November 2020

Sun Mon Tue Wed Thu Fri Sat
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28
29 30          
Blog powered by Typepad

« Changing Landscape of Academia | Main | Lewis Lehrman on Deficits, Debt and Debasement »

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Don't forget your Bastiat, Steve. The seen and the unseen. Think in counterfactuals. This graph proves nothing except that forecasting is hard.

What I see is that the problem was of magnitude, that is to say the situation was much worse than originaly thought.
That said, the curves (the dark blue one and the red dots) both start to bend a tthe same time

Too many variables.

If forecasting is hard and we misread the magnitude and there's too many variables, doesn't that suggest caution before we run off spending $787billion to "solve" the problem?

who says there was not caution?

Daniel Kuehn: if (I agree) that graph proves nothing regarding the effectiveness or lack thereof of the stimulus, it proves an even worse thesis: politicians are unaccountable, and cannot be checked by democratic control because there is no feedback mechanism informing voters whether politicians are doing the right thing or not.

The essence of the claim that the stimulus package created or saved jobs (net) is a counterfactual (theoretical) experiment.

To do this well you need the correct model of the economy such that when you omit stimulus you get a picture of the effects and then when you include stimulus get get another picture. The latter should track the actual economy in this case.

The discrepancy may be explained by either the economy being in worse shape than economists thought at the outset. In this case the model of the economy without stimulus is wrong.

Alternatively, the discrepancy can be explained by the fact that the model generating the effects of stimulus is wrong.

In either case, our confidence is not boosted.

The debate over Say's Law has never been settled, and few economists realize they are replaying it again over the stimulus.

Mario -
Our confidence is not boosted, but then again I don't think anyone who knew anything about these issues before hand ever expected arbitrating between those cases to be easy. So while our confidence is not boosted, it shouldn't really be reduced either, should it? I think you lay out the issues accurately - I just don't think it's some great failure to begin with. And it is notable that some of the biggest stimulus advocates - Krugman, for example - said last January that these forecasts looked way too rosy, even by his simple back of the envelope calculations. Nothing decisive - it still doesn't establish the right counterfactual for us - but it's an interesting thing to note.

Steve -
I would agree completely. Of course we should proceed cautiously. I'm strongly against reckless, rash policy making - I just think you and I may have different definitions of what is reckless.

Yeah the graph only says the foreceasters were wrong. It doesn't resolve the "stimulus sucks" vs. "stimulus is awesome but we haven't done enough of it yet" argument.

Oh come on, Daniel. The point is that we should not take the guy in the white lab coat too seriously. I think Steve posted the graph to say that we have a certain tendency to think the experts are wise and able to pierce the fog of future, whereas they are not. We should be afraid of any doctor -- medical or economic -- who prescribes strong medicine on the basis of an unreliable diagnosis. Saying the stimulus was too small is a bit like saying that my magical incantation keeps dragons away. You don't see any dragons around here, do you? So, yes, if you are committed to the view that the stimulus was needed, then the graph will not dissuade you. But it counts for Steve's POV to point out how wrong the good economic doctors were.

Mario Rizzo's observation re "measuring" aggregate effects of stimulus is exactly on target.

But doesn't the debate over aggregates miss the more important point? Even if stimulus spending lowers unemployment and raises GDP over the short run, is that necessarily desirable? I would suppose the goal should be eliminating false price signals, capital distortions, dodgy finance etc. such that the economy returns to wealth creation. Even if stimulus could be shown to push GDP and unemployment to pre-recession levels, it would not follow that it had been successful.

It never gets old!

"politicians are unaccountable, and cannot be checked by democratic control because there is no feedback mechanism informing voters whether politicians are doing the right thing or not."

feedback mechanism or not, those who make the important decisions on the ground are unaccountable to the voters. anyone who has taken federal administrative law understands this. democratic theory of justification is a farse.

Hi - I'm the guy who made the graph. I look at it a bit differently. My original motivation for making the graph was to show that Romer et al's timelines for pumping money into the economy were ridiculously naive, and exemplary of the lack of experience of Obama's economic team.

There was no way that the federal government could allocate and disburse the stimulus funds, which would then be allocated and obligated by the state governments, which would then be invoiced by the "shovel-ready" contractor - all in one or two quarters. Anyone with the least bit of familiarity with federal and state procurement knew that from the outset. And we've seen that to be true in the ARRA spending rates.

As far as whether the stimulus helped or harmed matters: I personally don't believe that we had a choice in spending money on extended unemployment benefits, or in supporting the states' Medicare/Medicaid obligations. The problem is that it was called "stimulus," which it clearly was not.

The remaining money, spent on tax breaks and bonafide attempts at stimulus projects, has been poorly conceived. It represents short-term spending that is propping up a small sector of business while waiting for the rest of economy to heal itself. It is not a program that ever had a chance of jumpstarting the economy.

In my opinion, while the stimulus has had a weakly positive effect on businesses, it has been the Obama administration's business-hostile rhetoric and programs that have had a large dampening effect. We've seen several polls of business owners showing that they are very wary of the future business climate, and so are reluctant to add staff.

Finally, let me say that if you read the report that spawned the graph, you may come to the same conclusions that I did:
1) Romer et al.'s analysis consisted of a 1-page spreadsheet based on gross assumptions and generalizations. From that she created her graph.
2) That graph, completely lacking in rigor, was the basis of the "3.6 million jobs saved" claim, which sold (oversold) the program. It was a terrible example of peer review and press scrutiny, and a terrible way to justify the spending of $787 billion.
3) Those who say "all that graph proves is that economic predictions are difficult" are completely wrong. Mark Zandi (Moody's Economy) ran an actual economic model at the same time, and his "without stimulus" projection has been very accurate (I'll update his prediction vs. reality chart next month). Romer's back-of-the-envelope calculation was invalidated within a month. The graph proves that wildly optimistic, poorly supported, economic predictions are crap.

The stimulus package actually provided the ground conditions that made Zandi's model successful. His model would not have been able to predict the widespread economic problems that would have occurred had more than 5 million people been left without unemployment benefits.

Geoff: Was Zandi's model a projection "without stimulus" or did "the stimulus package provided the ground conditions" for it?

If a model designed to describe the economy without the stimulus replicates the outcomes of the economy after the stimulus, it looks like an argument for its lack of effectiveness, provided that (that's a point raised several times in the comment thread) reliable conclusions can be drawn out of the data.

Was Zandi's model a projection "without stimulus" or did "the stimulus package provided the ground conditions" for it?

Zandi projected both the "with" and "without" stimulus cases, though he didn't know where the stimulus money was going at the time of his projection. Here's his prediction compared to reality for last quarter. Based on the results thus far this quarter, it looks like the 1Q10 number will be around 10%, so his "without stimulus" prediction will start to overpredict unemployment.

But it is clear that for the first year he was able to fairly closely track the rise in unemployment to 10%.

Of course the relevance of the unemployment numbers themselves are of some concern.

Geoff:

So the first year of with-stimulus has looked like Moody's prediction for without-stimulus, and you see this as success? What did the actual with-stimulus Moody projection look like for the first year - was it identical to the without-stimulus one?

Wasn't the pro-stimulus argument that the stimulus would begin to take effect in less than a year not more than a year? Because, correct me if I'm wrong, if it was going to take more than a year to even begin to take effect why bother? There is little reason to believe that the economy would not begin to heal itself in that period (think 1920-1921 "depression" or more recently 1990-1991 or 2001-2002).

This is why the Obama administration had Romer and Bernstein project the disastrous 2009 effects of not-stimulating. If they had told the country that 2009 would look the same either way - 10 % unemployment - but if they did not pass a huge stimulus package then in mid-2010 or 2011 unemployment would fail to drop I think the people would be skeptical. They would realize that by mid-2010 or 2011 the economy would recover on its own.

So the first year of with-stimulus has looked like Moody's prediction for without-stimulus, and you see this as success?

I see it as success of his model, I certainly don't see it as success for the stimulus portion of ARRA.

Wasn't the pro-stimulus argument that the stimulus would begin to take effect in less than a year not more than a year?

That is exactly the concern that led me to make that graph. As I said - I didn't believe the unrealistic timelines that were used to sell the program.

Hmmm. Looks like my attempt to link the Zandi comparison didn't take in the earlier comment. Here's the URL: http://michaelscomments.wordpress.com/2010/01/08/an-accurate-unemployment-projection/

Dang, do HTML codes just not work here? All my italics and links have come to naught.

Sorry Geoff, no html in the comments, though links are picked up automatically.

the discrepancy can be explained by the fact that the model generating the effects of stimulus is wrong.

The comments to this entry are closed.

Our Books