The Hollowing Out

Menlo Park, Calif.

It has become a campaign ritual. Immediately after the release of unemployment figures on the first Friday of every month, Democratic and Republican spin shifts into high gear.

“Our mission is not just to get back to where we were before the crisis. We’ve got to deal with what’s been happening over the last decade, the last 15 years – manufacturing leaving our shores, incomes flat-lining – all those things are what we’ve got to struggle and fight for,” Obama declared at the Dobbins School in Poland, Ohio.

Romney took the opposite tack in Wolfeboro, N.H.: “This is a time for America to choose whether they want more of the same; whether unemployment above 8 percent month after month after month is satisfactory or not. It doesn’t have to be this way. America can do better and this kick in the gut has got to end.”

Both candidates are only tinkering at the edges of the most important issue facing the United States: the hollowing out of the employment marketplace, the disappearance of mid-level jobs.

The issue of the disappearing middle is not new, but credible economists have added a more threatening twist to the argument: the possibility that a well-functioning, efficient modern market economy, driven by exponential growth in the rate of technological innovation, can simultaneously produce economic growth and eliminate millions of middle-class jobs.

Michael Spence, a professor at N.Y.U.’s Stern School of Business, and David Autor, an economist at M.I.T., have argued that this “hollowing out” process is a result of twin upheavals: globalization and the hyper-acceleration of technological progress.

Just two weeks ago at the Aspen Ideas Festival, Alan Krueger, Chairman of the Council of Economic Advisers, stressed this theme:

If you look at the decade before the recession, the U.S. economy was not creating enough jobs, particularly not enough middle class jobs, and we were losing manufacturing jobs at an alarming rate even before the recession. And I would also put together, combined with those two problems, the polarization of the U.S. job market, the fact that we are getting more and more people at the very top and the very bottom and the middle has been shrinking.

In recent months, Erik Brynjolfsson, a professor at the M.I.T. Sloan School of Management, and Andrew McAfee, a research scientist at M.I.T.’s Center for Digital Business, have raised the stakes in this discussion with the publication of “Race Against the Machine” and a collection of accompanying essays and papers by the authors.

McAfee has graphically illustrated the key findings that he and Brynjolfsson find worrisome. The red line in the figure below, the employment to population ratio, tracks the ratio of the number of people working to the total number of working-age men and women in the United States.

On his blog, McAfee explains the graphic:

Since the Great Recession officially ended in June of 2009 G.D.P., equipment investment, and total corporate profits have rebounded, and are now at their all-time highs. The employment ratio, meanwhile, has only shrunk and is now at its lowest level since the early 1980s when women had not yet entered the workforce in significant numbers. So current labor force woes are not because the economy isn’t growing, and they’re not because companies aren’t making money or spending money on equipment. They’re because these trends have become increasingly decoupled from hiring – from needing more human workers. As computers race ahead, acquiring more and more skills in pattern matching, communication, perception, and so on, I expect that this decoupling will continue, and maybe even accelerate.

This view is controversial – especially McAfee’s argument that the decoupling of jobs from other positive economic developments “will continue, and maybe even accelerate.” In other words, the downward employment and jobs spiral will keep going, driven by structural forces. Policies to ameliorate the process – a shorter work week, a massive investment in education (for example, at the community college level), the disaggregation of complex tasks into simple functions that could be executed by mid-skill workers – may only slow the decline, not stop it. This is a deeply pessimistic vision.

“In my dystopian vision of the future, that red line (in the chart) keeps falling down – or suddenly drops off a cliff,” McAfee told The Times, adding: “All of the trends that I see and can identify are contributing to the hollowing out of the economy.”

In a videotaped interview on Bloomberg News, Brynjolfsson was somewhat more cautious:

I have to be brutally honest, I don’t think Andy and I are sure whether it’s different this time around. If you look at the data, this time it seems to be a lot more difficult. So it’s possible we are facing a regime change, a fundamental change in the way technology and employment interact with each other.

The Brynjolfsson-McAfee message has been generally well received in the high-tech community. On July 5, McAfee held the attention of an audience of young researchers and prospective entrepreneurs here at Singularity University. For over an hour after his lecture, students met with McAfee to explore the consequences of his argument.

The students’ questions:

How much can wealth accumulate for a small slice of the population at the top, while large numbers of people are forced to work for ever lower pay or to drop out of the workforce altogether? For such a future society to function, would wealth need to be (coercively) redistributed from the top to those below, in order for the mass of the jobless population to survive? Who would have power and how would tax and spending policies be determined in such a radically bifurcated, automated, workless society?

Many reviews of “Race Against The Machine” have been favorable, including those in publications supportive of free markets, including the Economist and the Financial Times.

McAfee noted in our interview that some critics have accused him and Brynjolfsson of accepting the “lump of labor fallacy” (the idea that there is a fixed amount of work available) in defiance of economic history. In the aftermath of major periods of technological advance, including the transition from agriculture to industry, employment has grown enormously.

James Hamilton, an economist at the University of California, San Diego, challenged the “Race Against The Machine” thesis in an e-mail to The Times:

I am very skeptical of the claim that technology itself is the problem. In 2005, the average U.S. worker could produce what would have required 2 people to do in 1970, what would have required 4 people in 1940, and would have required 6 people in 1910. The result of this technological progress was not higher unemployment, but instead rising real wages. The evidence from the last two centuries is unambiguous – productivity gains lead to more wealth, not poverty. The unemployment since 2007 was not caused by gains in productivity or increased automation, but instead by loss of demand for the product that the workers had been producing, for example, a plunge in the demand for new home construction.

Amar Bhidé, author of the book “The Venturesome Economy,” and senior fellow at the Center for Emerging Market Enterprises at Tufts, did not mince words responding to a request for comment from The Times:

As you might guess I find the storyline rather unconvincing and Luddite. What’s new about automation? I’ve been banging away for years about the phenomenon of non-destructive creation as a vital complement to creative destruction. The two don’t move in lock step but I have no reason to believe that non-destructive creation has ceased. Until someone persuades me it has, I will limit my anxieties to global warming, financial misregulation, a screwed up health care system, etc.

McAfee countered in an e-mail that “this time really is different,”  arguing:

All previous waves of automation affected only a small subset of human skills and abilities. To oversimplify a bit, the industrial revolution was about building machines that had (much) more brute strength than we did. For all mental work, the industrial revolution was meaningless – you still needed people.

Until recently, the digital revolution also didn’t affect that many human skills and abilities. Computers became better at math, and at some clerical abilities, but we people were still miles ahead in other areas. So employers needed to hire humans if they wanted to listen to people speak and respond to them, write a report, pattern-match across a large and diverse body of information, and do all the other things that modern knowledge workers do.

Employers also needed people if they wanted lots of physical tasks done, including driving a truck or vacuuming a floor. The same with most tasks involving sensory perception, such as determining if a soccer ball has crossed a goal line.

All of the above abilities have now been demonstrated by digital technologies, and not just in the lab, but in the real world. So employers are going to switch from human labor to digital labor to execute tasks like those above. In fact, they’re already doing so. I expect this process of switching to accelerate in the future, perhaps rapidly, because computers get cheaper all the time, are very accurate and reliable once they’re programmed properly, and don’t demand overtime, benefits, or health care.

Brynjolfsson, who is more optimistic, said in an interview with The Times, “we are hopeful that that (job growth) will happen, but there is no guarantee of it. There is no economic law that says everyone benefits from technological improvement.” He also pointed out that the surge in inequality driven by rising incomes at the very top of the distribution suggests strongly that the benefits of digitization have not been widely spread.

“The problem is not tech stagnation,” as some have argued, “but the opposite,” Brynjolfsson contends. “Technology is rushing ahead faster than humans can adapt.” The difficulty of human adaptation is, in turn, likely to get worse, he added, because technological innovation – as in Moore’s Law (predicting a doubling of computer capacity roughly every two years) – grows exponentially in scope. The total number of non-farm jobs in the country is now 5 million less than in January, 2008. The 3.7 million jobs added to the economy have not been enough to make up for the 8.7 million jobs lost in 2008-9.

Brynjolfsson and McAfee have a list of 19 proposals that they support – which range from massive investment in education, infrastructure and basic research, to lowering barriers to business creation, eliminating the mortgage interest deduction and changing copyright and patent law to encourage new (as opposed to protecting old) innovations.

Any effort to ameliorate the damaging consequences to the employment marketplace stemming from technological innovation, according to Brynjolfsson, requires substantial government action at a time when “the political system is the most dysfunctional part of our society.”

McAfee and Brynjolfsson argue that in a race against machines, humans will lose. In their view, “the key to winning the race is not to compete against machines but to compete with machines.” The question, then, will be whether humans can adapt at anywhere near the pace needed to keep up.