of these fields, Silicon Valley is starting to carve away at the establishment. Just think of services that consumers can use to get a rough self-diagnosis without going to the doctor, such as WebMD or HealthTap; or sites where students can learn college-level material without ever setting foot on campus, like Khan Academy, Udacity, and Coursera; or online legal services sites such as Nolo Press, Rocket Lawyer, or LegalZoom. Such services aren’t putting downward pressure on wages yet, but the technology is still in its infancy. “The process appears to have a long way to run,” Cowen speculates. “To be blunt—while I know I can’t prove this—I wonder how much of the middle class consists of people in government or protected service-sector jobs who don’t actually produce nearly as much as their pay.”

And if true machine intelligence emerges in our lifetimes, all bets are off. So far, progress toward systems that can reason and act as effectively as humans has been limited. Even Google’s self-driving cars only work under normal traffic conditions (which is why there’s always a human in the driver’s seat, ready to take over). But somewhere down the road, maybe a few decades from now, maybe longer, that will probably change. As Robin Hanson, a colleague of Cowen’s at George Mason University, has shown in a widely read paper, an economy where machines can do most jobs will be almost unrecognizable to us. At first, there will likely be rapid and enormous productivity gains and economic growth, as robots build everything people need for ever-lower prices. But eventually, the wages of the remaining human workers will start falling even faster than prices, making it impossible for people to afford the things the machines are making. The potentially dystopian result: impoverishment and starvation for everyone except the owners of the machines.

Scenarios like that are enough to bring out the Luddite in the most hard-core tech optimist. But renouncing technology isn’t the answer to labor market polarization. Short of abandoning free-market capitalism—which even China has embraced, in its own way—there’s no way to stop companies from adopting machines and systems that improve productivity.

So what can individuals do to adapt to an era when skill-biased technical change is the norm? What can policymakers do to put middle-skill people back to work?

Here are some ideas, drawn partly from Cowen, Brynjolfsson, and McAfee, and partly from talks I heard at the Innovation for Jobs Summit. (The meeting was held under Chatham House rules, meaning I can’t reveal who said what.) Some of these proposals are more plausible than others, given the political realities in Washington, D.C., and in the nation’s state houses and city halls. But they’re all worth weighing, and even testing through pilot programs.

1. Get Used To It. Accept the idea that there will be a large and permanent class of unemployed people. Improve and extend public benefits to reduce the stigma and suffering of joblessness. Pay for these benefits partly through increased taxes, and partly through productivity gains from modernizing the agencies that administer benefits. (You might also call this the Nordic Solution, and it has a downside. In countries like Finland and Sweden, which have an overall tax burden in the 45-percent range, unemployment benefits are so generous that—by the admission of the countries’ own policymakers—there’s little incentive for hundreds of thousands of long-term unemployed to look for jobs.)

2. Grow Our Way Out of It. Pursue monetary, fiscal, budget, and tax policies that stimulate economic expansion. In theory, increased demand for goods and services will eventually force companies to start hiring at all levels.

3. Get On the Retraining Train. Reengineer public services to vastly improve the scope, quality, and responsiveness of job retraining programs. The U.S. Department of Labor is beginning to have some with success with career search sites like MyNextMove and the Transition Assistance Program for military veterans reentering the civilian workforce, but the advice these sites offer isn’t always tailored to the skills employers need today. In Finland, an experiment is underway to hand over some retraining programs to the private sector, which presumably understands the job market better, and would be offered financial incentives to get trainees back into the workplace.

4. Replace Old Technology Clusters with New Ones. Figure out what skills people in a given region already have, and give private industry incentives to innovate in areas that match those skills. In the Skåne region of southern Sweden, the contraction of companies like Ericsson and Sony Mobile has left many people with skills in the mobile industry out of work. At Lund University, there’s a new research institute, the Mobile and Pervasive Computing Institute, dedicated to supporting startups that are exploring the Internet of Things—the emerging network of distributed, cloud-connected devices, from thermostats to digital contact lenses. Much of the underlying technology comes directly from the mobile industry, and those companies will need experienced workers.

5. Reinvent Education. Shore up mass education by rewarding the best teachers with much higher salaries. Encourage experiments with other ways of learning, such as massive online open courses (MOOCs), and agree on a system of certificates or credentials that will help the people who’ve completed MOOCs find jobs. At the same time, make room for new ways to finance traditional higher education. Look to examples like Pave, Upstart, and Michigan’s proposed pay it forward plan, which help students raise the money for college tuition in return for a percentage of their future income.

6. Outwit the Computers. Through college counseling and retraining programs, nudge future workers and job seekers toward roles that are unlikely to be disintermediated by technology: sales, marketing, finance, support, personal services, management. These are irreducibly complex jobs where the human touch matters.

7. Support Startups and Small Businesses. As the Kauffman Foundation has been arguing for years now, all net new job growth comes from startups. Yet while the overall rate of new business formation is still healthy, many startups have just one employee: the founder. The rate of formation of larger companies is actually falling, perhaps due to immigration restrictions and overly burdensome regulations. It’s time to clear these thickets, and uncork the banking system so that the operating capital small businesses need in order to hire people will be cheaper. Crowdfunding could be a help: sites like Indiegogo and Kickstarter are giving more and more ventures the ability to gauge market demand for their products even as they raise seed funding. But it’s important that the emerging regulations around equity-based crowdfunding aren’t so restrictive that they end up driving investors away. And this isn’t just about technology startups: it’s better to create 10,000 small services companies with 10 jobs each than to create one Google, which has only 44,000 employees.

8. Rethink What It Means to Have a Job. My suggestion above that we just get used to long-term unemployment was mostly facetious. There are good social and moral reasons, not just economic ones, for believing that everyone needs work: it’s one of our main sources of self-worth, dignity, and purpose. But our definition of “work” is curiously narrow. It doesn’t seem to include raising children, for example: stay-at-home moms and dads aren’t counted in the workforce and aren’t compensated for the absolutely vital work they do to give children a foundation of learning and support.

In an ideal world, people who choose to add value to society by caring for children or elders, volunteering for community groups, or maybe even being poets or painters or actors, would be appropriately compensated through mechanisms like a government-guaranteed basic income or a negative income tax. Or maybe we should think about lowering, rather than raising, the minimum retirement age. In such a “pretirement” scheme, people could opt to begin receiving their Social Security and Medicare benefits early, in return for a commitment to doing community service.

Or we can just do nothing at all and wait for technology to evolve, trusting that it will reveal new kinds of jobs that we never anticipated. (In a world where Amazon is sending us packages via quadcopter, for example, I’m betting there will be a need for UAV repair technicians and micro-air-traffic-controllers.) This laissez-faire approach to labor force adaptation is how we’ve handled technological unemployment for the last couple of centuries.

The problem is, it doesn’t seem to be working anymore. There wasn’t a job waiting for you on the other side of the buggy-whip era if you were a horse. Millions of people face the same dilemma today, from warehouse pickers being replaced by robots to postal workers who have less and less physical mail to deliver.

An idea often attributed to sci-fi author William Gibson says that the future is already here, it’s just not evenly distributed yet. And it never will be, unless we come to grips with the fact that technology doesn’t welcome all workers equally. If we want to survive the tsunami, we’re going to need more routes to the hills.

By posting a comment, you agree to our terms and conditions.

8 responses to “The Missing Middle Class: Jobs in the Second Machine Age”

  1. KJ says:

    9. Mitigate the oversupply of labor by having fewer children?

    In his book Baby Bust, Stewart Friedman found that the rate of graduating Wharton undergrads who plan to have children has dropped by nearly half over the past 20 years. Maybe these students are presciently sensing that as automation climbs the skill ladder so that worldwide demand for all types of labor declines, and as working lives lengthen, only a reduced birth rate can balance the supply of labor with declining demand. The utopian idea that increasing labor productivity will lead to more leisure time with improving living standards doesn’t conform to the laws of supply and demand for labor.

    Kensho claims their software performs jobs that students like those surveyed might have filled, “…the kind of work usually done by well-educated junior (financial) analysts.” “There are several hundred thousand people employed in that capacity. We do it with machines…We’re competing with people.” ( )

    • Wade Roush says:

      KJ — I think the world is already moving furiously in the direction of slower population growth. Reduced labor demand (you don’t need as many kids to support you in old age) is one of the big reasons, along with reduced infant mortality, increased availability of contraception, and the education and empowerment of women. Even so, the world population is going to hit 9 or 10 billion, and that’s a lot of people who will need jobs.

      Even if software replaces junior financial analysts, though, I tend to think an MBA will still be a pretty good investment. Management is one of those human-touch professions that computers will have a hard time taking over.

      • Todd McKissick says:

        I hate to be the bearer of bad (???) news but managers will be one of those fields where “it can’t be done” until one day when it can and then is done… overnight. When the mindless duties of process, employee or resource oversight is finally seen as a target, it won’t take but months before the first software bot is in place. After that, it’s just copy, paste and adjust the parameters and there goes the middle chunk of your labor costs. What top level department head wouldn’t jump on that in a heartbeat.

        Sorry, but short of truly creative and innovative positions, there is little that’s safe from being e-sourced. Not that this is bad because it just ushers in the switchover that much faster. That being when we stop evaluating everyone on what they can offer us and begin to evaluate them on what the person is.

        “From each according to his desire and from all that mechanisms have to offer, to each according to his existence.”

        • Wade Roush says:

          You could be right, Todd. It’s probably foolhardy to say that any given profession is un-computerizable — even journalism (the news about the earthquake in Los Angeles this week was broken by an algorithm). But I’m skeptical. As soon as some organizations start replacing managers with software, it would give an advantage to the competitors who still have really good humans running things. So while automation might put pressure on human managers to be better, I can’t see them going extinct. Maybe the lesson for MBAs is: make sure you focus more on creativity than process.

          • KJ says:

            Even if the management function isn’t automated, a decline in the labor workforce would lead to a decline in the number of managers needed to manage that workforce.

            > …an MBA will still be a pretty good investment…

            To clarify, Stewart Friedman surveyed graduating Wharton undergraduate students, not MBA students.

          • Wade Roush says:

            Good point, but if it’s a question of where to position yourself on a sinking ship, management will still be one of the last roles to be automated.

            I confess I didn’t realize that Wharton has an undergraduate program separate from U. Penn. Thanks for the clarification.

          • Todd McKissick says:

            Applying simple algebra to this would look something like this.

            The economy consists of W individual businesses.
            All businesses are a collection of X individual process.
            A given process has traditionally consisted of Y managers per process.
            The average process has a ratio of 1 manager to Z workers.

            So there’s a total of W*X*Y*Z workers with a W*X*Y managers at that same Z ratio. If automation reduces X, Y and Z, and you’re recommending to continue the trend of workers to become managers, any reduction will DECREASE the total number of manager, percentage-wise as well, from the system. In short, managers will become over abundant sooner and faster than any other group.

  2. 4Gbill says:

    The solution is a fundamental shift in the economy in which consumers become producers of energy, education, food, healthcare, and information to become more self-reliant and autonomous but also become able to sell specialized goods and services. Technology empowers people. The shift is in who owns the machines. When a consumer can buy and operate a 3D printer to produce goods, the ownership of a “factory” shifts because the barrier of large capital requirement is eliminated. And distributed production reduces shipping costs and delays. Cities are at risk because they represent an obsolete, high cost paradigm that is destroyed by autonomy and high speed global networking such as the Internet. Cities and other organizations were needed to enable efficiency of interactions before autonomy and global communication networks. Ronald Coase, famous economist, is best known for two articles in particular:”The Nature of the Firm” (1937), which introduces the concept of transaction costs to explain the nature and limits of firms, and “The Problem of Social Cost” (1960), which suggests that well-defined property
    rights could overcome the problems of externalities in economics. Technology has now empowered people such that large firms as owners and operators of production are becoming obsolete and inefficient in comparison to distributed production by individuals. Open software is another example of the paradigm shift.