Unemployment Benefits Create Unemployment Like Aspirin Creates Headaches

The argument that paying benefits to unemployed people will keep them from seeking real jobs has been around for a long time. It was, indeed, one of the leading arguments put forth during the Great Depression to oppose the system of unemployment insurance that was eventually adopted. It’s hard to square with today’s politics, but there was a time when this argument was made by labor unions themselves. “The dole system develops a paternalism that is demoralizing and destructive. It stifles ambition, destroys initiative and blights hope,” said Philip Green, president of the American Federation of Labor, in 1930.

In the ensuing three-quarters of a century, the nation has at times come pretty close to full employment, which ought to make clear the point that almost everyone who is capable of working would prefer to have a job than to receive government benefits. But little is ever settled in the world of economics, and so we face arguments like the one atop today’s Wall Street Journal opinion page, “The Folly of Subsidizing Unemployment.” Harvard economics professor Robert Barro makes the 21 st -century version of the argument: It’s not that paying people who aren’t working is in and of itself destructive; it’s paying them to do so for too long a period. Specifically, his beef is with the government decision to extend benefit eligibility to 99 weeks, which he labels “a welfare program that resembles those in many Western European countries.”

Barro burrows into government data to produce some interesting historical and contemporary facts about joblessness, but the calculation he performs with them is worthy of a cheap stage magician. He focuses specifically on how long the average person receiving unemployment has been getting it and the percentage of jobless who’ve been out of work for six months or more. And this is where the so-called thesis kicks in: “The duration of unemployment peaked (thus far) at 35.2 weeks in June 2010, when the share of long-term unemployment in the total reached a remarkable 46.2%. These numbers are way above the ceilings of 21 weeks and 25% share applicable to previous post-World War II recessions.”

I’m with him that far; the 46 percent number in particular is chilling. But then he says: “The dramatic expansion of unemployment-insurance eligibility to 99 weeks is almost surely the culprit.”

“Almost surely” is a handy construction, isn’t it? He’s fudging the language because his argument cannot hold water. First of all, a large number of unemployed people did not have access to the 99-week eligibility until President Obama signed a retroactive measure into law on July 22 . A bill signed into law in July almost surely could not have had a profound effect on the millions of unemployed people who’ve been out of work six months or more. Second, even the 99-week benefits that went into law earlier were not uniform among all the states, so Barro would almost surely have to pull out millions of people from his presumed numbers of people collecting unemployment if he wanted them to be accurate.

But most importantly, this argument rests on one of the most time-honored logical fallacies: post hoc ergo propter hoc . These days, commenters on Slate prefer to say “correlation does not mean causation.” Put simply, the fact that a 99-week extension took place at sort of the same time as persistently high unemployment does not mean that it caused it; in fact, the causal arrow—based on what Barro has presented—may well be in the other direction. Or the causation could come from somewhere else (such as an economy that’s not creating enough jobs).

This doesn’t mean, necessarily, that Barro’s argument is wrong; extending unemployment benefits probably does keep some percentage of jobless people from seeking full-time work. But it’s harder to take that argument seriously when it’s made so speciously.