Krugman and DeLong mount a chivalrous defense of IS-LM
I’ve posted a bunch of critiques of IS-LM, but naturally Tyler Cowen’s criticism (which I agree with) got more attention. Brad DeLong had this to say:
The right thing for Tyler to have said, from my perspective at least, would have been that IS-LM does not provide us with enough insights to satisfy us, and here is a slightly more complicated model–a four-good or a three-good two-period model–that actually helps us think coherently about (some of) the issues of nominal versus real interest rates, short-term versus long-term interest rates, safe versus risky interest rates, moral hazard and adverse selection in the bond market, non-interest bearing and interest bearing assets, liquidity and means of payment, flows and stocks, expectations, government reaction functions, and so forth.
Both DeLong and Krugman insist that macro needs to start with simple models. I agree. And that those models must at a minimum include money, bonds and output. Here I don’t entirely agree. I think a model with money and goods, plus sticky prices, can get at many of the key features of the business cycle. BTW, I am not envisioning a model with constant velocity; I agree that would be almost entirely useless. But I’m willing to provisionally go along with the three market minimum for reasons that DeLong lays out here:
But the mechanical quantity theory is simply wrong for us today: the Fed has tripled the monetary base since 2007, and yet the flow of nominal spending has not tripled: not at all. IS-LM at least starts you thinking about the issues around the concept that has been called the “liquidity trap” which the mechanical quantity theory does not.
A quantity theoretic monetary model need not be the mechanical quantity theory. So I see DeLong making a pragmatic argument here. He’s saying that thinking in quantity theoretic terms is likely to lead us astray. We know that V might change, but we are likely to forget that problem when thinking about policy options at the zero bound. Fair enough. But this criticism applies equally to IS-LM, which is also likely to lead one astray, especially at the zero bound.
The IS-LM model led economic historians to argue money was easy in 1929-30, because rates fell sharply. It led modern Keynesians to assume that money was easy in 2008, because rates fell sharply. And IS-LM proponents underestimated the importance of monetary stimulus in late 2008, because they thought the IS-LM model told them that monetary policy is ineffective at the zero bound. Brad DeLong himself was one of those IS-LM proponents who underestimated the importance of monetary stimulus in late 2008. Now he’s bashing the Fed almost every day.
Some IS-LM defenders argue that there is nothing wrong with the IS-LM approach; it’s just that the model is misused by its supporters. After all, there has to be some sort of general equilibrium in the goods, money, and bond markets. The markets all interact with each other. And the IS and LM lines merely depict that general equilibrium. Yes, but a model that general would be pretty useless. IS-LM proponents also tend to argue that the IS curve is downward sloping. Nick Rowe recently argued that it is upward sloping. I think Nick’s right, at least if we use the yield on T-securities as “the interest rate,” and use a time frame that is relevant for business cycle analysis (a few months or years.) The problem is that most Keynesians identify changes in monetary policy by changes in interest rates, and hence misidentify monetary shocks.
So has Nick “fixed” the problem with IS-LM? Not really, because it’s a mistake to think of their being a “true” IS-LM model, untainted by the misuse of its adherents. Models aren’t out there in some Platonic realm, they are tools created by humans. The value of any model is instrumental, not intrinsic. If IS-LM is misused by almost everyone, then ipso facto, it’s not a good model.
Paul Krugman makes an anti-elitist argument in favor of IS-LM:
Here’s the problem: Macro I (that’s 14.451 in MIT lingo) is a quarter course, which is supposed to cover the “workhorse” models of the field – the standard approaches that everyone is supposed to know, the models that underlie discussion at, say, the Fed, Treasury, and the IMF. In particular, it is supposed to provide an overview of such items as the IS-LM model of monetary and fiscal policy, the AS-AD approach to short-run versus long-run analysis, and so on. By the standards of modern macro theory, this is crude and simplistic stuff, so you might think that any trained macroeconomist could teach it. But it turns out that that isn’t true.
. . .
Now you might say, if this stuff is so out of fashion, shouldn’t it be dropped from the curriculum? But the funny thing is that while old-fashioned macro has increasingly been pushed out of graduate programs– it takes up only a few pages in either the Blanchard-Fischer or Romer textbooks that I am assigning, and none at all in many other tracts – out there in the real world it continues to be the main basis for serious discussion. After 25 years of rational expectations, equilibrium business cycles, growth and new growth, and so on, when the talk turns to Greenspan’s next move, or the prospects for EMU, or the risks to the Brazilian rescue plan, it is always informed – explicitly or implicitly – by something not too different from the old-fashioned macro that I am supposed to teach in February.
I think Krugman’s right that real world policymakers use IS-LM to frame the issues. And to me that’s precisely the problem. The policymakers understand the basic IS-LM model, but not its weaknesses. They think there is “a” fiscal multiplier, ignoring monetary policy feedback. They think that low rates mean easy money. That’s why when I started arguing that money became ultra-contractionary in late 2008 I was regarded as something of a kook. Policymakers also tend to assume the Fed is out of ammo at zero rates. Where does this crazy idea come from? Krugman constantly like to praise Hick’s 1937 model, but in that paper Hicks said that the liquidity trap was the only revolutionary idea in the entire General Theory. The rest was putting already understood concepts (i.e. money demand depends on interest rates, or wages and prices are sticky) into a different language. There’s no question that the liquidity trap view comes from IS-LM, even its supporters admit that. And the liquidity trap view that is out there in the real world is the main reason we are letting central banks off the hook, the reason Obama thinks the Fed has “shot its wad.”
We don’t need policymakers that rely on IS-LM; we need policymakers that rely on cutting edge macro. Who rely on arguments for why level targeting is an extremely powerful tool at the zero bound. Those should be the standard model, if we insist on teaching our policymakers a standard model. We need useful models, not models that fulfill our urge map out a 3 market general equilibrium framework.
Here Krugman trashes Tyler Cowen:
Brad DeLong comes down hard on Tyler Cowen over his attempt to critique the IS-LM model “” but not hard enough.
. . .
In macro “” or at least macro that tries to get at monetary and fiscal issues “” what you need, at minimum, is to understand an economy in which there are three goods: money, bonds, and economic output.
. . .
There’s something about macro that seems to invite this sort of thing: more even than the rest of economics, macro seems afflicted with people who mistake confusion for insight, who think their own failure to understand basic ideas reflects a failure of those ideas rather than their own limitations.
Tyler shouldn’t feel too bad about this. After all, Krugman doesn’t identify a single flaw in Tyler’s critique. The post is just a string of personal insults. And recall that Michael Woodford creates models without money, so he’s also “confused.” And of course Milton Friedman was no fan of IS-LM—so he’s another guy who just doesn’t get it.
I favor an ad hoc approach to models–use the simplest model that gets at the issues you are interested in. Start with a simple economy with money and goods, no bonds. The supply and demand for money determines the price level and/or NGDP. That’s most of human history. Add wage price stickiness and you get demand-side business cycles. Add interest rates and you get . . . well it’s not clear what you get. Interest rates almost certainly have an influence on the demand for money. Do they play a major role in the transmission mechanism between money and aggregate demand? Hard to say. Short term Treasury yields probably don’t have much impact. Other asset prices might, but then there is generally no zero bound for other asset prices. On the other hand monetary policy often operates through purchase of short term T-securities. Bottom line, it’s complicated.
Now let’s add another asset, NGDP futures contracts. Now the modeling process gets much easier. We model monetary policy as changes in the price of NGDP futures contracts (accomplished through central bank purchases of financial assets in order of safety and liquidity, as much as it takes.) Then we have a Philips Curve or SRAS curve to translate NGDP shocks in fluctuations in real output. Since NGDP futures prices are monetary policy, fiscal policy is 100% classical.
Some will object that we don’t have NGDP futures contracts, so we are currently forced to stop at the money/bonds/goods stage of human progress. Not so, we can construct a pseudo-NGDP futures price by modeling expected NGDP as a function of lots of variables (past NGDP, current asset prices, TIPS spreads, consensus forecast of economists, etc.) That pseudo-NGDP futures price is available to Fed officials in real time. They can peg it if they want to. The policy has flaws related to the circularity problem (which NGDP index futures convertibility does not have), but it’s workable.
Of course you’ve probably noticed that this is also my model. I think it’s also in the tradition of Milton Friedman, although obviously it differs in certain respects. Friedman thought it was more useful to take a partial equilibrium approach to macro. By doing so he was able to avoid the mistakes of those who looked at the Depression from an IS-LM perspective. He was interested in how monetary policy determined NGDP, and then used a separate Phillips Curve approach with a natural rate to explain output fluctuations, to partition NGDP into RGDP and P. He viewed interest rate movements as a sort of epiphenomenon. Monetary policy affected rates in a complex way, which made interest rates an unreliable indicator of the stance of monetary policy.
Of course the indicator Friedman choose, M2, also turned out to be somewhat unreliable, which is why I replaced it with NGDP futures. Expected NGDP (or something similar that incorporates the Fed’s dual mandate–like the Taylor Rule) is the goal of monetary policy. There’s quite a bit of slack between changes in M2 and changes in expected NGDP. In contrast, changes in the price of NGDP futures contracts ought to track changes in expected NGDP (the policy goal) pretty closely. I find the NGDP perspective to be much more useful than the interest rate perspective.
BTW, I think this might have been what Tyler Cowen had in mind here (first Tyler, then Brad):
“The most important points… one can derive from a… nominal gdp perspective…”
What is this “nominal GDP perspective”? The Google reports that as of this writing the phrase “nominal GDP perspective” appears only once on the internet–in Tyler’s post.
I want all 5000 of my readers to Google “Scott Sumner nominal GDP perspective” 100 times. Each time, please link to my blog if it appears on the list.
I think DeLong and Krugman need to lighten up a bit. They are defending the IS-LM model like it’s some sort of bride whose virginity has been challenged. Models are only valuable if they are useful. We critics are convinced that other approaches are much more useful. Contrary to DeLong, there is nothing “tribal” about all this. I’ve discarded the old monetarist preference for M2 targeting, and accepted the Krugman argument that temporary monetary injections are ineffective at the zero bound. I’m not tribal, I’m eclectic. When we see people use models in ways that we think are wrong, indeed that we think helped cause the Great Recession, we are naturally going to be critical of those models. Especially if we find alternative approaches that seem more fruitful—like viewing monetary policy through the lens of changes in NGDP expectations, and viewing fiscal policy in essentially classical terms (except where a bizarrely perverse central bank allows fiscal decisions to alter its inflation or NGDP target.)
PS. Yes, I do understand that the strongest criticism of my approach is that we do live in a world with bizarrely perverse central banks. We’ll fight that issue another day.