Correlation and causation
By Scott Sumner
Bryan Caplan has a new post where he claims that people can avoid poverty with three simple steps:
If you live in the First World, there is a simple and highly effective formula for avoiding poverty:
1. Finish high school.
2. Get a full-time job once you finish school.
3. Get married before you have children.
This made me wonder if Bryan was confusing correlation with causation. He denies this:
A more agnostic criticism doubts causation. Sure, poverty correlates with failure to follow the success sequence. How, though, do we know that the so-called success sequence actually causes success? It’s not like we run experiments where we randomly assign lifestyles to people. The best answer to this challenge, frankly, is that causation is obvious. “Dropping out of school, idleness, and single parenthood make you poor” is on par with “burning money makes you poor.” The demand for further proof of the obvious is a thinly-veiled veto of unpalatable truths.
I am not at all convinced by this argument. Indeed I don’t see any real argument being made here. It seems equally plausible to me that the sort of person who doesn’t finish high school is different, on average, from those who do. The dropout may (on average) be less smart, less interested in classes, less motivated, and/or perhaps a bit anti-social. None of those traits are normally associated with financial success. If you put a gun to their heads and forced this cohort to finish high school, would that by itself change those personal characteristics? Maybe slightly, but how much? Would this group then become identical to other high school grads? I doubt it.
As for marriage, the Nordic countries tend to have a much higher share of births out of wedlock, and yet typically have relatively low rates of poverty:
You might argue that their culture is different, and that in Scandinavia even unmarried men often take an interest in raising their children. I accept that, but again it just makes me wonder if it’s marriage that is the key, or if the deciding factor is the personal characteristics of those who fall into poverty.
I certainly agree that working hard and being responsible are useful traits, and that some people are poor due to unfortunate life choices. I would push back, however, against any suggestion that there are simple public policy fixes, such as policies that discourage people from dropping out of high school or encouraging marriage. Those policies might work, but simple correlations don’t prove that. (BTW, I lean toward policies that make work more attractive, such as low wage subsidies and housing deregulation, as opposed to basic income programs that might discourage work.)
Also keep in mind that definitions of poverty are based on “households”, where the poverty line increases only modestly each time a person is added to a household. Thus if two single people making $10,000 each decide to double up and live in the same apartment, that pushes them above the poverty line. It’s not obvious their situation improved (otherwise no one would ever chose to live alone), but the US government treats the decision to share an apartment as an improvement of living standards. This biases the statistics toward the conclusion that marriage improves one’s economic well being.
Thus you might just as well argue that poverty could be almost eliminated if everyone lived like Chinese college students in the 1980s, with eight people per apartment.
Even with minimum wage jobs, a household of eight will earn far more than $47,650. But would those “households” be better off, or would people get on each other’s nerves? (My wife shared a room with 7 other college students in the 1980s, in Beijing.)
Finally, most women have a strong preference to have children. Finding a suitable husband is not always a “simple” process.