Vanishing Jobs

  America has long cherished revolutions in technology.  It is an article of faith that they are always for the good, spawning vital new industries, spurring economic growth, creating more jobs than they destroy.  GDP, productivity, profits and wages increase,  all boats rising on a waxing tide.

  Indeed, this rosy scenario has repeatedly been validated.  Although mechanization of agriculture displaced most of the country's farmers in the 19^th and 20^th centuries (they accounted for 90% of the working population in 1800, only 2% in 2000), they and their children were fully absorbed into a growing urbanized society.  Replacement of the horse and buggy by the automobile created millions of net new jobs in the 20^th century, not only in car manufacturing but in the mobile economy created as Americans took to the roads.  Motion pictures, radio and television displaced other forms of entertainment, but spawned megalithic replacements.  Following all these innovations, GDP and the standard of living persistently increased.

  This is not to say that technology revolutions did not cause painful dislocations for workers whose skills were no longer needed, but the pain was temporary as needed new skills were acquired in a changed economy.  Workers ultimately benefited with shorter hours and higher pay.  When I was a teen-ager, there was frequent talk about a likely 30-hour work week and a nagging concern that Americans wouldn't know how to use all of their leisure time.

  Could that virtuous cycle have come to an end?  In an exquisite irony, MIT—that great bastion of technology—has an article in this summer's issue of its Technology Review declaratively entitled "How Technology is Destroying Jobs."  Much of it is based on the work of Erik Brynjolfsson and Andrew McAfee of the MIT Sloan School of Management.  The striking graphic below, central to the article, shows how the growth of productivity is now greatly outpacing the growth of employment in the U.S.;  innovation has caused productivity to continue skyrocketing, while employment has recently remained almost flat.

The increasing gap between productivity and employment.

[Source: MIT Technology Review, July-August 2013, p. 31.]

  Part of the flat-lining of employment is surely the effect of the two recessions in the first decade of this century, shown by the two dips in the employment curve during those years.  However, according to the Technology Review article, Brynjolfsson and McAfee "believe that rapid technological change has been destroying jobs faster than it is creating them, contributing to the stagnation of median income and the growth of inequality in the United States."  Robots and computers, they say, are replacing humans faster than they are increasing the need for humans having mid-level skills.

  As a result, the fraction of jobs requiring a mid-level skill—those increasingly easily done by automation—seems to be declining for good.  Many blue-collar jobs, mostly in manufacturing, continue to be permanently replaced by robots; more and more white-collar positions like those in middle management, bookkeeping and paralegal research are being taken over by computers.  The swath of jobs being replaced by automata is therefore steadily widening. Contrarily, the fraction of jobs not easily subject to replacement by automata (mostly low-skilled jobs in the service industries like home care and hairdressing, and high-skilled jobs like designing iPads and apps) continues to increase.  Altogether, though, employment as a fraction of population is falling.

  The first chart below shows the slump from 1980-2005 in employment in mid-skill levels.  The second chart shows an associated redistribution of income during that period—a more lopsided curve with most of the change going to people having the very top skill levels.  In effect, the great middle class, defined by mid-level skills and incomes, which once embraced the preponderance of the population, is diminishing rapidly.  Not all of these changes can be attributed to automation alone; off-shoring of mid-skill jobs to countries with lower wages has contributed.  However, note that both charts cover a quarter century prior to the Great Recession, which therefore cannot be blamed for the trend, but certainly has accelerated it.

Changes by skill level in share of total unemployment and hourly wages, 1980-2005.

[Source: MIT Technology Review, July-August 2013, p. 33.]

  If Brynjolfsson and McAfee are right, and I believe they are, new norms for unemployment and underemployment, much higher than we have traditionally come to expect, are here to stay.  The situation will be made worse as access to retirement benefits moves upward from 65 to 70.

  It's not a pretty picture.  The bifurcation of the country into those who have substantial wealth and skills and those who have little of either—with fewer and fewer in the middle—will further reflect itself in partisan politics, as if its present polarization were not disastrous enough.  The "normal" future that we have been awaiting will not come.  The future is already here, with depressingly different norms.

  I don't think that all this is just a new Luddist scenario, fated to fade away. The digital revolution is exponentiating so much faster than previous technology revolutions that I don't believe employment will ever be able to catch up. The future of American society under such regime is anyone's guess.

  To find out more, you might want to read Brynjolfsson and McAfee's very short and readable book, Race Against the Machine: How the Digital Revolution is Accelerating Innovation, Driving Productivity, and Irreversibly Transforming Employment and the Economy.  Its first three chapters elaborate the statistics given above.  The last two attempt to cast a more optimistic light on the trend by proposing numerous steps that could be taken to reverse it.  Those cures seem to me like bromides, not likely to be implemented and probably ineffective if they were.

  If you weren't already aware of the likely permanence of the new economy, sorry to spoil your day.  In fact, a lot more than a day has been spoiled.