Why it’s rotten to tell someone they’re poor because of when they had their kids

The “success sequence” is an idea from Ron Haskins and Isabel Sawhill at the Brookings Institution. They want to balance government investment and “personal responsibility” to reduce poverty. By personal responsibility, they mean adherence to what they call “the three norms”: complete high school, work full time, wait until you’re 21 and married to have children. If you do that — and smile while doing it — they’re willing to spot you a little welfare and some education.

I’ll describe it, and some criticism of the idea, then show a little data analysis.

Haskins and Sawhill claim to have analyzed data to show that when you follow all three of these norms, you have a 98% chance of not being poor. This is how they illustrate it (from this slideshow):

Matt Bruenig at Demos has an important post explaining how misleading — and wrong — this is. There are three main problems, briefly:

1. The high school degree and full-time job is doing almost all of the work. With those two hurdles complete, you’re already down under 4% poverty. So the marriage stuff is mostly moralizing for political purposes.
2. The data they used does not include the information necessary to see whether people were married when they had their children — it doesn’t have marital history. So they didn’t even do the analysis they said they did.
3. Family complications mess this up badly. In particular, if a person (say, a man), has children with a partner and never lives with them, he shows up as having met the “norms” because the data don’t show him having any children — it’s a household survey, so absent parents aren’t parents in the data.

So if someone gives you the “success sequence” thing, just remember, the analysis is baloney, and the bottom line is decent full-time jobs are what keep people out of poverty (by the official poverty measure, of course).

Analysis

Anyway, I can go a little further using the American Community Survey, which includes data on the year of each person’s most recent marriage, and the number of times they’ve been married. So, limiting the data to first-time married parents, I can check the age of their oldest child and see whether it was born before they were married, and before the parent was age 21. Some of the above problems still apply, but this is something. And it enables me to underscore Bruenig’s point that step three of the success sequence is not pulling its weight.

(Note this analysis is just about the timing of births for people who are currently married. Single parents of course have higher poverty rates that you can’t attribute to the timing of their births without more information than the ACS has.)

Using the 2013 ACS provided by IPUMS, I took all married parents, living in their own households, age 18 or older, married for the first time, with a child under 18 in the household. Then I used the job norm (self or spouse full-time employed), the education norm (high school complete), and the parent norm in two parts (child born after marriage, child born after age 21), as well as other variables, to see their relative contribution to not being poor. The other variables were additional education (BA degree), race/ethnicity, age, sex, disability, and nativity

This figure shows the marginal effects. That is, how much does the chance of being in poverty change with each of these conditions, holding all the others constant at their means? Click to enlarge:

If the oldest child in the family was born before the year of the parents’ marriage, the chance of being in poverty is increased by 0.4%. If the child was born before the parent was 21, the chance goes up by 0.6%. This seems reasonable to me, given the potential hardships associated with single and early parenthood. But compare: Not having a high school diploma increases the chance of poverty by 2.2%, and neither spouse having a full-time job increases the chance by 6.4%.

Remember, these are all effects holding constant everything else in the model. If you just look at the difference between those who fulfill the parenting “norm” and those who don’t, it’s much bigger. Among people with a full-time job in the family and a high school degree, the poverty rate is 2.8% for people whose oldest present child was born after they were married and 21, versus 9.1% for the people who let us all down on the childbearing norm. But that big difference is mostly because of education and race/ethnicity and disability, etc.

In short, this exposes how rotten it is to tell someone they are poor because of when they had their kids. A decent job and some education would mean a lot more than your sermon.

Code

Here is the IPUMS codebook for my download, and the Stata .do file for the analysis.