ARTICLE CONTENTS INTRODUCTION 257 I. "THIS TIME IS DIFFERENT": THE NEXT WAVE OF AUTOMATION AND WHY WE SHOULD WORRY 263 A. What Is New About the New Wave of Automating Technologies? 264 B. Estimating Automatability and the Pace of Job Destruction 267 C. Estimating New Job Creation Through Labor Complementarity, 271 Consumer Surplus, and New Products and Services D. Should We Welcome or Worry About a Future with Much Less 275 Work? E. Alternative Futures and Why the Pace of Automation Matters 280 II. HOW AUTOMATION ALTERS THE LANDSCAPE OF WORK AND CONFOUNDS 283 EFFORTS TO SHORE UP THE FORTRESS OF EMPLOYMENT A. Fissuring and the Growth of Outsourcing, Offshoring, and 283 Platform Work B. Three Common Causes of Automation and Fissuring: 286 Technology, Heightened Competition, and the Costs of Employing People C. Why Automation Is Different 291 D. How Automation Confounds Prevailing Approaches to the 295 Regulation of Work III. A WAY FORWARD IN THE FACE OF UNCERTAINTY 301 A. Framing a Strategy in the Face of Uncertainty 301 B. Employee Entitlements that Are Inextricable from Employer 303 Mandates C. Employee Entitlements that Can and Should Be Detached from 305 Employer Mandates D. Paying for Entitlements that Are Detached from Employment 315 E. A Broader Case for Reconstructing the Platform for Social 319 Rights and Benefits F. Some Objections and Alternatives 320 CONCLUSION 325 INTRODUCTION
Three major threats to American jobs have grabbed headlines in recent years. One is the migration of manufacturing jobs to China, as perhaps best exemplified by Foxconn, the Taiwanese firm that employs over a million Chinese workers in the production of iPhones, iPads, and other consumer electronics. To labor-law cognoscenti, the outsourcing of manufacturing to China, and the feared "race to the bottom" in labor standards, is mostly yesterday's news. Since 2015, they have been more preoccupied with a second development--the rise of platform-based "gig" work in place of real jobs, epitomized by Uber's treatment of its drivers as independent contractors rather than employees.
Yet both of these threats to American jobs and workers arguably pale beside the threat of automation. If Uber has its way, its drivers will soon go the way of lamplighters, replaced by self-driving vehicles. (1) And if Foxconn is representative, then Chinese factory jobs are also at risk: by 2016, Foxconn had replaced 60,000 production workers with robots and was planning to replace most of the others within several years. (2)
For some observers, Uber's autonomous vehicles and Foxconn's robots are harbingers of a jobless future, as machines and algorithms threaten to put vast swaths of the labor force in the United States and worldwide out of work or into desperate competition for the jobs that remain. (3) These commentators describe an exponential growth of technologies that replicate or surpass humans in an ever-wider range of tasks. (4) Putting aside the more fantastical predictions about artificial intelligence (AI) dominating or even devouring its human creators, (5) the prospects for job destruction are eye-opening. Robotic and digital production of goods and services, coupled with advances in AI and machine learning, is poised to take over both routine or repetitive tasks and some more advanced tasks. In one much-cited reckoning, nearly half of the jobs in the current economy are at risk. (6) Although some new jobs are readily foreseeable--especially skilled jobs working with technology--no large new sectors or industries yet visible on the horizon appear likely to absorb the multitudes of human workers who might be displaced. Within this camp, predictions range from a tsunami of job losses to a more manageable rising tide.
For other observers, the real threat from automation is not a net loss of jobs but growing polarization of the labor market. (7) These observers note that automation generates large productivity gains and profits for some, while destroying many decent midlevel jobs. They predict a growing economic chasm between those who create or own the new technology, or whose high-end skills are complemented by that technology, and most workers who are stuck competing for the less-skilled but still human jobs that remain. In this scenario, labor shortages in some skilled-job categories will coexist with labor surpluses and downward wage pressure outside those categories.
The prospect of large net job losses and sharper income polarization has generated new interest in some old ideas, such as a universal basic income (UBI), (8) reduction of maximum working hours, (9) and public investments in job creation. (10) Before taking up responses to feared job losses, however, we must turn to the other side in the debate over the impact of automation on jobs--that is, those who discount the claim that "this time is different," and would predict the future of automation from its past.
For many economists, the current wave of automation anxiety amounts to misguided scaremongering by modern-day Luddites. (11) After all, they point out, the prediction that automation will supplant human labor on a massive scale has recurred in both Utopian and dystopian flavors throughout the history of industrialization. (12) Futurists of the past have predicted that mass automation will usher in an era of human liberation from toil, or that it will immiserate all but the fortunate few who own or create the machines. (13) Time and again, however, the economy has defied such predictions. For centuries, automation has been destroying some jobs while creating other jobs--usually better paid and less grueling--and driving economic growth and prosperity. In short, the history of automation's impact on the labor market has been one of "creative destruction," a mantra to which many economists adhere today. (14)
The debate over automation and jobs rages within and between the fields of economics and information technology. (15) Among technology experts canvassed in 2014, nearly half said they believed that "networked, automated, [AI] applications and robotic devices [will] have displaced more jobs than they have created by 2025." (16) By contrast, a 2014 survey of academic economists found wide agreement that "[a]dvancing automation has not historically reduced employment in the United States" (17) (which sidesteps the claim that "this time is different"). Yet one-third of them agreed that automation was "a central reason why median wages have been stagnant in the US over the past decade, despite rising productivity." (18)
The debate over the impact of automation also divides economists on the left who are chiefly concerned with the well-being of ordinary workers. For example, economists at the left-leaning Economic Policy Institute (19) reject both the job-killing story and the income-polarization story about automation. (20) For them, the very real problem of wage stagnation stems not from technology but from globalization of trade and finance, declining union strength, and lagging enforcement of labor standards. By contrast, labor-friendly Harvard economist Richard Freeman believes that this time really is different, and that technology has already contributed to a historic shift in the distribution of income over the past two decades "toward robots/capital and against labor." (21) Freeman expects technology to affect wages more than employment: "The 'iron law' of the effect of robots on pay is that increased substitutability with human skills puts downward pressure on the wages of persons doing competing tasks--a pressure likely to grow in the future as technology improves the competence of robots and lowers their cost." (22)
This vigorous debate is not likely to be resolved within the legal academy; yet legal scholars, and especially scholars of labor and employment law, cannot afford to ignore it. In the United States, as in many developed countries, the employment relationship has long served as the platform for delivering a plethora of rights, protections, and benefits that enable people to live a decent life. Automation-related job losses threaten to further destabilize that already-precarious platform. It is thus imperative that we try to understand how automation will affect the law of work--and how the law of work will affect the future trajectory of automation. This Article aims to advance that understanding and to propose an optimal strategy for the law of work in the face of uncertainty about automation's impact on the labor market.
An optimal response to automation must take into account three other labor-market problems that are far less controversial among those who focus on workers' well-being: rising income inequality, the erosion of labor standards for low-wage workers, and the role of "Assuring"--the substitution of outside contractors for employees--in both trends. (23) If we can find ways to meet the still-contested challenge of automation that will also address--or at least not exacerbate--the more certain challenges of Assuring, inequality, and deteriorating labor standards, then we will be on solid ground.
It is crucial to recognize at the outset that the existing law of work adds to the costs of employing human labor, and that new and improved worker benefits and higher labor standards would further increase those costs. As such, the law of work contributes both to firms' flight from direct employment through Assuring and to their substitution of machines for human workers. In response to Assuring, many scholars and advocates seek to shore up what I call the "fortress of employment" by extending Arms' legal responsibility for workers in their supply chain, in most cases by expanding the definitions of "employee" and "employer." (24) But automation confounds that strategy by offering Arms a more complete exit from the costs, risks, and hassles associated with human labor. Extending Arms' responsibility for workers in their supply chain...