Jobs saved and counterfactuals

Brad DeLong catches Allan Meltzer claiming that there’s something nonsensical about the Obama administration making estimates of jobs saved thanks to its policies. But it’s not just Meltzer — Greg Mankiw has done the same thing.

They should be ashamed of themselves.

The Obama administration’s “jobs created or saved” is just a way of saying “other things equal” in non-economese. Of course it makes sense to ask how many more people are working than would have been the case without a given policy — and every administration makes assertions along those lines. During the 2001 recession and its aftermath, how many times did the Bush administration claim that the recession would have been worse without its tax cuts? And while many of us quarreled with that claim, I don’t think I ever argued that other-things-equal arguments are nonsense on their face.

The willingness of conservative economists to fall in line behind such cheap shots says something sad about them, not about the Obama administration.

Yes, counterfactuals in economics are important. But they’re hard to do. Very hard to do and prone to bias. They’re a great place for faith-based econometrics.

Back in September when the administration claimed ONE MILLION jobs saved or created, this was the methodology:

Romer said the White House advisers came up with their estimate using two methods: An analysis of the law’s effect on the national economy and employment using historical data and another using a statistical model of the effects of tax cuts and government spending on job and economic growth. Independent economists’ estimates and data from other countries that implemented their own stimulus plans supported the administration’s conclusions, Romer said.

Does anyone (even Krugman or DeLong) think that the $100 billion or so of stimulus money that had been spent at that point had really saved or created one million jobs? (I’m ignoring the “tax cuts” which were mostly saved.)  It’s possible of course. But does anyone think that the claim of one million was the result of careful objective analysis to control for everything necessary it would take to measure the number of jobs that would exist in the absence of the stimulus? And it’s not even clear that the exercise is econometrically possible. Yes, economists try to estimate things like the multiplier and the estimates vary widely for purely statistical reasons and not just the challenge of measuring confidence say, or why all government spending is not alike or any of a myriad of other challenges.

Then there’s this lovely story from the Sacramento Bee:

Up to one-fourth of the 110,000 jobs reported as saved by federal stimulus money in California probably never were in danger, a Bee review has found.

California State University officials reported late last week that they saved more jobs with stimulus money than the number of jobs saved in Texas – and in 44 other states.

In a required state report to the federal government, the university system said the $268.5 million it received in stimulus funding through October allowed it to retain 26,156 employees.

That total represents more than half of CSU’s statewide work force. However, university officials confirmed Thursday that half their workers were not going to be laid off without the stimulus dollars.

“This is not really a real number of people,” CSU spokeswoman Clara Potes-Fellow said. “It’s like a budget number.”

Macro is hard to do. Pointing that out isn’t shameful.