I have written before that it is OK to worry about the effects of automation on work. One frequently encountered response is that this worry is wrong, that having it is committing the Lump of Labor fallacy. Marc in particular has been bringing it up a lot, including on Twitter, in a blog post, and a recent interview.
But this is a red herring. One can legitimately worry about the transition even if one believes that the Lump of Labor fallacy is in fact a fallacy. So let me posit for a moment that it is a fallacy (more on that later). Even then the adjustments in the labor market are likely far slower than the displacement effects from automation. Employers can fire people quickly. People take a long time to acquire new skills and find new jobs (and these jobs need to be created first). We have ample of evidence for that from the industrial revolution.
In the industrial revolution agricultural work disappeared much more rapidly than factory work was created. This resulted in an oversupply of labor which allowed factories to pay a pittance, have horrible working conditions, no vacations, etc. The social upheaval that resulted brought us the revolutions of 1848 in Europe and contributed significantly to both World Wars.
We are again experiencing something similar. Why is it possible that ISIS has been able to recruit over two thousand Europeans to join them? One of the key reasons is that youth unemployment in European countries is above 20% on average – that’s 1 in 5 young people without work.
So Lump of Labor or not, we need to do way more than we are about the adjustment. Especially because we have created a situation where the returns to Labor are low at the same time that the returns to Capital are high, which is causing the unprecedented levels of income and wealth inequality. We had a similar situation at the outset of the industrial revolution (due to the effects of land ownership) but the two World Wars managed to destroy enough of the existing wealth accumulation to create more equality for a while. That is really the central point of Piketty’s book.
Now let me address Lump of Labor itself. People who think it is a fallacy argue as follows: we use machines to do something, that frees up labor to do new things which are additive to the economy (and which we can now afford to do as the machines are doing other stuff). So: we used to work in agriculture, then we worked less in that but added manufacturing, then we worked less in manufacturing but we added services.
Now we will work less in services but we will add X. That’s historically correct up to now but it is legitimate to disagree on (a) what is X and (b) how much will anyone get paid to do X. Here are some possible concrete examples
X = healthcare including caring for older people (the last part is particularly promising due to demographic shifts)
X = cultural production (making music, art, etc)
X = cleaning up the environment
X = traveling to the stars
Marc says he is “long on human creativity” which I interpret to mean we won’t run out of answers to what is X. And I completely agree with that! We will always come up with interesting things to do. But that is only one half of the equation.
Here is the other half of the equation: just because we work on X doesn’t mean we get paid to do it or paid enough. We do plenty of things today already that used to be work at one time but we now actually pay to do them, such as sailing a sailboat or riding a horse. So what determines whether you can get paid for doing X?
First, it is supply and demand for the output of X. And many of the logical X’s currently have insufficient demand (in price terms not necessarily in quantity terms). For instance what is the paid demand for cleaning up the environment? Even caring for older people has an issue here — yes we have lots of people getting older but many of them have no retirement savings, so the paid demand for care can’t come from them directly. Others of the logical X’s suffer from over supply, such as cultural production (eg 100 hours of vide uploaded to Youtube every minute). So it is entirely possible that the market clearing price for many of the possible X’s is low or even zero.
Second, it is the potential for substituting machines in the production of X. In order for X to produce a return to human labor, human labor must be cheaper for that activity than machine labor. There is simply nothing in economics that says this must be the case. You have to go outside of it and somehow construct an argument that machines cannot be cheaper. One place you can look to for that is where people specifically are paying to interact with a human but the list of those isn’t super long.
So in order for Lump of Labor to be a fallacy two conditions must be true. We must have enough new activities X that have a meaningful market size and where humans are better than machines at producing them. Nothing inside economic theory can tell you whether these conditions will be met. You have to go outside of it and appeal to say history, as in “Lump of Labor was a fallacy during the first industrial and second industrial revolutions so I believe it will be a fallacy again” against which someone else could argue “Lump of Labor turned out to be true for horses so there is precedent for other factors of production.”
Where I come down personally is that this is first and foremost a (re)distribution issue both in the transition phase and once we are past that. Here I am supporting for something along the lines of a Basic Income Guarantee which I believe is a better solution than growing the public sector. If enough people have access to some basic income we vastly increase the likelihood that both conditions are met in the long run and Lump of Labor will once again be a fallacy. I also believe this is an issue of creating new markets where there currently aren’t any and one prime opportunity here is the environment and within that the leading candidate is carbon. That would open up a very substantial X, eg reforestation, solar installation, etc. that is currently under priced.
