The Agenda

Technological Unemployment vs. Technology-Driven Inequality

Scott Winship argues that fears of technological unemployment are overblown:

[P]eople seem to resist the idea that if technology reduces demand for labor by a quarter, that might translate into everyone working 25 percent less rather than unemployment rising by one-fourth. The late economic historian Robert Fogel predicted that the increase in leisure between 1995 and 2040 would exceed the gain Americans saw in the 115 years before 1995.

Will Fogel be proven right? I certainly don’t know. What I’m pretty sure of, however, is that in 2040 we will not look back at 2013 and think, “We’ve made a huge mistake,” ruing the day we failed to listen to those afflicted with technophobia. Technological progress is not a trick we have played on ourselves to throw ourselves into poverty. It is a means for fulfilling our material wants and needs and continues only to the extent that it does so.

Yet Scott also cites a new report by Frank Levy and Richard Murnane, which warns that mechanization might lead to stagnant or declining wages for less-skilled workers, a report I will discuss at greater length in my next column. While I agree with Scott’s basic view that the emergence of increasingly sophisticated machines that can automate a wide range of tasks now performed by human workers won’t in itself lead to unemployment or underemployment, the real question is how technological change is likely to interact with labor market policy. That is, if it becomes increasingly expensive for firms hire low-wage employees and if it becomes increasingly less attractive for less-skilled workers to take part in the labor force, as compensation declines and non-participation in the labor force becomes more tolerable for a variety of reasons (a decrease in the cost of the goods and services needed to maintain a dignified life is one possibility, an increase in means-tested transfers is another), it isn’t too difficult to imagine that technological development will co-exist with a shrinking labor force and declining work hours. This isn’t necessarily troubling provided declining work hours and labor force participation reflect changing preferences, hence Scott’s reference to leisure time. But if this increase in leisure will does indeed occur, it will most likely be distributed unevenly. Mark Aguiar and Erik Hurst famously observed that we have seen an increase in “leisure inequality“:

This paper examines the changing allocation of time within the United States that has occurred between 1965 and 2003-2005. We find that the time individuals have allocated to leisure has increased in the U.S. for both men and women during this period, with almost the entire gain occurring prior to 1985. We also find that post 1985 there has been a substantial increase in leisure inequality, particularly for men. Over the last 20 years, less educated men increased the time they allocated to leisure while more educated men recorded a decrease in leisure time. While the relative decline in the employment rate of less educated men is important, trends in employment status explain less than half of the increase in the leisure gap.

The Economist drew on this finding in 2006 to make the following point:

Does this mean that income inequality increased only because the poor and middle class work less? Not necessarily; the causation may run the other way. Leisure is a good like any other, and has a positive income effect; as people have more income, they want to consume more of it. Since low-paying jobs tend to be more unpleasant than higher paying jobs (journalism excepted), the income effect may be stronger at the bottom of the wage distribution. Since all income groups have seen real incomes rise over the last forty years, those in the bottom half may simply have decided that since they get paid more, they don’t have to work as much.

On the other hand, the substitution effect predicts that as changes in technology have disproportionably increased the salaries of skilled workers, the better paid are likely to have worked more. People with higher wages have a greater incentive to forgo leisure time. The less skilled face a smaller opportunity cost for their leisure. So they will work less.

If the prestige associated with less-skilled, low-wage work declines as mechanization proceeds apace, the income effect at the bottom of the wage distribution might grow stronger still. And if mechanization exacerbates the polarization of the labor market, as seems plausible, the opportunity cost of leisure at the bottom of the wage distribution will shrink in relative terms. Scott is right that this scenario would not represent a straightforward case of machines throwing people out of work — rather, it would represent an intensification of a number of existing trends that have tended to exacerbate wage and income dispersion. Some left-liberals are resistant to the notion that technology has been a significant driver of increasing inequality, instead emphasizing the policy environment, e.g., the prevalence of collective bargaining, etc. Dean Baker writes:

In this story the upward redistribution of income was a conscious policy by those in power. This story points to a number of different policies that had the effect of redistributing income upward. For example, exposing manufacturing workers to direct competition with low-paid workers in the developing world, while protecting highly educated professionals (e.g. doctors and lawyers), would be expected to lower the wages of both manufacturing workers and the large number of workers who will compete for jobs with displaced manufacturing workers.

Central banks that target low inflation even at the cost of higher unemployment will also increase inequality. When a central bank like the Fed raises interest rates to slow the economy and reduce inflationary pressures, it is factory workers and retail clerks who lose their jobs, not doctors and lawyers. Even an economist can figure out that this will depress the wages of the former to benefit the latter.

Baker’s points are reasonable, yet it does seem as though a technology-driven erosion of mid-skill employment has contributed to pre-tax-and-transfer wage and income dispersion. Reducing the protections afforded credentialed professionals might have mitigated this tendency to some extent, but it seems implausible that it would have halted or reversed the deterioration of the labor market position of less-skilled workers. Highly educated professionals in other domains — management consultants, corporate executives, coders — have also seen their relative position improve, yet they don’t have organizations with the clout of the AMA and the ABA working to build barriers to entry on their behalf, which is not to say that there haven’t been attempts. 


The Latest