For most of our history Americans have supported increased productivity, even if it led to economic disruption and worker displacement. Unfortunately, over the last decade that support has weakened. Productivity gains are now portrayed as a tool for greedy corporations to jack up their profits at the expense of workers. Moreover, workers are now seen as so fragile that all automation must be feared. So, rather than support productivity, the dominant narrative favors redistribution and protection from change.
The emergence of this new “productivity doesn’t benefit workers” narrative matters because government policies in a wide array of areas — from taxation to science to regulation — affect productivity growth for good or ill. And if the view is that productivity gains are no longer widely shared and that society should oppose job-displacing automation, support for growth and progress will wither and be replaced by a focus on redistribution and stasis.
Let’s start with the claim that productivity growth no longer benefits workers. Thomas A. Kochan and Barbara Dyer write: “Productivity growth — where workers produce increasing amounts of goods or services per work hour — no longer drives pay.” American Prospect editor David Dayen agrees: “Increased productivity and tight labor markets should lead to higher wages. But in the U.S., wages for the typical worker have been flat for four decades.”
This gloomy view can trace its source to work by economists Thomas Piketty and Emmanuel Saez, who purported to show in 2014 that U.S. median real-income growth declined by 8 percent between 1979 and 2014, a period when labor productivity doubled. If this were true, people would be well justified in turning their back on productivity.
But as labor economist Stephen Rose notes in a study for the Urban Institute, the original Piketty and Saez work (2013) and its 2014 update suffer from a number of serious methodological problems, including looking at individual tax files rather than those of households and not including the value of rapidly growing employer-sponsored health care. It turns out that measuring real-income growth is fraught with difficulties and chock-full of assumptions. As Rose notes, different studies get different answers because they use different definitions of income, adjustments for inflation, and units of analysis. However, there is a methodology developed by an international group of experts to measure the growth of income (the Canberra method), and while no U.S. study has exactly met this standard, some have come close. When Rose adjusts for these methodological differences he estimates that real median income grew 43 percent over this period. The Congressional Budget Office’s estimate is even higher, 51 percent. Even Piketty and Saez (with co-author Gabriel Zucman) found in 2018 that using a more appropriate methodology showed that median income grew by 33 percent. To be sure, this does not mean that income inequality did not grow — it did — but 33 to 51 percent growth in worker incomes is a far cry from an 8 percent decline.
One would expect the current narrative on this issue to reflect updates from Piketty and Saez, and numerous other studies, that have shown that the median worker still benefits from productivity growth. As John Maynard Keynes famously stated after someone accused him of changing his view, “When the facts change, I change my mind.” Apparently we are not all Keynesians today. Consider David Leonhardt, a New York Times economic columnist, who this year wrote that before the 1970s workers enjoyed consistently rising wages but that now “profits have soared at the expense of worker pay.” He adds: “The wealth of the median family today is lower than two decades ago.” Leonhardt is not an outlier. Despite solid evidence to the contrary, many people have made up their mind that productivity doesn’t benefit most Americans and have turned to redistribution as the only valid economic-policy goal.
If productivity still benefits the average worker, what should government do to boost productivity growth, particularly in ways that benefit the bottom half of earners?
Let’s start with what probably won’t work. Doing nothing and leaving it only up to firms, as some on the right would counsel, won’t work. To be sure, preventing regulation from squashing productivity-enhancing innovations is important. But as I have written, when it comes to productivity a host of market failures require smart government-policy responses. For example, when firms buy new machinery and equipment they don’t capture all of the benefits; some “spill over” to other firms and society. Consequently, absent a tax code that incentivizes investment in new capital equipment and software, firms will invest less than is societally optimal. Likewise, federal funding for scientific and engineering research is a public good that can boost productivity.
But focusing only on redistributing income — as many on the left would counsel, in the mistaken belief that there is nothing government can do to boost productivity — will also not work.
Neo-Keynesian economist Frank Levy argues: “We cannot legislate the rate of productivity growth. . . . That is why equalizing institutions are so important.” Such views are why so many on the left default to a redistribution strategy, grounded in policies such as reducing taxes paid by the bottom 80 percent of Americans while supporting more overtime pay, profit-sharing, unionization, and subsidization of goods and services (from housing to infrastructure to education). These Huey Long Democrats want to govern on a share-the-wealth agenda, not a grow-the-pie agenda.
It’s time to break this stalemate and forge a bipartisan consensus that holds that productivity is the most important goal of economic policy and that the federal government should do more to spur productivity growth. This should include more-generous tax incentives for investing in research and development, capital equipment, and worker training; as well as significant funding increases for research and development focused on areas that will likely have high productivity payoffs (e.g., material sciences, artificial intelligence, and robotics).
But while America desperately needs faster productivity growth, if we want faster wage growth, especially for low- and moderate-income workers, those who point to the growth of inequality are not completely wrong. Policymakers do need to pay more attention to improving the earnings of these workers. But automation, rather than being an obstacle, is a key to achieving that goal. Here’s why.
Many agree that the emerging wave of technological innovation in fields including robotics and artificial intelligence holds the promise of boosting productivity. But this next wave will likely have differential impacts on occupations. Estimates of the impact of these technologies on occupations show that there is a significant negative correlation between the average wage of an occupation and its risk of automation. In other words, this next wave of innovation is more likely to automate low-wage jobs than middle- and upper-wage jobs. If true, this means that there will be a larger proportion of middle- and upper-wage jobs as lower-wage jobs are automated at higher rates and therefore employ fewer people. This would have the beneficial result of there being fewer lower-paying jobs and more better-paying jobs — a plus for many workers now employed in occupations whose productivity and wages remain low and stagnant. This means that rather than protect lower-wage cashiers and toll-takers from automation we should encourage more and faster automation so that we get more middle-wage carpenters and sales reps.
But while automation is likely to be a progressive force, policy can and should do more to ensure that lower-wage workers gain, including increasing the federal minimum wage (which will have the added benefit of spurring more automation of low-wage jobs), shifting support away from four-year colleges that serve higher-income households and toward two-year technical schools and other kinds of occupational training, and expanding apprenticeship programs nationally.
Finally, what about job loss? Shouldn’t we be worried about individuals’ losing their jobs, or even about the massive unemployment that some have warned of? No, no, a thousand times no! No matter how many times a purported expert claims we are facing an epochal technology revolution that will destroy tens of millions of jobs and leave large swathes of human workers permanently unemployed, it still isn’t true. The simple reason is that when companies invest in robots, they usually cut costs and pass a significant share of those savings to consumers in the form of lower prices (with some going to workers as higher wages and to shareholders in the form of higher profits). These savings are not buried, but rather are recycled. As they are spent or invested, they create additional jobs.
While we won’t run out of work, a faster pace of technological innovation and productivity will likely lead to increased labor-market churn as some occupations decline (e.g., truck-driving) and others grow (e.g., health-care work). This means that the challenge for policymakers is not to slow down change or enact a wasteful universal-basic-income scheme, but to put in place a more robust system to help workers make employment transitions.
Still, at the end of the day, higher productivity is the worker’s friend.