Which race is helped most by transfers?

Ronald Brownstein has a bizarre article in The Atlantic about which race is helped most by transfers, which is basically just a rewrite of a CBPP post from yesterday.

in both percentage terms and absolute numbers, federal programs reduced poverty among working-age whites without a college degree more than they did among non-college-educated Hispanics, African Americans, or members of other races, and far more than they did among college-educated adults of any race
the new CBPP study challenges the frequent assumption that government anti-poverty programs primarily benefit minority communities
African Americans, Hispanics, and members of other races without advanced degrees confronted even higher poverty rates than working-class whites. But they didn’t gain quite as much from the federal anti-poverty programs.

That benefit programs help more whites than individuals in any other group is something anyone could determine by looking at population: whites are far and away the largest racial group.

The finding that they help more whites than individuals in other groups in percentage terms however? That is a finding that anyone with any background in these statistics knows is screwy on its face. And indeed, when you look at the actual CBPP report, you find that’s not really true.

To see where the thing goes off the rails, consider these numbers pulled from the report:

On this measure, 10.7 percent of white people without a college degree are pulled out of poverty by transfer income (24.3 minus 13.6). For blacks, it is 18.7 percent. On its face, a significantly larger portion of working class blacks are lifted out of poverty by benefit programs than working class whites are.

So how does CBPP and then Brownstein reach the opposite conclusion? What they do is they divide the percent pulled out of poverty, not by the overall population of the group, but instead by the number of people in market poverty. So 10.7 is divided by 24.3 to give you 44 percent for whites. Then 18.7 is divided by 43.1 to give you 43.4 percent for blacks. Since 44 is greater than 43.4, it is demonstrated that actually transfer programs benefit whites more even in percent terms.

I’ll leave it to you to decide whether you think this transformation makes sense. But in making that decision, consider the following: under this metric, a population whose market poverty is 1% and disposable income poverty is 0.5% is helped more than a population whose market poverty is 100% and whose disposable income poverty is 51%. Does that make sense to you? Is that what anyone in the real world means when they talk about who government benefits help the most?

The deception on this is even worse than the above suggests. The CBPP post only analyzes people with less than a college degree. So what they are comparing is the segment of each racial group that has less than a college degree. There is nothing wrong with doing that, but if you are trying to decide who is helped more by the benefits, it is necessary to note that the share of black and hispanic people without a college degree is considerably larger than the share of white people without a college degree. So a much greater percentage of black and hispanic people are in the non-college-educated bucket that generally receives these benefits.

In fairness to CBPP, they are not the only ones who have tried this last move recently. The Center for American Progress did the same thing just two days ago when it released a post saying that “Immigrants access fewer public benefits than U.S.-born individuals.”

But as the subhead of the graph notes, they achieve this outcome by cleverly only comparing low-income natives to low-income immigrants, without mentioning that the share of immigrants who are low-income is larger than the share of natives who are low-income. Comparing all immigrants to all natives actually shows immigrants are slightly more likely than natives to be receiving benefits of one sort or another, but it’s not that much higher and the benefits they receive tend to be less generous than the ones natives are receiving (see my SIPP analysis from 2015).

In the age of Trump, perhaps these cleverly-crafted alternative facts are useful (though I doubt it). But for the audience at home, I think it is important to continue to understand that the welfare state actually does a good deal to smooth out racial inequality, which is one of the reasons it is so good and so cool.