In this post, I’ll outline the past economic trends that provide the context for understanding our current and future situation. In summary, the trends are:
- A growing gap between productivity and worker compensation
- A growing gap between the wages of the most and least educated worker
- A decrease in available jobs for middle skilled workers
- A flattening of wages for all but the highest skilled workers
There are many forces driving these changes, but accelerating technology is, in my opinion, a dominant force. I plan to talk about the people who have been the thought leaders in analyzing recent trends, so let’s begin by providing a cast of characters. Daron Acemoglu and David Autor are MIT economists who (among others) have provided many important studies, backed by statistical data, that are central to these arguments. Martin Ford is a technologist, who wrote The Lights in the Tunnel (2009), which describes how the accelerating pace of technological change will disrupt the economy. Erik Brynjolfsson and Andrew McAfee, both from MIT, wrote Race Against The Machine (2011), and continue to write about the impact of technology on business. As the story unfolds, you’ll see that there are other voices as well, but these are the people at the tip of the arrow.
In this series of posts, I will give a reader’s digest version of these emerging arguments, and relate these discussions to my experience. I’ll show why I feel that the job market will remain difficult for many years, and I’ll talk about what people can do to cope between now and the day when the technologists and economists sort all of this out.
In 1984, I began research that resulted in a paper that was published in June 1990. It was about how a robot could automatically grind weld beads flat, which was important to the auto industry. At that time, people did this job, and it was dangerous and dirty. The photo at the left shows the prototype created by three graduate students and two advisers. My contribution was an early attempt to give the robot dexterity: specifically, to vary the force applied to the weld depending of the depth of cut required. Another student created the vision system and another modeled the dynamics of grinding. An early personal computer, the IBM PC AT, did all the processing, powered by an Intel 80286, that is more than 60,000 times slower than a typical desktop computer today. Back in 1990, we collectively believed that the nature of our work was simple: automation increased economic productivity, which increased jobs, which increased prosperity. Little did we know that we were contributing to a major economic shift that would not be fully understood until the 21st century.
The Raging Debate: Are we at the beginning of an epic change in human history?
Agreement about what is going on in our economy is elusive. Many economists still believe that what we experience today is just another chapter in a long series of technological disruptions, from the plow to the steam engine to electric power distribution. Consider that this 2011 report about the causes of long-term unemployment, from the Federal Reserve Bank of Richmond, does not talk about technology at all. Technological disruption of labor has emerged as a much discussed idea, but there are significant groups of economists that dismiss it. There is a raging debate going on. Is history repeating itself, or are we at the beginning of an epic change in human history?From 1940’s until 1970’s economic productivity and worker compensation moved hand in hand: as productivity went up, workers benefited in proportion. In the 1970’s this relationship broke down and the gap between productivity and compensation has grown ever since. The figure Productivity vs. Worker Earnings illustrates the concept. Essentially, our economy generates more wealth each year as productivity increases, but less of that wealth is distributed to workers in the form of wages. Instead, more of that wealth is distributed to the people who own the companies: the stockholders. We all benefit from lower costs for goods, but many struggle with stagnant wages.
This much is not in dispute, but the reasons for the change have been debated for a long time. Some economists, such as Mark Perry, from the American Enterprise Institute, attribute the gap to market forces. He expects that it will narrow as the economy recovers, saying “It’s just taking a while.” I disagree with people in this camp. The gap has grown for over 40 years, and I believe that this is a fundamental change. Other economists, such as Lawrence Mishel, John Schmitt, and Heidi Shieholz, from the Economic Policy Institute, believe that the main reasons for this gap are political forces such as stagnant minimum wage, de-unionization, trade liberalization and deregulation. Specifically, this camp offers an explanation of these trends that does not rely on technological change. While these other points are valid, I disagree with their assertion that technological change plays no role in creating wage inequality. In terms of economists, I most closely align myself with people such as Daron Acemoglu and David Autor, from MIT, who argue that new technologies “directly substitute capital for labor in tasks previously performed by moderately skilled workers.” Dylan Matthews reports on this debate in the article Inequality is rising. Should we blame robots or the government? Or both?
What technological changes have emerged since 1970 that might be impacting this trend? Personal computers, which emerged in the late 1970s, allowed almost everyone to create and publish written materials. Worldwide standards emerged. People in all countries began to join the economic mainstream. The internet, which became popular in the 1990s, created new ways to collaborate and created new channels to market and sell products. As personal computers and the internet matured, technology emerged that streamlined business transactions, manufacturing processes, and product distribution. All of this facilitated the use of a worldwide labor pool with major improvements in productivity.
Autor and Acemoglu’s paper also show a growing divergence between the most-educated and least-educated workers. The authors show that, over the last 40 years, wages for those with a high school degree have fallen while those with a college or graduate degree have risen (see figure Trends in Wages). Economists call this “Skill-Biased Technological Change” (SBTC), which means a technological change that increases demand for high-skill labor while reducing or eliminating demand for low-skill labor. A classic example is workers displaced by robots that do routine tasks in a factory. Looking at the chart, one can see that this trend has accelerated since microprocessors were introduced in 1971 (one marker in the emerging digital era). Many economists (Autor, Katz and Krueger; Levy and Murname) have shown a correlation between demand for skilled labor and advances in digital technology.
In 2013, Autor co-authored another paper, this time with David Dorn, that shows another economic trend: in the past decade demand is rising for people with the lowest and highest skills, but is falling for those in the middle of the skill distribution. Federico S. Mandelman, Federal Reserve Bank of Atlanta, gives a condensed explanation. One reason for this trend is that jobs in the middle of the spectrum (book-keeper, bank teller, cashier, factory worker) are relatively easy to automate compared to either the low-end (hairdresser, gardener, and home health aide) or the high-end (managers and professionals). Many low-end jobs are hard to automate because they require physical dexterity and interpersonal interactions; many high-end jobs are hard to automate because they require creativity, problem solving and persuasion. In addition, the trend in some industries has been to off-shore many mid-skill jobs.
In 2013, Mandelman and Zlate build upon the work of Autor and Dorn. Essentially, they point out that wages are growing significantly only for the most skilled workers. At the low-end of the spectrum, wages are flat even though the number of people performing these jobs is increasing. Their hypothesis is “sizable inflows of low-skilled immigration into the United States during the last 30 years weakened the increase of wages for this skill group.”
Let me elaborate on my personal experience. In the 1980s and 1990s, I spent most of my time working on using computers for factory automation. In the beginning, I worked on designing a specialized factory for a U.S. government agency that cut, packaged and delivered packages of photos to clients. I then did my master’s thesis in robotics, which I briefly described above. After graduation, I worked on computer systems for process industries, such as systems to control the flow and temperature of liquids. Later, I worked on computer systems for high-speed motion control, which is a component of modern robotics. These systems increased productivity, but also affected someone’s job in the factories that used them.
As a technologist, I’ve observed and thought about the impact of technology my whole career. As such, I tend to align myself with those economists whose explanations align with my experience. That said, I think even the most open-minded economists underestimate the rate at which technology is accelerating; they place too much value on past experience in trying to predict the future. I disagree with those who believe that they can predict our future economy simply by studying the past; that history is repeating itself. I believe we are on the cusp of a new era.
This is post two of five. In the next post I’ll talk about the current situation in the job market.