David Brooks is wrong about poverty and the safety net.

This passage from David Brooks’ Times column today got my attention:

The United States spends far more on education than any other nation, with paltry results. It spends far more on health care, again, with paltry results. It spends so much on poverty programs that if we just took that money and handed poor people checks, we would virtually eliminate poverty overnight. In the progressive era, the task was to build programs; today the task is to reform existing ones. [Emphasis added]

I’m not exactly sure what David Brooks is saying here. My best guess is that he’s pointing out that the government spends a lot of money on in-kind (non-cash) benefits to the poor, but virtually none of these programs (think Medicaid or food stamps) are counted as income by the Census for the purposes of calculating the official poverty rate (though the new Supplemental Poverty Measure does incorporate some of them). Brooks implies that if we just converted these benefits into checks, poverty would be history.

The first problem with this perspective is that it’s purely definitional. We can define “poverty” and “income” however we want: if “poverty” were decreed to be any income under $1 a week I dare say the U.S. would be free of it. It just so happens that the Census defines income in a way that includes wages and unemployment compensation, but excludes realized capital gains and the earned-income tax credit. Converting in-kind programs into cash programs as a way of addressing poverty is essentially just redefining the income measure without fundamentally making people better off (and I emphasize “fundamentally” because of the important economic point that if you assume people are rational decision makers, then they’ll be happier on the margin if they receive cash instead of in-kind benefits).

The second problem with Brooks’ argument is that it’s wrong on the merits. The Census has a nifty tool that allows you to play around with its definitions of poverty and income and see how they affect the resulting poverty rate (hat tip to Scott Winship for pointing me to this oh-so-many moons ago). You can include as income the fungible values of all the in-kind government programs that the government normally excludes. It turns out that even counting as income all of the major in-kind and cash transfers aimed at the poor would only reduce the poverty rate by about a third, far from eliminating it as Brooks implies:

And expanding the definition of income ignores the strong arguments for adding in expenses that the Census currently excludes, among which are taxes and out-of-pocket medical bills.

While there’s a robust discussion among social policy experts about the best definitions of poverty and income, the bottom line is that under the most plausible definitions, we would need additional resources devoted to anti-poverty programs to see significant reductions in the poverty rate.