The One Where Larry Summers Demolished the Robots and Skills Arguments

February 20, 2015

Everyone should take it easy on the robot stuff for a while.


There’s been a small, but influential, hysteria surrounding the idea is that a huge wave of automation, technology and skills have lead to a massive structural change in the economy since 2010. The implicit argument here is that robots and machines have both made traditional demand-side policies irrelevant or naïve, and been a major driver of wage stagnation and inequality. Though not the most pernicious story that gained prominence as the recovery remained sluggish in 2010 to 2011, it gained an important foothold among elite discussion.

That is over. They say Washington DC has had a huge crime decline, but I just saw one of the most vicious muggings I’m likely to see, one where David Autor and Larry Summers just tore this idea that a Machine Age is responsible for our economic plight apart on a panel yesterday at the Hamilton Project for the launch of a new Machine Age report. Summers, in particular, took an aggressive tone that is likely to be where liberal and Democratic Party mainstream economic thinking is in advance of 2016. It is a very, very good place.
As Larry Mishel noted in an American Prospect piece on the eve of the event, the robots and skills story has many problems. But I was genuinely surprised at how poorly those pushing the argument – Erik Brynjolfsson and Andrew McAfee, authors of the influential The Second Machine Age, were there – could address the pretty obvious counterarguments that were brought up. You can see the event at CSPAN here. This piece will mostly be blockquotes, because the quotes are too good to try and summarize. (Any transcribing errors are my own.)
First up, economist David Autor of MIT demolished the core claims in about a minute of speaking. For those with ears to hear it, this is him also moderating and walking back portions of his “job polarization” arguments from 2010. Autor:

I think there’s reason for some skepticism about how fast things are actually moving. There’s a lot of aggregate data that don’t support the idea the labor market is changing or the economy is changing as rapidly as this very dramatic story. The premium to higher education has plateaued over the last 10 years. We see evidence highly skilled workers have less rapid career trajectories and are moving into less skill occupation if anything. Productivity is not growing very rapidly, and a lot of the employment growth we’ve seen in the past 15 years has been in relatively low education, in-person service occupations.

The second point I want to make, when we think about how technology interacts with labor market we think of substitution of labor with machinery. […] What is neglected is that it complements us as well. Many activities require a mixture of things. it requires a mixture of information process and creativity, motor power and dexterity. if those things need to be done together if you make one cheaper and more productive, you increase the value of the other.

It got worse for the robots. Here’s Larry Summers:

On the one hand we have enormous anecdotal evidence and visual evidence that points to technology having huge and pervasive effects. Whether it is complementing workers and making them much more productive in a happy way, or whether it is substituting for them and leaving them unemployed can be debated. In either of those scenarios you would expect it to be producing a renaissance of higher productivity.
So, on the one hand are convinced of the far greater pervasiveness of technology in the last few years, and, on the other hand, the productivity statistics on the last dozen years are dismal. Any fully satisfactory view has to reconcile those two observations and I have not heard it satisfactorily reconciled.

Summers also pointed out something that’s fairly obvious once you think about it: if this robot argument is true, doesn’t it mean a short-term job boom? (JW Mason once pointed out this is how Schumpeter thought of these types of “recalculation” business cycles.) Summers:

If you believe technology happens with a big lag and it’s only going to happen in the future, that’s fine. But then you can’t believe it’s already caused a large amount of inequality and disruption of employment today. […] Let’s take retailing. You can imagine you can have all kinds of spiffy technology so you no longer have to have people behind cash registers. The problem is you wouldn’t expect the people behind the cash registers would get fired before the people working the systems got the new systems working. […] I understand why it might take years for it all to have an effect. What I have a harder time understanding is how there can be substantial disemployment ahead of the effect of the productivity.

That is, if you thought that it just was impossible to put in these systems and so forth, then you might think that in the short run it would be a big employment boom. You have to keep your old legacy system going and you have to have a million guys running around figuring out how to put the new computer system in.

The panel goes into a thing about how education will save us. Moar education! This is where Summers went harder than I had expected:

I think the [education] policies that Aneesh is talking about are largely whistling past the graveyard. The core problem is that there aren’t enough jobs. If you help some people, you could help them get the jobs, but then someone else won’t get the jobs. Unless you’re doing things that have things that are effecting the demand for jobs, you’re helping people win a race to get a finite number of jobs. […]

Folks, wage inflation in the united states is 2%. It has not gone up in five years. There are not 3% of the economy where there’s any evidence of hyper wage inflation of a kind that would go with worker shortages. The idea that you can just have better training and then there are all these jobs, all these places where there are shortages and we just need the train people is fundamentally an evasion. […]

I am concerned that if we allow the idea to take hold, that all we need to do is there are all these jobs with skills and if we can just train people a bit, then they’ll be able to get into them and the whole problem will go away. I think that is fundamentally an evasion of a profound social challenge.

But, but, but, if we don’t just educate people more, what can we even do? Summers:

What we need is more demand and that goes to short run cyclical policy, more generally to how we operate macroeconomic policy, and the enormous importance of having tighter labor markets, so that firms have an incentive to reach for worker, rather than workers having to reach for firms. […]

I think that the broad empowerment of labor in a world where an increasing part of the economy is generating income that has a kind of rent aspect to it, the question of who’s going to share in it becomes very large. One of the puzzles of our economy today is that on the one hand, we have record low real interest rates, that are expected to be record low for 30 years if you look at the index bond market. And on the other hand, we have record high profits. And you tend to think record high profits would mean record high returns to capital, would also mean really high interest rates. And what we actually have is really low real interest rates. The way to think about that is there’s a lot of rents in what we’re calling profits that don’t really represent a return to investment, but represent a rent.

The question then is who’s going to get those rents? Which goes to the minimum wage, goes to the power of union, goes through the presence of profit sharing, goes to the length of patents and a variety of other government policies that confer rents and then when those are received, goes to the question of how progressive the tax and transfer system is. That has got to be a very, very large part of the picture.

Two bonus quotes. First, someone immediately followed up that instead of the minimum wage, why don’t we just expand the earned income tax credit? Summers:

If we had the income distribution in the United States that we did in 1979, the top 1% would have $1 trillion less today, and the bottom 80% would have $1 trillion more. That works out to about $700,000 a family for the top 1%, and about $11,000 a year for a family in the bottom 80%.

That’s a trillion dollars. I don’t know what the number is, but my guess is that the total cost of the Earned Income Tax Credit is $50 billion. Nobody’s got on the policy agenda doubling the Earned Income Tax Credit. The big, aggressive agendas are probably to increase it by a third or a half. So, I’m all for it, but we are talking about 2.5% of the redistribution that has taken place. So, you have to be looking for things and there’s no one thing that is going to do it. My reading of the evidence, it’s a fairly general evidence, is that while there may be some elasticity, the elasticity around the current level of the minimum wage is very low.

Nice. And from his introductory remarks, Robert Rubin casually mentions collective bargaining might be a solution to inequality, but also probably redistribution and a cultural and policy shift towards more free time and more leisure. Ya know, no biggie. Rubin:

We may need an increase in the income tax credit, not only for those who receive it at the present time but perhaps much further up the income scale. Measures that facilitate collective bargaining can result in a broader participation in the benefits of productivity and growth […] If we have ever rapid technological development and it is labor displacing, at some point in the future — as I say, that may be some distant point in the future — should that lead to some basic change in our lifestyles with less work, more lecture and a richer, more robust use of that leisure? […] In addition to everything that needs to be done to enhance growth, tighten labor markets and to improve the position of middle and lower income workers, should there be increased redistribution to accomplish the broad objectives of our society?

(I looked at the left-liberals I knew active in policy circles in the 1990s who were in the room, wondering how they kept their heads from exploding at that moment.)
Perhaps this turn is just reflecting this very specific historical moment, and it could change again just as quickly. But the problems are real, and terrifying stories about robots taking all the jobs can no longer have the double function as a form of relief that we have no responsibility to try and address these problems. And it’s great to see prominent liberal economists doing that, especially in advance of the 2016 election.