David R. Henderson  

Mankiw on the CEA Stimulus Study

The Somali Model... What is the Fed Thinking, 2?...

Arnold laid out one of the two ways the CEA estimated the effects of Obama's "stimulus" package. Greg Mankiw has a particularly nice statement about the other way. Mankiw writes:

That is, the CEA took a conventional Keynesian-style macroeconomic model and used those set[s] of equations to estimate the effect the stimulus should have had. Essentially, the model offers an estimate of the policy's effect, conditional on the model being a correct description of the world. But notice that this exercise is not really a measurement based on what actually occurred. Rather, the exercise is premised on the belief that the model is true, so no matter how bad the economy got, the inference is that it would have been even worse without the stimulus. Why? Because that is what the model says. The validity of the model itself is never questioned. (italics added)

Greg was a little too easy on them, though. He went on to write:
In the end, I do not find this CEA document very persuasive. At the same time, I feel the CEA's pain. The stimulus act instructed them to do the (nearly) impossible.

Sure it did, but it didn't instruct the CEA not to point out some qualifiers. And it didn't instruct them not to mention that they had way overestimated the effect of the stimulus package 18 months earlier.

Comments and Sharing

COMMENTS (9 to date)
Daniel Kuehn writes:

"And it didn't instruct them not to mention that they had way overestimated the effect of the stimulus package 18 months earlier."

Could you explain how you're aware of the fact that they overestimated the effect of the stimulus package and didn't, say, underestimate unemployment? I'm at a loss for how you're privy to that information.

Daniel Kuehn writes:

And I should note that I'm only making the point that Greg Mankiw is making - that we have no reliable way of testing any of these theories. You're criticizing the CEA for making claims on paltry evidence - and I agree they should be qualified - and then you turn around and make the same heroic claims on paltry evidence without qualifying your claims.

You can't get a grip on the counterfactual any more firmly than they can, David. So don't write posts lecturing them on how they should qualify what they write, when you barrel ahead without qualifying what you write.

Doc Merlin writes:

The burden of proof for an opinion piece or blog post is significantly lower than for a paper. Blog posts and opinion pieces are meant to start the dialog, papers are supposed to be very strongly backed up and polished.

Daniel Kuehn writes:

Doc Merlin -
Right... and I'm agreeing with Mankiw and Henderson on the lack of polish in the CEA paper, and then I'm engaging in a dialogue on the blog post. Aren't I?

What's the issue exactly? I agree with you completely.

David R. Henderson writes:

@Daniel Kuehn,
Good point. Attribute it to my writing it late at night while I was tired.

david (not henderson) writes:

Daniel Kuhn says:

"You can't get a grip on the counterfactual any more firmly than they can, David."

Exactly. Perhaps then we should rely on a combination of a) theory (not favourable to fiscal stimulus), instead of faux-models, and b) predictive accuracy. The pro-stimulus camp said unemployment would be x% with the stimulus and now it's greater than x.

At a minimum, sticking just to empiricals, we can say we don't know fiscal stimulus works, it doesn't appear to have and we know for certain that it has added significantly to the fiscal budget problems. So why do it again? Sounds like the proverbial triumph of hope over experience.

Daniel Kuehn writes:

David (not henderson) -
I think it is a combination of theory and predictive ability, and maybe some empirical work on past episodes (we can't get a counterfactual on this one episode, but we might be able to get something somewhat plausible by looking at an expanse of time in the past).

As for predictive accuracy - one thing this post fails to mention is that while the predictive accuracy on unemployment has been off, it has been pretty good for GDP. And output is the primary mechanism through which stimulus is expected to work, so that's notable.

So there are two solutions if, like me, you think that past experience and theory gives us reason to believe stimulus will work:

1. Unemployment would have been a LOT worse without, but unfortunately we can't say one way or another because we don't have a counterfactual (I actually think this is plausible, but obviously we're up a creek as to determining it either way), or

2. Argue that stimulus did work for output, but that for some reason Okun's law isn't holding true in present circumstances, which isn't allowing the impact on output to work through to jobs.

I think the second is the more fruitful way to go. I think there probably was some effect on employment, but the bigger story here is that the output-employment relationship is behaving in an unexpected way, not that the stimulus-output relationship is behaving in an unpredictable way.

Why is Okun's law seemingly being flauted? That's the million dollar question.

Daniel Kuehn writes:

And I should note - this gets back to what I think was Hayek's most important critique of Keynes - that he didn't spend enough time on the relationship between output and employment. Generally speaking, I think the Keynesian case is only strengthened by everything that we've been seeing in the last couple years. This, though, is the area where we could do a little more thinking.

John V writes:

This is why economics is the overly contentious field that it SHOULDN'T BE.

Politics, Politics, Politics. I hate it.

What other field of somewhat scientific study is almost totally driven by so many priors assumptions and biases and so easily twisted and manipulated for political gain?

None. Clime Science? Not even close.

You really don't need a PhD in economics (or even a bachelor's) to understand what happens in this field with those who have an expert-level grasp of it and those who need it to say what they want it to say.

The models in this case are indeed inadequate. They almost always are. When you really stop and think of what does and doesn't go into the models, it's a wonder anyone takes them seriously at all...or at least serious enough to allow billions and billions of dollars to ride on them. It's no wonder it's advocates will defend the results somehow...how could they not? That's an expensive blunder for which, "It wasn't really ironclad and we were wrong" just doesn't cut it.

Comments for this entry have been closed
Return to top