Unless we send humanity on a permanent paid vacation, the future could get very bleak
Newspapers are full of bewildered economists scratching their head at the emerging jobloss recovery. The right reassures us that job growth is right around the corner, although it wouldn’t hurt to have more tax cuts, deregulation, freer trade and lower minimum wages. Liberals counter that we can cut unemployment with more job retraining, free higher education, more protectionism, more demand-side tax stimulus and nonmilitary public sector investments.
The problem is that none of these policies can reverse the emerging structural unemployment resulting from automation and globalization. The economists and policy makers are still wedded to the idea of full employment, or at least to a labor market with a job opportunity for everyone who wants one.
Without a clear strategic goal of a humanity freed from work through the gradual expansion of automation and the social wage, all policies short of Luddite bans on new technology will have disappointing and perverse effects. If liberals and the left do not re-embrace the end of work and the need to give everyone income as a right of citizenship, unconnected to employment, they will help usher in a much bleaker future of growing class polarization and widespread immiseration. If libertarians and the right do not adapt to the need to provide universal income in a jobless future they may help bring about a populist backlash against free trade and industrial modernization.
In other words, it’s time to make a choice: Luddism, barbarism or a universal basic income guarantee.
The compensation thesis
Techno-utopians have predicted the end of work through automation for a century or more, and yet employment has continuously expanded. This has made the compensation thesis, the idea that technological change creates as many jobs as it destroys, seem like common sense. As the need for farm labor shrank, mines, factories and mercantile trades grew. In the late 20th century, industrial jobs were replaced with service sector employment. Supposedly, whatever jobs are lost to robotics, AI and nanotech in the 21st century will also then be replaced by new jobs.
- Compensation via new machines and products. Each new set of machines requires new occupations to build and service them, and make possible the production of new goods and services.
- Compensation via decrease in prices. Each technological revolution reduces the cost of inputs and goods, stimulating greater demand, and therefore creating more employment.
- Compensation via new investments. Technological modernization increases the profit margins of the owning class, who then invest in the creation of more employment.
- Compensation via decrease in wages. If wages are allowed to find their equilibrium point, all unemployed workers can find new jobs at lower wages.
- Compensation via increase in wages. And, directly contrary to the prior model, Keynesian social policies and Fordist labor practices distribute some of the increased profitability to workers as wages, with a consequent demand stimulus on the economy and employment.
Empirically, the evidence for the compensation thesis up through 2000 is mixed. On the one hand, the proportion of the population participating in the labor market has steadily increased in the industrialized world. On the other hand, the number of hours worked per year has steadily declined in most industrialized countries from about 3,000 hours a year at the turn of the 20th century to about 1,500 to 1,900 hours per year at the turn of 21st century. So whatever gains in employment that technological change has facilitated in the last century appear to have been happily accompanied by a decline in total average hours worked.
The US is the biggest exception, however, since we appear to be working more hours over the past 30 years, due in part to a stronger preference for income and consumption over additional leisure. This is taken as another confirmation of the additional products part of the compensation thesis. We could just work a couple of hours a day to attain the standard of living of our ancestors, but we all in fact believe that we have to work longer hours each year to provide ourselves more of the novel goods and services being produced by the supersize-it society. Americans have also demanded and won fewer workplace and social wage benefits, such as universal health insurance, a month of vacation or fully subsidized university education. So the US has been able to sustain a lower unemployment rate, and stay competitive with robots and the developing world a little longer, largely because of the growth of low wage service jobs with few benefits.
But now we are facing the same structural unemployment that was supposedly due to overregulation of the European labor market and their generous safety net, their so-called “Eurosclerosis.”
Robots for people
All of these compensatory mechanisms fail to take account of the possibility that capital investments in technology will become increasingly more productive and profitable as substitutes for investments in labor. The number of workers required to make the new machines and products will then be fewer than the number of workers displaced each year regardless of increased demand or the accumulation of investment capital. Any increase in demand resulting from falls in prices, declines in the cost of displaced labor or increases in the wages of the remaining workers will only stimulate further capital investment in labor-saving machinery. Any effort to stimulate the demand in the economy, Fordist or Keynesian, by sharing the benefits of increased profits and productivity with workers through increased wages or benefits, or shorter work weeks, will only increase the cost of labor and the profitability of investing in labor-saving machines.
This is part of the story behind the disappointing experience with shorter work weeks as a way to redistribute employment. In the end it is more expensive to hire 10 workers to provide 350 person hours of labor, rather than nine, even if you pay both teams the same wage per person hour. So laws that reduce the exploitability of human labor, such as reductions in the work week, make investments in automation all the more comparatively profitable, even when they were supposed to redistribute employment. In an automating economy, redistributing employment through shorter work weeks will result in less employment, not more.
Nor is this substitution of capital for labor, machines for people, only occurring in the expensive labor markets of the developed world. Labor-saving technology also appears to be eliminating agricultural and industrial jobs in the developing world. Former US Labor Secretary Robert Reich recently noted that between 1995 and 2002, despite dramatic economic growth or perhaps because of dramatic economic growth, China lost 15% of its manufacturing jobs. During the same period, Brazil lost 20% of its industrial jobs, the US lost 11% and Japan lost 16%.
The compensatory theorists argue that new employment will soon open up in fields that are one step ahead of automation. This was of course the solution to globalization’s job losses offered by people such as Robert Reich in the mid 1990s: Retrain displaced assembly-line workers to be computer programmers. Now, as computer programming slips away to the developing world and one intellectual or managerial occupation after the other is routinized into an expert system, the argument seems less convincing.
The brief reign of the knowledge worker could soon be over. What if there are no more types of labor accessible to humans that artificial intelligence and robotics cannot do more cheaply and efficiently, with higher quality? Marshall Brain makes a very convincing case that there are very few occupations ultimately exempt from automation. Even Reich recently opined, “The issue we really ought to be talking is what jobs Americans, and everyone else, will be able to find when machines are able to do just about everything.”
We won’t prefer humans
An optimistic thought frequently offered at this point in the argument is that, even if machines can do everything better than humans, humans will still prefer some things to be done by other humans rather than machines. But I think the growth in the percentage of American meals provided through highly automated fast food franchises puts the lie to this idea. Apparently many people find fast food more appealing on the grounds of at least convenience and cost, and perhaps also quality, compared to meals they could make themselves in the kitchen or consume in a more traditional, labor-intensive restaurant. Similarly, the wealthy might spur a strong demand for performance art involving millions, for vegetables grown labor-intensively in window boxes or for other personal services that might take up the slack of structural unemployment. But it doesn’t seem likely.
The compensatory mechanism of giving increased profits from technological advances to consumers to stimulate demand will undoubtedly be tried in a variety of ways, starting with expanding public employment. But what are those occupations that the state can hire displaced workers for? The largest sectors of public employment are administrative-managerial tasks, mail handling and the military, all of which are being downsized by information technology and automation.
We could also provide public sector subsidies to make hiring the unemployed more profitable and attractive. For instance, in the US we could expand national health insurance to cover all Americans, rationalizing and generally reducing the burden of health care insurance on employers. Then we could expand our nursing care benefits to give employment to more nurses, and income to the women who currently take care of their sick relatives for free in the informal economy. But that assumes that the growing elderly population will need more nursing, when in fact seniors have a rapidly declining rate of disability. Even if we do only figure out how to make seniors live to 120, without making them healthy and able-bodied in the process, nursing and health care are also undergoing a robotics and information technology revolution permitting sick seniors to be monitored and cared for in their homes, and when necessary institutionally, by fewer and fewer caregivers.
Another laudable strategy to expand the economy would be some form of global Keynesianism, efforts to stimulate global economic demand by distributing the benefits and wealth of technological innovation to the majority of the world’s peoples who still live at substantially below the Western standard of living. Setting aside the political and ecological constraints, dramatic expansions in the global market might delay the global decline in employment, but would not stop it, as the declines in Chinese manufacturing demonstrate. If a global Marshall Plan allowed every Bangladeshi to buy a car, the most competitive cars would roll off increasingly automated assembly lines, providing only a temporary increase in global employment.
A way to avoid structural unemployment rarely considered outside of transhumanist circles is upgrading human beings themselves. Using the converging technologies of nanotechnology, genetic engineering, psychopharmaceuticals and cognitive science it might be possible to upgrade workers to keep them competitive with machines. But computers are doubling in memory and speed every two years. It is pretty unlikely that human beings will ever be able to keep up with the wave of automation without merging with it completely, which is a whole different essay.
Luddism, barbarism or basic income
In the end, business would rather invest first in people in the developing world and then infotech and robots rather than expensive human workers in the developed world. Because of this, wages in the developed countries will fall in competition with the lowest wage human competitors around the world, and then in competition with the increasingly inexpensive robots and expert systems. Jobs will disappear, wages will fall and we will face three choices: Luddism, barbarism or basic income.
The Luddites might win a temporary battle here and there, delaying one or the other labor-saving device or innovation. But in the end they will lose, and the technologies will come. Then the question will be what happens to the displaced, and to the economy.
Without an expanded social wage (benefits and income from the government) in general, and a guaranteed basic income in particular, we face widespread immiseration, economic contraction and polarization between the wealthy, the shrinking working class and the structurally redundant.
Or we can avoid this bleak future by re-embracing the techno-utopian vision and consciously striving to shrink working life by reducing the work week, mandating paid vacations, raising the minimum wage, improving workplace protections and providing health insurance and a basic income as a right of citizenship. All these policies will make human labor more expensive and investments in automation increasingly attractive. Employment will shrink, social wealth will grow and be shared more equally, and we can start rejoicing instead of despairing about the end of work. As Marshall Brain says, humanity can go on a permanent vacation.