On a New Lifecycle Income Study

Danielle Paquette at Wonkblog reported on a new study about income trends across individual lifecycles. The findings confirm the basic point that income rank does not differ much across the lifecycle (a person whose income puts them in the 30th percentile of 30-year-olds at age 30 is likely to have an income that puts them around the 30th percentile of 50-year-olds at age 50). The study also found that the personal incomes of those at the 20th percentile and below do not increase much (if any) between age 25 and 45 and fall significantly between age 45 and 55.

A couple of people asked me what implication this study has for the long-standing observation that poverty is related to income lifecycles (i.e. younger workers have much higher poverty rates than older workers). The suggestion is that this undermines those observations, especially because the poorest workers don't see their personal income change much over the lifecycle. So I address that question here.

Only Personal Income

The first thing to note is that the study only tracks personal income of the selected individuals. But poverty is a function of family income, not personal income.

This problem is compounded by the fact that the bottom 20 percent has very little personal income at all points of the lifecycle. You can see this in the following graph I pulled from the ACS 2005-2007 3-year file (which is cross-sectional rather than panel data, but nonetheless makes the point):

The bottom percentiles of personal income will be comprised in large part of people who have very little income because, among other things, they rely on others (e.g. spouses) for the family's market income. These individuals' personal incomes do not tell us much about the lifecycle income of their family because that is being driven entirely by the other person or persons in the family who are doing the actual "earning."

If we draw the same graph as above, but this time use each person's family income rather than personal income, a different picture emerges:

It can be hard to see, but the lifecycle phenomenon is present at every percentile. At age 25-34, the 10th percentile has a family income of $12.9k. At 35-44 it is up to $16.6k, an increase of 30%. At age 44-54, it is up to $17k. At age 25-34, the 20th percentile has a family income of $22.7k. It is up to $29.3k at age 35-44, also an increase of 30%. It is up to $30.8k at age 45-54.

So when we look at family incomes near the bottom (rather than personal incomes near the bottom), the lifecycle trend clearly emerges.

Family Size

Of course, poverty is not solely a function of family income. It is also a function of family size. This study (since it is not designed to do this) takes no account of family size and, since it is using SSA data, couldn't even if it tried. Family size is important when discussing lifecycle poverty because younger adults have higher poverty, not simply because their income is at a lower part of its lifecycle, but also because their family size is often larger on account of children.

Because the poverty line for a given family depends on the size of the family, one quick and easy way to create an income measure that is sensitive to family size is to represent income as a percentage of the poverty line. The following graph does this (note 501% is the top code, which is why the 80th percentile is just a flat line):

The 10th percentile goes from 67% of the poverty line to 87% of the poverty line to 104% of the poverty line. The 20th percentile goes from 130% of the poverty line to 158% to 190%.

Conclusion

To be completely clear here, this post is not meant as a criticism of the study. The study was simply not about poverty lifecycles, which it never purported to be about. Those who think it may undercut the long-standing poverty lifecycle observations have simply misapplied it. The bottom 20 percent in personal income is not the same thing as the bottom 20 percent in family income and will contain many people who simply do not do much market labor because they depend on others in their family to do so. In fact, some people who have little personal income actually live in families with very high family incomes.

Personal incomes are not totally irrelevant to poverty analysis, but this particular analysis of them simply does not shed much light (good or bad) on any poverty-related questions.

Comments