Does Technology Make Most Workers Worse Off?

Since the early dawn of industrialization, many have argued that new technological advances would make things worse for large swaths of the working class. Most famously, the Luddites in the 1810s in England destroyed the new textile machines that threatened their jobs. In the early 1970s, longshoreman in many US ports went on strike to stop the spread of containerization as the dominant form of transport of goods between countries, and they were no more successful than the Luddites in stopping innovations that reduce costs. More generally, in the 1950s and 1960s, many critics argued that “automation” would lead to a permanent class of excess workers.

The current crop of negativists are once again predicting the decline of middle class jobs and purport to have some data to back their claims up. The most prominent proponent of the “disappearing middle jobs” thesis is David Autor from MIT. In series of papers and presentations, he finds a U-shaped curve in which low and high end jobs are growing while middle jobs are declining.

It appears to be a convincing case until one looks more closely at his approach. First, the male and female patterns are completely different. For women workers, middle jobs are declining but virtually all of the gain is due the high-end jobs growing. Male workers, by contrast, have a declining middle with equal gains at the upper and lower ends.

Second, there are other ways to define very good, good, and bad jobs. In an approach that I developed 15 years ago, I found that the share of bottom tier job (“less-skilled jobs) declined by a lot, the share of middle jobs declined by a bit, and the share of the top tier—managerial and professional jobs—grew substantially for both genders.

Hmm. The fact that two approaches can lead to much different results over the same period show how sensitive this analysis to the exact manner in which job quality is defined. I define my three grand job tiers on the basis of education and earnings with education weighted more heavily; Autor defines his tiers on the basis of earnings alone.

The results of both analyses are driven by the decline of manufacturing jobs—often held by unionized male workers. In my categorization, these are in the bottom tier of low-skilled jobs while in Autor’s approach they are in the middle tier. This group was particularly vulnerable to job losses because of new technologies (i.e., we produce the same amount of steel and coal today as we did in the 1960s with one-quarter of the workers) and because their pay relative to their schooling was unusually high (i.e., they were particularly vulnerable to outsourcing).

Ross Perot in 1992 colorfully proclaimed that “if NAFTA passed, we would hear a great sucking sound of jobs leaving the country.” Of course, the 1990s were a time of fast job creation and job declines have been associated with business cycles and not changes in technology or global trade.  There have been many other predictions of impending economic decline that didn’t come to fruition. I think that the experience of women workers is more indicative of the future of the American labor market and that the Autor’s U-shaped curve finding applies only to a certain time frame and specific methodology. We certainly have problems—particularly high inequality, but technology is a good thing that enlarges the economic pie.