Economic Institutions Should Keep Americans Out of Poverty, not Norms

Rich Lowry wrote this at Politico:

It’s closer to the truth that they, like all Americans, are in a much better position to succeed if they honor certain basic norms: graduate from high school; get a full-time job; don’t have a child before age 21 and get married before childbearing. Among the people who do these things, according to the research of Ron Haskins and Isabel Sawhill of the Brookings Institution, about 75 percent attain the middle class, broadly defined.

This comes to us from this 2009 chart produced by Isabell Sawhill and Ron Haskins.

Dylan Matthews has a good piece explaining some of the issues with this at Vox. Jordan Weissman has a decent piece explaining some of the issues with this at Moneybox. I would like to underscore some of the things they said and add my own observations based upon a replication attempt I made.

First, this analysis proceeds by excluding all families headed by people who are below the age of 25, above the age of 64, and who receives disability income. This means 60.7 million americans were dropped from the analysis altogether, which amounts to excluding 20.3% of the entire US population. By excluding this fifth of the US population, they also dropped 10.1 million poor people from their dataset, which was equal to 27.2% of all the poor people in the US. Needless to say, ignoring more than 1 in 4 of the nation's poor is not exactly a great way to get terrific poverty eradication insight.

Second, as Matthews points out, the number of people in the 0 norms category is vanishingly small. They put it at 1.4% of their sample population. Against the overall US population, it's actually 1.1%.

Third, the middle column of one or two norms is entirely unhelpful without a breakdown of which norms they are talking about (more on this later). In their book, they note that "looking at each norm separately, working full time has the strongest association with ensuring one's position in the middle rungs of the income distribution." Instead of breaking out how much work the full-time screen is actually doing in this, though, the authors instead drop a footnote that says you can contact them to get more information about their methodology (I've tried to do this, but so far to no avail).

Fourth, the breakdown is hopelessly mired in selection effects problems. People with higher incomes are more likely to find marriage partners than lower incomes, in which case it's the income that's driving both the marriage and the low poverty. Additionally, people who have full-time work are more likely to find marriage partners than those without it, presenting the same basic analytical problem.


I set out to replicate the Sawhill/Haskins figures, which was met with mixed success. I did manage to basically replicate the initial screen they put on the 2007 data, which involves excluding all families headed by people below the age of 25, above the age of 64, and who receive disability income. No matter how much I tried, I could never get the figures to line up exactly, but I got close enough:

Efforts to replicate their norm groupings were a total failure. They don't explain how they code anything, and figuring out what to do with certain things like related subfamilies, unmarried people, the childless, and so on is not obvious to me. Further, when you go hunting for footnotes for elucidation, you are met once again with the instruction to contact them personally (which I have attempted to do).

Nonetheless, because I have isolated the overall population Sawhill/Haskins is working with, I am able to do some fun calculations of my own. Here is what those tests reveal.

First, if you take this population and simply isolate families whose head meets the full-time work definition Sawhill/Haskins uses, that alone gets you almost entirely to the low poverty conclusion. Specifically, people in families in this sample with a head that worked at least 40 weeks and at least 35 hours per week had a poverty rate of 4%. This norm alone gets you within 2 percentage points of the Sawhill/Haskins 3-norm poverty rate of 2%.

Second, if you take this population and isolate full-time work + high school education or greater (which is 2 of the norms), you get a poverty rate of 2.7%. This puts you just 0.7 percentage points shy of the Sawhill/Haskins 3-norm poverty rate. This means that without ever bringing in timing of children or marriage, we basically get all the way to the Sawhill/Haskins figure.

Third, marriage and children increases the likelihood of poverty. If you take this population and isolate full-time work + high school or greater + lives alone, you get a poverty rate of 1.5%, which is lower than the Sawhill/Haskins 2% figure.

Of course, none of this stuff is particularly surprising. Taken alone, marriage actually increases poverty risks because it adds people to your family (the spouse), which increases the amount of money you need to be out of poverty. To the extent that marriage does anything at all to poverty, it is not the marrying, but the possibility of two incomes (where the marginal income of adding a spouse is greater than the marginal increase in the poverty line that results from adding them). Sometimes they may have this income. Other times they may not. But at all times, as with all of poverty stuff, it is the income that is doing the anti-poverty work.

Additionally, the full-time work thing is also pretty known. Earlier this year, I broke down the impoverished over time based on their individual life status and got this:

If we exclude the two-thirds of all poor peple in the bottom four categories, we are left with just people who worked at least one week in the year, people who were involuntarily unemployed for the entire year, people who were out of the labor force because they were taking care of home/family, and then the rest. The top three categories are not working, and so that's obviously a decent enough chunk of the poor.

But also, if you break down the people who did work at least one week last year, and you see what percentage of them say they were also involuntarily unemployed or involuntarily part-time for at least one week in the year, you get this:

Which is to say, workers who are poor in a given year are also workers who tend to have faced spells of involuntary unemployment during the year. This happens every year because, as you may know, the labor force turns over. Selecting out these people by only screening for the full-timers magically disappears the transitory poverty caused by the permanent reality of people becoming involuntarily unemployed.

What's Immoral About Having Kids Young and Not Graduating HS?

The last thing I'd like to point out here is that it's deeply strange to me what's supposed to be compelling about the delay-kids-until-21 and graduate-high-school norms. Marriage has a moral component to it for some. And of course other things people like to talk about regarding poverty like drug use are generally bad ideas (though overblown). But a person without a high school degree having kids before the age of 21 was basically the median lifecourse in America in the middle of the last century. Was the median American a half century ago an immoral wreck?

I understand that the point (as unsuccessful as it has been) is simply to say you need to do those things to stay out of poverty. But unless there is an independent compelling case for why not having a high school degree and starting a family young are bad, then this analysis proceeds not by showing bad actions lead to poverty, but instead by finding what leads to poverty under our current set of institutions in our current period and then deeming it a bad action because of it. Which is to say it falls into the same circularity trap that most meritocracy discussions fall into.

In any society, there are going to be people who don't do well in education and also people who want to start families young. We shouldn't build our economic institutions to just brutalize those kinds of people when it's totally unnecessary to do so. A good family benefit regime (like the one I spelled out last week) should lead to comfortable enough lives for almost all families in the country, enabling people to take different paths based on their own preferences, while leaving nobody in the gutter. That's the kind of society we should want to have and it's one we could have by simply changing our economic institutions.