Has the Keynesian IS-LM model made good predictions lately?

I’ll skip context and links and cut right to the chase.  Reinhart-Rogoff and nominal gdp perspectives and TGS views also have been predicting a slow recovery, so while IS-LM has done OK here it wins no special prizes.

What about the “no crowding out” prediction?  Since at least the early to mid 1980s, it has been well-known in macroeconomics that U.S. budget deficits do not forecast real interest rates very well and that includes under periods of full or near-full employment.  Here is a brief survey by Alan Reynolds on the topic (you can follow up on his references), and he is usually considered a villain by the Keynesians and so he is hardly a Keynesian himself.

There may be a few reasons for the general lack of a connection between deficits and real interest rates in the United States:

1. The supply of capital to the United States is fairly elastic, either domestically or internationally.

2. We don’t have good identifying restrictions on the empirics in the first place.  For one thing, controlling for monetary policy is tricky.

3. We haven’t yet seen budget deficits big enough to matter.

4. We are not measuring budget deficits correctly because what matters is the consolidated fiscal stance of the U.S. government, a’la Robert Eisner.

5. Ideas related to Barro’s Ricardian Equivalence hypothesis.

Anyone — Keynesian or otherwise — paying attention to the last thirty years of empirical macro never expected much crowding out of financial capital in the first place.  It simply has not been in the cards.

To put it more bluntly, the “no crowding out” result is not much of a predictive victory for Keynesian economics, IS-LM, the liquidity trap, and so on, even though I have read it claimed as such many times.  It is a strike against some predictors who were wrong in the first place, especially in the right-wing popular press circa 2009-2010, plus some Republicans who jumped ship on the issue, perhaps because they wanted to attack Obama.

What’s a unique prediction we might look at?  It is a common Old Keynesian claim these days, at least from Krugman, that the AD curve is upward-sloping because of a liquidity trap.  That would imply that harsh and binding minimum wage hikes, and other wage-propping mechanisms, should prove expansionary.  That claim, at least for the Great Depression, has been knocked down fairly conclusively by Scott Sumner.  If there is no comparable test on today’s data, it is because we have grown that much wiser.


Looking forward to an interesting discussion on this one . . .

What is the point of discussing the predictive successes of competing models? If you want to test theories, you should focus on their failures. You can always write down a model that explains a given stylized fact. However, this model might fail in other regards. I read the defense of the IS-LM model as claiming that it is a simple model that is able to explain a quite large set of empirical regularities that can otherwise be explained by a larger set of different/distinct rather models that are rather small range in their explanatory scope. A case in point is that the post offers 5 different potential reasons for not having a relationship between interest rates and debt that all would need to be tested. In contrast, the IS-LM predicts (as far as I understand the discussion) that no crowding out requires only one thing: the zero lower bound must be binding. Hence, increases in government spending when this is not the case, should give rise to crowding out. If the empirical literatur shows that this is not the case, this a predictive failure of the IS-LM model. Failures like this are more informative than predictive successes.
Moreover, arguing that one should look at "empirical macro" does not help either. Anyone familiar with the Lucas Critique knows that historical relationships need not hold anymore under policy changes. Hence, you need a model to form predictions that are not based on potentially spurious correlation (due not structural non-invariance). However, in this regard, it is doubtful whether the static IS-LM without expectations is of much help either.

Interesting points. Scott also has some interesting arguments re #2. I think #1 is true in large part because the customers for the debt are different.

It is a strike against some predictors who were wrong in the first place, especially in the right-wing popular press circa 2009-2010

It does appear interest rates are not an argument against deficits.

If there is no comparable test on today’s data, it is because we have grown that much wiser.

Unfortunately, it doesn't seem likely a modern test would change anyone's mind either. They would just claim the interventions hadn't been large enough to generate the desired effect. Economics has become too politicized for empiricism n some quarters.

I'm curious if anyone has gone so far as to endorse Jesse Jackson Jr.'s plan to just hire the 15 million unemployed and pay them $40K each. For the crowd that doesn't like gov't inefficiency or deficit arguments and thinks AD is the central problem, it seems like the logical endpoint.

"We haven’t yet seen budget deficits big enough to matter."

I assume you have no intention of responding to Kash Mansori's demonstration that you could have indeed predicted the mysterious whimsy of the elusive bond market vigilantes at least in Europe NOT by government deficits, but by current account imbalances that Euro-zone governments could do nothing about without finding a way to impose some strict capital controls (which I assume you would also have been opposed to on principle).

As a layman and debater, my problem with the TC approach is that it gets detached from real-world, present-time choices. At some point, TC needs to fly-speck the "austerity now" and "hang Bernanke" arguments/models, as well as those of Mankiw and freshwater economists and see if they have anything to recommend them. If they get 2 stars and the Krugman approach warrants 4 stars, TC needs to make that clear. Otherwise, TC looks a bit like a not-so-relevant troll in the current context.

Thanks again for lots of interesting content, from a long-time follower.

I looked at the Scott Summer piece, and maybe there is a reason for plotting things as ln(value) but it certainly seems odd; in any event, he has some graphs, that sort of look, without any detailed statistical analysis, like they follow each other...so what.
I could probably get as good a graph with sale of diapers.
That the best ya got ?

Well, seems then that IS-LM has made excellent predictions. Compared to the ALTERNATIVES, which have predicted NOTHING.

And no, Scott Sumner has demonstrated nothing of the sort you claim he has. Krugman appears to be right. As usual.

While wage hikes might not do much for aggregate demand, what would they do for outstanding private sector debt? (particularly consumer debt).

'Ideas related to Barro’s Ricardian Equivalence hypothesis.'

Really? I mean, really? Come on.

I don't know how to include links here, so I did this elsewhere; pls excuse the [is there a word for inappropriately using the comments section of someone else's blog to promote your own blog?].



I guess Gene Fama (and John Cochrane) haven't been paying attention. (or John Taylor, Michael Boskin, Robert Barro, etc)

Interestingly, Ricardian equivalence does not rule out government expenditure being stimulative in the short run. You should read Krugman sometimes.


Professor Cowen, I read both yourself and Prof. Krugman, and sometimes feel like you're operating in two entirely different worlds.

It is a strike against some predictors who were wrong in the first place, especially in the right-wing popular press circa 2009-2010, plus some Republicans who jumped ship on the issue, perhaps because they wanted to attack Obama.

"The right-wing popular press circa 2009-2010, plus some Republicans..." are hugely important, powerful institutions in the world.

Dismissing your serious critics, as Krugman does at least on his blog, by only taking on lightweights does little to advance the state of economics as a field. But dismissing the powerful intellectual lightweights who use specious economic arguments for political reasons, as I think you're doing here, does little to advance the state of economics in the world.

Comments for this post are closed