If you were in charge of the economy, you’d probably care that it could produce a lot, that it had high productivity, that it provided lots of jobs, and that these jobs offered decent pay on average. You might well care about other things, too, but if these four indicators were all headed in the right direction over time (in other words, UP) you’d be pretty happy.
So how have we actually been doing in the US? Well, as the graphs below show we’ve been experiencing a long, slow decoupling between the first two of these — output and productivity — and the last two — jobs and wages. For more than three decades after the end of World War II all four of these measures went up together:
The French call the thirty years after the war les trente glorieuses, reflecting the shared economic prosperity of the period. Well, we had a bit more than trente spectaculaires of our own.
In the early 1980s the picture started to change for the average American worker. There were still a lot of jobs available, but they started to pay less well. Median household income became decoupled from the other three stats and grew more slowly than they did. By the time of the 2001 recession, median income was lagging behind pretty badly. If we’re going to stick with Gallic labeling, the years between 1982 and 2001 were the vingt troublantes:
By the end of 2011, things had become much worse in two ways. First, median household income was actually lower than it was a decade earlier. In fact, it was lower than at any point since 1996. And second, the American job creation engine was sputtering badly. Between 1981 and 2001 the economy generated plenty of low-paying jobs. After 2001, though, it wasn’t even generating enough of these, and employment growth started to lag badly behind GDP and productivity growth (on all three graphs here, GDP growth is charted on a separate axis because it grows more quickly than the other three). The last ten or so years have been les dix déprimantes.
What’s going on? Why have the things that workers care about – jobs and wages – become decoupled from the the other things that economy-watchers care about? So far, explanations for this unhappy phenomenon include tax and policy changes, and the effects of globalization and offshoring. These are clearly powerful forces, but there’s one other one: technological progress.
I’ve been talking a lot about this latter force here and elsewhere, and it’s the subject of Race Against the Machine, a short e-book Erik Brynjolfsson and I wrote that came out a bit more than a year ago (we’re working on a full-length sequel now).
Our argument, in brief, is that digital technologies have been able to do routine work for a while now. This allows them to substitute for less-skilled and -educated workers, and puts a lot of downward pressure on the median wage. As computers and robots get more and more powerful while simultaneously getting cheaper and more widespread this phenomenon spreads, to the point where economically rational employers prefer buying more technology over hiring more workers. In other words, they prefer capital over labor. This preference affects both wages and job volumes. And the situation will only accelerate as robots and computers learn to do more and more, and to take over jobs that we currently think of not as ‘routine,’ but as requiring a lot of skill and/or education.
As a result, I don’t see the four lines in the graphs above re-converging any time soon.
Over the past couple days this argument has gotten some attention and support from Paul Krugman, a Nobel-prize winning economist, New York Times columnist, and incredibly popular and prolific blogger. On Saturday he put up a post titled “Rise of the Robots” in which he wrote:
If this is the wave of the future, it makes nonsense of just about all the conventional wisdom on reducing inequality…
I think our eyes have been averted from the capital/labor dimension of inequality, for several reasons. It didn’t seem crucial back in the 1990s, and not enough people (me included!) have looked up to notice that things have changed. It has echoes of old-fashioned Marxism — which shouldn’t be a reason to ignore facts, but too often is. And it has really uncomfortable implications.
But I think we’d better start paying attention to those implications.
Erik and I sent him an e-copy of RAtM after reading the post. Krugman is evidently a fast reader, because his Monday Times column, entitled “Robots and Robber Barons” had this to say:
There’s no question that in some high-profile industries, technology is displacing workers of all, or almost all, kinds. For example, one of the reasons some high-technology manufacturing has lately been moving back to the United States is that these days the most valuable piece of a computer, the motherboard, is basically made by robots, so cheap Asian labor is no longer a reason to produce them abroad.
In a recent book, “Race Against the Machine,” M.I.T.’s Erik Brynjolfsson and Andrew McAfee argue that similar stories are playing out in many fields, including services like translation and legal research. What’s striking about their examples is that many of the jobs being displaced are high-skill and high-wage; the downside of technology isn’t limited to menial workers.
Still, can innovation and progress really hurt large numbers of workers, maybe even workers in general? I often encounter assertions that this can’t happen. But the truth is that it can…
We agree completely. As we wrote,
computers are now doing many things that used to be the domain of people only. The pace and scale of this encroachment into human skills is relatively recent and has profound economic implications. Perhaps the most important of these is that while digital progress grows the overall economic pie, it can do so while leaving some people, or even a lot of them, worse off.
Krugman writes that “I think it’s fair to say that the shift of income from labor to capital has not yet made it into our national discourse… but it’s time to get started, before the robots and the robber barons turn our society into something unrecognizable.”
The national discourse needs to acknowledge the Great Decoupling, and also acknowledge that it’s not going to be reversed by a couple quick policy fixes or even, I believe, by deeper changes to our educational and entrepreneurial systems. I believe it’s a simple fact of the technological era we’ve been creating.
I want us to continue this work of creation — as I’ve said before, unplugging the computers would be about as bad an idea as ripping up all the roads and closing all the schools — but as we do so we need to rise to the grand challenge of dealing effectively with the Great Decoupling.