The blogpost here is a little bit brief, and grabs data to make a point they favor. It's well worth reading the full conference paper and understanding the data presented, as well as the discussion by the authors on their model and projected impacts of programs to improve results for disadvantaged children. It's rather well-written and approachable, and interesting that it advocates programs to add extra tutoring time, and social counseling, to help children gain the skills needed to be upwardly mobile instead of simply throwing money at parents.
I'll start by linking more of the plots below. These are social mobility matrices, on the x axis is the income quintile of the parents, and the y axis gives the percentage of children in quintile at 40 years old.
I list out my observations:
- Overall the middle 60% is pretty close to unbiased, and bottom and the top tend to have sticking power.
- Dropping out of high school provides almost certainly poor outcomes.
- Even the drop-out kids of the top 20% have an 81% chance of winding up in the bottom 60%.
- I wish the authors included data on high school grads with no college.
- Going to college generally pays off.
- Those who make it through from the bottom, are 20x more likely to reach the top 20%.
- Worth noting that the distribution for college grads from poor parents is pretty fair, but skewed upward. Suggests you get out what you put in.
- Even with a college degree just a bit more than 1/3 kids from parents in the top 20% stay in the top 20%.
- Family coherence is critical.
- Children from single mother homes have much worse prospects.
- Further being poor hurts, but the chances of moving up to the top are about the same for all parent incomes
- Having both parents is about as good as a college education (likely because those are highly overlapping)
- Even having a two parents for part of childhood improves outcomes.
Overall I feel like the blogger from the Post cherry picked some graph bars to make a point without giving the fuller context. Overall I would say that family structure and education were far better indicators of future success than parental income.
If you average the percent of college grads ending up in the bottom %20, it's about 11.2% which is the same percent of rich kids ending up in the bottom 20%. Averaging across the college grads that end up in the top 20%, 31%, one percent better than being born to high income parents.
Seems to me like a college education tends to be an equalizing factor.




~edit~ I should point out that I don't show two plots discussing race and income results. I feel the authors did a disservice by only including a black vs. white comparison, and not breaking down race by education and family coherence as well. As such I see race as a red herring, a result variable covering up latent variables concerning education and family structure (and in turn more fundamental latent variables that are the drivers of those.) The poor level of detail in the reporting in the paper makes me feel like the authors were trying to hit some paper guidelines instead of making a serious investigation, and as such the results should be ignored until the authors produce a more thorough discussion.