The history of the robot future’s future history: The graph below (click to enlarge) represents three decades of U.S. middle class employment shrinking as a share of the employed labor force, with the occupations along the graph’s X-axis proceeding left to right from the least- to the highest-paid. The top four occupations and three of the bottom four have increased their share of employment, at the relative expense of the middle three.
This comes via this Third Way paper by Frank Levy of MIT and Richard Murnane of Harvard, which includes an extended section regarding the trends (emphasis ours):
The hollowing out is the result of multiple factors, but it is consistent with the idea that occupations subject to computer substitution grow relatively slowly. Low wage work—Personal Care, Personal Services, Food Preparation, and Building and Grounds Cleaning—have all grown in importance and all involve non-routine physical work that is hard to computerize. Technicians and Professional and Managerial Occupations also have grown in importance. All involve abstract, unstructured cognitive work that is hard to computerize. Moreover, all rely on computers as complements including jobs like Network Manager that wouldn’t exist without computers.
By contrast, occupations in the middle of the distribution—Machine Operators, Production, Craft and Repair Occupations, Office and Administrative—have declined in importance. From the perspective of 1979, each of these occupations contained significant amounts of routine work that could be expressed in deductive or inductive rules and so were candidates for computer substitution and/or offshoring.
And the BLS expects these trends to roughly continue:
The BLS projects total employment growth of 14.3% between 2010 and 2020, with the fastest growing occupations involving unstructured problem-solving, working with new information, and non-routine physical activity: Healthcare Support Occupations (+34.5%), Healthcare Practitioners and Technical Occupations (+25.9%), Community and Social Service Occupations (+24.2%), Construction and Extraction Occupations (22.2%), Computer and Mathematical Occupations (+22%).
Conversely, occupations with potentially strong computer substitution are projected to grow by less than 14.3%: Production Occupations (+4.2%), Office and Administrative Support Occupations (10.3%).
The period of the Great Stagnation has brought the double-whammy of a slower pace of overall productivity improvement combined with an intense, protracted labor-saving quality in many of the economic sectors that have most contributed to the improvement.
The Levy-Murnane report is an effort to understand the second whammy, making the case that even the graph above understates the evolution in labor markets because it captures only the employment distribution of certain occupations, not the changes within them.
The researchers identify workplace tasks according to five kinds…
– Solving Unstructured Problems
– Working with New Information
– Routine Cognitive Tasks
– Routine manual tasks
– Non-Routine Manual Tasks
… and their point is that jobs heavily emphasizing the two routine tasks
are easy to replace with computers or to outsource. These have traditionally
been middle income jobs, while the other, non-routine tasks are concentrated in
high-paying (executives, entrepreneurs) and low-paying (furniture movers, truck
Here’s a passage from Robert C Allen’s Global Economic History: A Very Short Introduction, which is a delightful cheat sheet for people who don’t know much about economic history but need to pretend they do for a blog post:
The rate of economic growth achieved in the century after 1760 (1.5% per year) was very low by the standards of recent growth miracles in which GDP has grown by as much as 8-10% per year. However, Britain was continuously extending the world’s technology frontier, and that is always slower going than catching up to the leader by importing its technology, which is how countries have grown very rapidly. Moreover, the great achievement of the British Industrial Revolution was that it led to continuous growth, so that income compounded to the mass prosperity of today.
Technological change was the motor of the Industrial Revolution. There were famous inventions like the steam engine, the machines to spin and weave cotton, and the new processes to smelt and refine iron and steel using coal instead of wood fuels. In addition, there were a host of simpler machines that raised labor productivity in unglamorous industries like hats, pins, and nails. There was also a range of new English products, many of which, like Wedgwood porcelain, were inspired by Asian manufacturers.
In the 19th century, engineers extended the 18th-century mechanical inventions across the board. The steam engine was applied to transportation with the invention of the railway and the steamship. Power-driven machinery, whose use was initially restricted to textile mills, was applied to industry generally.
A question for historians of economic thought: If you had asked economists in 1759 whether such a fundamental shift was ever likely to happen, would they have thought the possibility ludicrous? Would they have argued that in the 12,000 years since the dawn of agriculture, humanity had yet to escape the Malthusian Trap — and therefore why believe that such an escape was even possible?
What about in 1780? In 1830? 1850?
At what point would it have been clear that the world had changed, permanently? What would have tipped them off? (I’m guessing it wasn’t the Wedgwood porcelain.)
The question is worth asking because the appeal to history is the most common counterargument to stories about the potential for robotics and automation-enabling technologies to massively displace workers while enriching the capital owners, leading to resurgent inequality. “Remember the Luddites!” goes the response. “These shifts are temporary, and eventually the workforce adjusts to the higher-productivity environment. You just don’t know your history.”
So surely it’s worth retorting that the history being appealed to is only two and a half centuries old? When compared against the roughly 50,000 years (or whatever) since people have been anatomically and behaviorally recognizable as what they are now, that’s nothing.
Which isn’t to say that the trends and regularities of the time since 1760 are no longer relevant or informative or applicable. That would be silly.
It is to say that such a counterargument is insufficient to dismiss the possibility that something new is happening. An economist arguing in 1759 that an industrial revolution was unlikely could have appealed to a much longer period of time, and still would have been wrong.
For the past few years, and especially since the publication of Race against the Machine, the economics blogosphere and commentariat has debated whether the same trends described by Levy and Murnane have been properly explained and whether they will continue.
It’s all fascinating, but the truth is that it’s very difficult to know what to look for. We can point to graphs showing the split between the share of income going to capital versus labor, but we can’t properly weight the conceivable causes. We can look at the disturbing growth of wage and wealth inequality since the downturn of 2008 (or even for the past three decades), but we don’t know how much blame should be assigned to, say, improper policy versus the under-aggressive application of proper policy.
Such is life in a discipline where the frame of reference never escapes the realm of counterfactual. That’s no excuse for nihilism, or to ignore the lessons of earlier experience. Sometimes the counterfactual can be reasonably inferred.
It simply means that understanding even the big recent trends is hard. And discerning the big trends of the future is even harder, especially when the only guide is a very recent past.