Recently, the famous Nobel laureate economist Paul Krugman criticized Andrew Yang, politician and UBI champion and candidate for mayor of New York, for not having done the math on the cost of a universal basic income in the US. One sentence in particular caught my attention:
Are we actually experiencing rapid automation — that is, a rapid reduction in the number of workers it takes to produce a given amount of stuff? That would imply a rapid rise in the amount of stuff produced by each worker still employed — that is, rapidly rising productivity.
But that’s not what we’re seeing. …This slowdown has been especially pronounced in manufacturing, which has seen hardly any productivity rise over the past decade.
So who is right? Yang when he says robots are putting us out of work? Or Krugman when he says that’s not happening because productivity is not increasing?
Well, let’s see…
Productivity can be defined as the output (e.g. in €) per hour worked. That is, if you work better – you produce more in less time – it increases. Everything points to the fact that productivity in industrialized countries should increase as technical improvements, robots, better software, etc. are incorporated. Common sense tells us that it should increase. There are always competent companies that improve what is there, aren’t there? However, the opposite is the case, productivity has never grown so little as in recent years.. See the attached graph:
So, technically Krugman’s statement is absolutely true. Productivity is not growing. Now, does this necessarily mean that automation is not putting us out of work?
Why isn’t productivity growing?
No one seems to know quite why. But let’s try to shed some light. On the one hand we have the productivity of every single company, on the other the productivity of the economy as a whole.
The productivity of the economy is the sum of the productivity of all firms, but that does not mean that if a group of firms increases their productivity a lot, the economy as a whole will increase productivity. It may seem counterintuitive, but let’s look at an example:
Let’s say a company with a turnover of 1.000€/h and 10 workers. Its productivity is 100€/h and employee. The company automates by introducing software and robots and, after a while, produces 1000€/h with 5 workers. Productivity is now 200€/h. The firm lays off 5 workers because they don’t need them. The former workers set up a hamburger restaurant and all together earn 200€/h. Their productivity is 40€/h and employee. The total productivity of the ten workers is 1,200 €/h, i.e. 120 euros/h and person. So far Krugman is still right, overall productivity has increased.
But what happens when there is a lot of competition, all companies automate and it’s not a growth market? The company has to adjust its prices. And then, instead of selling at the usual price (let’s say it was €10 per product), it has to sacrifice part of its new margin and sell 20% cheaper, at €8 per product.
It turns out that the company now has a turnover of 800€/h with 5 workers. The total productivity is 160€/h + 40€/h of the hamburguers. In other words, despite the doubling of productivity in the first company, total productivity in the system is the same as before. It does not matter if we look at productivity per hour or per person. However, the situation is not the same. Now, apart from the unknown products of the first company, everyone can enjoy burgers!
So we have a situation that corresponds quite closely to our real world: automation has improved process efficiency and total output has gone up, but productivity is stagnant. The same people work the same hours but probably the hamburger guys now earn less than before and the owner of the company earns more. We talked about the latter in The reason for inequality
Stagnant productivity, a symptom of misallocation and poor growth
In a market of declining turnover and poor reallocation of resources, productivity stagnates or declines. Competition between companies is not the only factor in lost productivity gains after automation. Even more important are the overall drops in turnover due to product substitution (discussed in The Substitution Effect) and the digitalization of products (The Matrix effect).
In sectors where these two effects do not exist or where there is high growth and resources are well utilized, productivity does not stagnate. For example, in the agricultural sector. A potato is a potato, I can’t replace it with a digital potato. That’s why productivity here continues to grow.
Is it possible to pay a universal basic wage?
I honestly don’t know. I believe that the money available for, for example, a universal basic wage, is the money that can be created by a government together with the central bank without triggering inflation. Technology is deflationary and therefore leaves considerably more room than in other times until inflation arrives. The amount of money that can be made available without triggering an inflationary spiral I don’t think Andrew Yang or anyone else knows how to calculate.
In short, my guess is that Yang probably hasn’t done the math, but he’s right that technology is creating circumstances that may allow for something like a universal basic wage. And the fact that productivity is stagnating is not a symptom that we are not automating as much as we think, but the direct consequence of a process of deflation due precisely to automation and digitization.