Pages

Friday, August 7, 2015

How Automation Affects Labor Markets: Rise of the "New Artisans"?

Will the plausible improvements in automation and robotics that seem to be on their way lead to mass unemployment? If not unemployment, what kinds of changes might they cause for the distribution of wages in labor markets? David H. Autor tackles these issues in "Why Are There Still So Many Jobs? The History and Future of Workplace Automation," which appears in the Summer 2015 issue of the Journal of Economic Perspectives.

(Full disclosure: I have worked as Managing Editor of JEP since 1986. Autor was the editor of JEP from 2009-2014, and thus was my boss during that time. All articles in JEP back to the first issue in 1987 have been freely available online since 2011, courtesy of the American Economic Association.)

Autor's argues that most discussions of automation and the labor market focus on one effect--that is, how automation might substitute for certain existing jobs. That's clearly one potentially important effect, but not the only one. Autor suggests that a fuller framework for how automation affects employment and wages needs to look at three broad factors: 1) how automation substitutes for some jobs but complements others; 2) whether it's easier or harder for workers to shift into certain new jobs (that is, what economists call "the elasticity of labor supply"); and 3) what goods and services are demanded as income expands.

Of course, it's obvious that modern worker would have a dramatically lower standard of living if they were deprived of machines and computers and instead had to use a stick they found in the forest to till fields and kill prey. But more generally, Autor writes:
"[T]asks that cannot be substituted by automation are generally complemented by it. Most work processes draw upon a multifaceted set of inputs: labor and capital; brains and brawn; creativity and rote repetition; technical mastery and intuitive judgment; perspiration and inspiration; adherence to rules and judicious application of discretion. Typically, these inputs each play essential roles; that is, improvements in one do not obviate the need for the other. If so, productivity improvements in one set of tasks almost necessarily increase the economic value of the remaining tasks.
An iconic representation of this idea is found in the O-ring production function studied by Kremer (1993). In the O-ring model, failure of any one step in the chain of production leads the entire production process to fail. Conversely, improvements in the reliability of any given link increase the value of improvements in all of the others. Intuitively, if n − 1 links in the chain are reasonably likely to fail, the fact that link n is somewhat unreliable is of little consequence. If the other n − 1 links are made reliable, then the value of making link n more reliable as well rises. Analogously, when automation or computerization makes some steps in a work process more reliable, cheaper, or faster, this increases the value of the remaining human links in the production chain."
However, if many workers are able to flock into the new jobs created by complementarities with automation, then wage increases in that job category are likely to be small. Only if it's hard for workers to enter the jobs complemented by automation are wage increases for that job likely to be higher. Autor writes:
"[T]he elasticity of labor supply can mitigate wage gains. If the complementary tasks that construction workers or relationship bankers supply are abundantly available elsewhere in the economy, then it is plausible that a flood of new workers will temper any wage gains that would emanate from complementarities between automation and human labor input. While these kinds of supply effects will probably not offset productivity-driven wage gains fully, one can find extreme examples: Hsieh and Moretti (2003) document that new entry into the real estate broker occupation in response to rising house prices fully offsets average wage gains that would otherwise have occurred." 
Finally, the capital investments that lead to automation and improved robots tend to increase output, and the incomes for those who are producing these additional outputs will also rise. As incomes rise, people demand additional goods and services. The particular goods that are demanded as income rises will also affect how automation and labor markets interact. Here's Autor:
[T]he output elasticity of demand combined with income elasticity of demand can either dampen or amplify the gains from automation. In the case of agricultural products over the long run, spectacular productivity improvements have been accompanied by declines in the share of household income spent on food. In other cases, such as the health care sector, improvements in technology have led to ever-larger shares of income being spent on health. Even if ... the sector shrinks as productivity rises—this does not imply that aggregate demand falls as technology advances; clearly, the surplus income can be spent elsewhere. As passenger cars displaced equestrian travel and the myriad occupations that supported it in the 1920s, the roadside motel and fast food industries rose up to serve the “motoring public” ( Jackson 1993). Rising income may also spur demand for activities that have nothing to do with the technological vanguard. Production of restaurant meals, cleaning services, haircare, and personal fitness is neither strongly complemented nor substituted by current technologies; these sectors are “technologically lagging” in Baumol’s (1967) phrase. But demand for these goods appears strongly income-elastic, so that rising productivity in technologically leading sectors may boost employment nevertheless in these activities. 
Autor argues that developments in information technology have dramatically reshaped labor markets in recent decades. But the primary effect is not through fewer jobs: after all, the US unemployment rate was below 5.5% for nearly four full years before the Great Recession took hold in 2008, and has now again fallen into that range. Instead, Autor argues that the effects of technological advances can be seen in job "polarization." In this phenomenon, which seems to arise not only in the US but also across many European economies, many previously middle-skill jobs including factory jobs and office jobs have been replaced by technology and become less important relative to overall employment. At the same time, low-skill manual jobs that can't easily be replaced by automation have tended to grow in number, but not in wages--because it is fairly easy for workers to shift into these jobs. And high-skill jobs that are complemented by technology have increased relative to the overall workforce as well as in wages paid.

Interestingly, Autor argues that this job polarization is not likely to persist in the future, although information technology will shape what are the middle-class jobs of the future. He writes:

My own prediction is that employment polarization will not continue indefinitely (as argued in Autor 2013). While some of the tasks in many current middle-skill jobs are susceptible to automation, many middle-skill jobs will continue to demand a mixture of tasks from across the skill spectrum. For example, medical support occupations—radiology technicians, phlebotomists, nurse technicians, and others—are a significant and rapidly growing category of relatively well-remunerated, middle-skill employment. Most of these occupations require mastery of “middle-skill” mathematics, life sciences, and analytical reasoning. They typically require at least two years of postsecondary vocational training, and in some cases a four-year college degree or more. This broad description also fits numerous skilled trade and repair occupations, including plumbers, builders, electricians, heating/ventilating/air-conditioning installers, and automotive technicians. It also fits a number of modern clerical occupations that provide coordination and decision-making functions, rather than simply typing and filing, like a number of jobs in marketing. There are also cases where technology is enabling workers with less esoteric technical mastery to perform additional tasks: for example, the nurse practitioner occupation that increasingly performs diagnosing and prescribing tasks in lieu of physicians. 
I expect that a significant stratum of middle-skill jobs combining specific vocational skills with foundational middle-skills levels of literacy, numeracy, adaptability, problem solving, and common sense will persist in coming decades. My conjecture is that many of the tasks currently bundled into these jobs cannot readily be unbundled—with machines performing the middle-skill tasks and workers performing only a low-skill residual—without a substantial drop in quality. This argument suggests that many of the middle-skill jobs that persist in the future will combine routine technical tasks with the set of nonroutine tasks in which workers hold comparative advantage: interpersonal interaction, flexibility, adaptability, and problem solving. In general, these same demands for interaction frequently privilege face-to-face interactions over remote performance, meaning that these same middle-skill occupations may have relatively low susceptibility to offshoring. Lawrence Katz memorably titles workers who virtuously combine technical and interpersonal tasks as “the new artisans” (see Friedman 2010), and Holzer (2015) documents that “new middle skill jobs” are in fact growing rapidly, even as traditional production and clerical occupations contract.
The same issue of the JEP includes two other articles on automation and labor markets. Joel Mokyr, Chris Vickers, and Nicolas L. Ziebarth look at the history of these concerns among economists over the last two centuries in "The History of Technological Anxiety and the Future of Economic Growth: Is This Time Different?"  From the abstract:
"Anxieties over technology can take on several forms, and we focus on three of the most prominent concerns. First, there is the concern that technological progress will cause widespread substitution of machines for labor, which in turn could lead to technological unemployment and a further increase in inequality in the short run, even if the long-run effects are beneficial. Second, there has been anxiety over the moral implications of technological process for human welfare, broadly defined. While, during the Industrial Revolution, the worry was about the dehumanizing effects of work, in modern times, perhaps the greater fear is a world where the elimination of work itself is the source of dehumanization. A third concern cuts in the opposite direction, suggesting that the epoch of major technological progress is behind us. Understanding the history of technological anxiety provides perspective on whether this time is truly different. We consider the role of these three anxieties among economists, primarily focusing on the historical period from the late 18th to the early 20th century, and then compare the historical and current manifestations of these three concerns."
Finally, Gill A. Pratt write: "Is a Cambrian Explosion Coming for Robotics?" Pratt is not an economist, but rather an expert in robotics who until recently was a program manager at the Defense Advanced Research Projects Agency, where one of his tasks was oversight of the DARPA Robotics Challenge. Pratt lays out a number of reasons why a dramatic increase in robot capabilities is likely to be near. Here are two of the most important:
Two newly blossoming technologies—“Cloud Robotics” and “Deep Learning”—could leverage these base technologies in a virtuous cycle of explosive growth. In Cloud Robotics— a term coined by James Kuffner (2010)—every robot learns from the experiences of all robots, which leads to rapid growth of robot competence, particularly as the number of robots grows. Deep Learning algorithms are a method for robots to learn and generalize their associations based on very large (and often cloud-based) “training sets” that typically include millions of examples. 
Pratt persuades me, at least, that "[i]t is reasonable to assume that robots will in the not-too-distant future be able perform any associative memory problem at human levels." This includes a wide array of jobs that include sensing and recognizing what is in a given environment, considering a list of tasks that should be done, and then acting autonomously to complete those tasks. But as Pratt also notes: "The human brain does much more than store a very large number of associations and access useful memories quickly. It also transforms sensory and other information into generalizable representations invariant to unimportant changes, stores episodic memories, and generalizes learned examples into understanding. The key problems in robot capability yet to be solved are those of generalizable  knowledge representation and of cognition based on that representation."

One might say that the "new artisans" of the middle class in Autor's description will be jobs that involve this kinds of "generalizable  knowledge representation and of cognition based on that representation." Autor writes: Machine-learning algorithms may have fundamental problems with reasoning about “purposiveness” and intended uses, even given an arbitrarily large training database of images ..." Pratt suggests that the changes in robotics may come so rapidly that human worker and labor market institutions and workers will have a hard time adjusting quickly enough.