Thursday, September 12, 2019

There's No Evidence of a Robot Takeover–This Is a Bad Thing

Noah Smith wrote a pretty compelling op ed in Bloomberg on Tuesday demonstrating that worries about a robot apocalypse–the eclipse of human labor by machine labor–are unfounded. He points to the rather steady employment-to-population ratio, the positive correlation between IT investment and jobs, and low productivity growth as proof that robots do not, in fact, cost at least the macroeconomy jobs. (We can't necessarily make assertions microeconomically.)

This is a bad thing.

First, let's go through the evidence. Smith points to the constancy of the employment-to-population ratio (between 65% and 80%, growing over time). To drive the point home, he could have pointed to the relative constancy of average weekly work hours to demonstrate that automation hasn't cost work hours either.

Smith admits that the stability of these statistics doesn't imply stability at the microeconomic level. Automation often makes certain jobs obsolete, requiring sometimes expensive retraining. Even still, Smith cites evidence from Mishel and Bivens that displacement from automation is dwarfed by that of displacement from trade with China.

Next, Smith insists that if there really were a robot effect, we'd see accelerating productivity, which is not borne out by the evidence. He also cites the stable wage premium for higher education. I think this would have been more elucidating if this were broken down by major since there are plenty of post-college jobs that are subject to automation.

The rest of the op ed is a critique of de Blasio's proposed robot tax. It's not really a robot tax so much as something you might find in a union contract: a guarantee of comparable employment if workers are automated out of work. I seriously doubt there would be the enforcement apparatus necessary to make such a policy actually stick.

More concerning in my view is the smug dismissal on Smith's part that robots don't cost jobs. Shouldn't we want that?! Shouldn't we want to collectively have to waste less of our lives occupied in production and more in enjoying the fruits of it?

If the myth of a robot apocalypse is the domain of science fiction as Smith insists, it is because we all assume that robots are there to make our lives collectively easier. If robots aren't costing us jobs, they aren't doing their job.

Perhaps this shouldn't be surprising in capitalism. Unlike other economic systems, capitalism requires labor, not because it is essential to production, but because it is essential to consumption. It is this tension that leads to the accumulation of consumer debt and eventual economic crises.

The problem with automation in capitalism is that the excess spoils of productivity increases accrue as profits and not wages. One reformist alternative that might alleviate (or accelerate) the eclipse of human labor would be the establishment of collectives which could collectively arrange to cut their hours at the same rate of pay (or more depending upon productivity gains from automation).

Of course, capital doesn't fall from the sky, and few business owners are amenable to sacrificing their authority and right to profit to reestablish their business as a non-hierarchical collective. In all likelihood, the robot apocalypse will require expropriations.