Branko Milanovic has an interesting discussion of a recent conversation he had with Joe Stiglitz over the potential outcome of the robot economy. And I should get this in early, that they’re both, both of them, more intelligent than I am and better economists (not that either is all that hard). However, the end state that Stiglitz is worrying about is simply an impossible end state. The world just couldn’t end up like that.
Essentially, what Stiglitz is saying is that under certain conditions the advance of the robots means that we all lose our jobs and the capitalists, the people who own the robots, get to have all of the economy. And one can see his mechanism to get there: if robots become ever more productive then yes, we can see the idea that there will be more robots and fewer people employed by the capitalists and so on. But it’s still impossible for that end state to arrive: simply because we non-capitalists aren’t going to let it.
Here’s the laying out of the idea:
It is always instructive to speak to Joe Stiglitz. In a conversation in Paris which we had after his talk at the INET conference, he pointed out that the elasticity of substitution between capital and labor greater than 1 (which is often assumed by Piketty in his “Capital in the 21st century”), combined with technological progress which does not fall like manna from heaven but develops in response to the existing factor prices, would lead to an explosive process that would end only with capital owning the entire net income of a country. How?
Suppose that we have a given r (you can imagine that it is 5% as is often mentioned by Piketty) and a given wage (w). Suppose also that at this ratio of factor prices, it is profitable to invest in more capital-intensive processes (that is, they reduce unit cost of output). So capitalists will replace labor by capital and K/L and K/output ratios will both increase. Since elasticity of substitution between K and L is greater than 1, r will only slightly decrease while wage will only slightly increase. Although factor prices, being sticky, will not have budged much they would have moved ever slightly further in making capital intensive processes even more attractive. So there would be another round of increased capital investment, and again K/L and K/output will go up with only minimal effects on prices.
This will continue round after round until the entire output is produced practically only by using capital and perhaps just an infinitesimal quantity of labor. Both r and w will remain almost as they were at the beginning, but instead of (say) 100 machines and 100 workers, we will, at the end, have 100 robots and 1 worker. Almost all output will belong to the owners of capital. Piketty’s alpha will be close to 1.
This is why, in my interpretation, Stiglitz argues that the elasticity of substitution greater than 1 combined with endogenous technical progress leads ultimately to an explosive equilibrium.
Agreed that that would be a dystopian outcome. But it’s also an impossible one. Simply, as above, because we non-capitalists wouldn’t let it happen. And no, I don’t mean that there will be a bloody revolution before it does and I’m not even referring to the most obvious outcome, which is that we each own a bit of the robots and thus we’re all a little bit at least capitalist. Rather, I mean that such an outcome really is a logical impossibility.
So, let’s recast that end state. There’s the 1% (the plutocrats, the capitalists, whatever) and then there’s us, the 99%. Robots become ever more productive and we the 99% all lose our jobs working for the capitalists. Hmm, tant pis in one telling of this story because as Karl Marx pointed out that’s a precondition for true communism, that we overcome the scarcity problem. But even leaving that aside what is going to happen next?
That 1% owns all the robots and gets all the production from them. We, the 99%, have no jobs and thus no incomes. We cannot purchase any of that robotic consumption from the capitalists. This is the very point that Stiglitz is making, that the capitalists will own and consume 100% of the robotic output. Well, yes, OK, but what are we the 99% going to do? This new peasantry: do we all just wander around the fields until we keel over from starvation? No, quite obviously we don’t. Sure, the robots are more efficient than we are, produce things for much lower prices. But we don’t have any of the currency with which we can buy that production. So, what are we going to do?
Well, we’ll end up doing pretty much what we do now actually. We’ll carry on producing things using human labour to produce them. For, after all, there’s 99% of the population that wants stuff like housing, clothes, food and so on, even maybe the occasional baseball game. That 99% can’t get any of the cool stuff being produced by the robots because the capitalists own it all and none of us are employed for the currency with which to buy it. So, we’ll carry on building houses with human labour, making clothes, growing food and so on.
It’s thus possible to think that we could end up with two economies, the one where the robots do all the work and the capitalists do all the consuming, and the one where the rest of us work and consume much as we do now. But it’s not possible for the capitalists to end up with 100% of it all. Simply because we, the 99%, want to be able to consume and if the capitalists won’t let us then we’ll go off and make stuff to consume ourselves. As, in fact, has been going on for a number of millennia now.
Even though the technical analysis there by Stiglitz is just fine, the thing that he’s worrying about simply cannot happen. This is actually, waaay underneath it all, an example of comparative advantage from David Ricardo. The way I try to shoehorn that into my brain is to point out that if we all do what we’re least bad at then collectively we’ll all be better off. This being true even if everyone else is better at doing everything that you or I are. It is still true that if we produce what we’re least bad at, however much better everyone else is than we are, that we’ve increased total production as much as we can. Here the robots are those people who can do everything better than we can. But, just as with the basic case for trade, even though we’re worse at doing everything we can still do something. And, of course, at the very barest minimum we’ll get to consume what we ourselves produce. Or, as above, we’ll carry on doing that Smithian thing of dividing and specialising labour and trading in the resultant production.
The point being that even if the robots have absolute advantage over all of us at producing and doing everything Ricardo was correct: there’s always a comparative advantage available. Thus we can’t end up with only the robots doing all the work and the owners of the robots doing all the consuming.