A while ago, I wrote how, even if we think that people should not be punished with bad lives for being unlucky with their upbringing or talents, we should favour some wealth and income inequality in many situations so as to achieve more overall justice. For example, if some choose to take more leisure instead of income, then it would not be just (on egalitarian grounds) to make sure their monetary income or wealth were equal—leisure is a sort of income. Similarly, it would not be just for people with more pleasant, safe or satisfying jobs to earn as much as those with less pleasant, more risky, or less interesting jobs (on egalitarian grounds).
There is one school of thought that understands and agrees with these caveat to equality. They may even note that teachers, who are respected and admired by society, and whose jobs are largely pleasant, satisfying and risk-free should not justly earn as much as despised unhappy bankers. But in the real world they see examples where people seem to have jobs that are high in pay and low in other benefits—boring or risky or low-status. Are markets failing to compensate people in pecuniary terms for the non-pecuniary costs of their job?
According to a new paper, they are not:
We use data on twins matched to register-based information on earnings to examine the long-standing puzzle of non-existent compensating wage differentials. The use of twin data allows us to remove otherwise unobserved productivity differences that were the prominent reason for estimation bias in the earlier studies. Using twin differences we find evidence for positive compensation of adverse working conditions in the labor market.
The authors look at twins to control for ‘unobservable’ factors about the person—basically their productivity and how employable they are—and find that workers are paid more for doing less desirable jobs. Though it often seems like they aren’t, this is only because some people (mainly people on early steps of the productivity/human capital letter without much in the way of experience or skills) are very unproductive. Of two twins, the one with a less pleasant or more risky job gets paid more.
The unique feature of our data is that we are able to estimate specifications that also control for unobservable time-invariant characteristics at the twin pair level (e.g. place of residence, spouse’s occupation, or family situation) and the common wage growth for the twins. We find evidence for positive compensating wage differentials for both monotonous and physical work using the data on MZs (monozygotic, i.e. identical, twins). Assuming that equal ability assumption holds for MZs, the results imply that both twin difference estimates and Difference-in-Differences (combined twin difference – time difference) estimates are unbiased for MZs.
I am perfectly comfortable with arguments that differences in endowments (basically upbringing and inborn talent) mean that there is a case for redistribution between individuals. But it bears repeating that markets, working properly, automatically produce results that in many ways elegantly fulfil the egalitarian concerns of leftist social justice.