Ross Douthat recently made this observation:

1.) The modern welfare state has succeeded in substantially cutting our country’s poverty rate. This is a point that both right and left sometimes obscure, the right because it complicates a simple “we fought poverty, and poverty won” narrative about the Great Society, the left because it complicates claims that Reagan or Gingrich gutted welfare spending and crushed the fortunes of the poor. But the basic evidence seems very convincing: Whether it’s Scott Winship analyzing the numbers from the center-right or Harvard’s Christopher Jencks doing the same from the center-left, you can see dramatic reductions in the poverty rate since the 1960s, with various public programs, means-tested and otherwise, pretty clearly playing a substantial role.

I agree with much of what Winship and Jencks had to say, but I do have one important reservation about this sort of claim. There’s a tendency to assume that the effect of anti-poverty spending can be measured by looking at the difference between market income and market income plus government aid. I’m going to argue that this is a very serious conceptual error, even if their conclusions end up being correct in the end.

Imagine someone who makes $8000 in 1960, working as a maid in a hotel. Ten years later the War on Poverty has been launched, and this lady now earns zero market income. She instead receives $10,000 in various government benefits. (I’m using 2015 dollars for simplicity; of course incomes and prices were far lower back then.) In the meantime, the wages of autoworkers rose from $40,000 to $60,000/year. If that were the data you had to work with, and that lady’s situation was typical of the poor, how would you evaluate the War on Poverty?

Taking the optimistic view, one could argue that this lady was kept out of abject poverty by the government aid. Without that aid, her income would have been zero. On the other hand if she was representative of the poor, you could say that the 25% rise in her income plus benefits was less than the 50% increase seen by middle class autoworkers. Here’s where the tricky counterfactuals come in. What would have been the 1970 income of the former maid if those welfare programs had never been established?

Here I’d like to switch from income to consumption. Economic theory says income is not the right measure of well-being, we need to look at consumption. Some of my critics insist the distinction is not that important for the poor, because they can’t afford to save anything. I’m not sure that’s true (it’s certainly not true in China, or among illegal Chinese immigrants to America) but I’ll grant that assumption if you’ll agree that consumption is the standard measure of well-being used in economics. So lets say income equals consumption of the poor.

Do you see the problem? If we return to our no War on Poverty counterfactual, and assume no government benefits and no effect of incentives on work effort, then everyone with no market income would have had zero consumption. Which is obviously impossible. I’ve oversimplified in all sorts of ways. People can get by for a while with aid from relatives. And there was some welfare even before the War on Poverty. Nonetheless, I think it’s pretty hard to take seriously the claim that the market income in 1970, or 2015, is the market income that would occur without welfare programs. We know how poor people (and mildly disabled people) behaved in the 1950s, and it was not the same way they behave today.

The bottom line is that the data alone cannot disprove the claim that (absent the war on poverty) that maid would have continued to work in a hotel and have earned $12,000 in 1970, receiving the same 50% wage increase as the autoworker. Yes poverty fell between 1960 and 1970, but that’s because America was getting richer. (Obviously the reduction in poverty with the elderly due to Social Security is another story, where my contrarian argument is much weaker.)

Given all this uncertainly, I might be inclined to take a fairly agnostic stance, and simply look at how the bottom 20% are doing. Did their economic wellbeing rise faster or slower than for the average American? If faster, that provides some (weak) indirect evidence against the hypothetical counterfactual I provided above.

Surprisingly, despite my methodological reservations, I actually end up agreeing with Douthat, Winship and Jencks; the War on Poverty probably helped reduce poverty (although by far less than Douthat, Winship and Jencks claim). But I get there in a rather surprising way, which may not please progressives. I believe that the Piketty obsession with inequality is way overdone, and that America’s poor have seen their living standards rise faster than the American middle class. (The top 1% have done very well, but they are so few in number that they hardly matter.) So inequality is becoming less of a problem for the 99%. That suggests to me that the War on Poverty may have been a limited success. However, much greater progress would have occurred if more emphasis had been put on programs like the Earned Income Tax Credit, which encourage work, and also more effort on avoiding the marriage penalty built into many programs (including the EITC.)

My fear is that progressives will wrongly conclude that inequality is getting worse despite the war on poverty, and advocate counterproductive policies such as higher minimum wages and taxation of capital income. We should improve the War on Poverty, not launch populist policies based on bogus economic theories.

PS. If I am wrong, and the living standards of the poor have improved by less than the middle class, then I’d reverse position on the War on Poverty. In that case I’d very strongly doubt claims that it was even a limited success.