At the recent American Economic Association meeting in San Diego, Brad DeLong chaired a panel on ” Stimulus or Stymied?: The Macroeconomics of Recessions“, and has posted a transcript. Paul Krugman was there and picked up my claim that macroeconomics has, on balance, gone backwards since 1958. I’ve extracted his section here. Lots of useful stuff, but I’d stress this:
the whole basis on which we constructed monetary policy during the Great Moderation, which is that stabilizing inflation and stabilizing output are the same thing, is all wrong: you can have a sustained period of low but not negative inflation consistent with an economy operating far below its potential productive capacity. That is what I believe is happening now. If so, we are failing dismally in responding to this economic crisis. This is in contrast to what some central bankers are saying—that we have done well because inflation has stayed relatively stable.
To push this a bit further, I’d argue that there will be no real recovery as long as central banks continue to treat the inflation-targeting polices of the (spurious) Great Moderation as the pre-crisis normal to which we should strive to return
Krugman remarks to panel on The Macroeconomics of Recessions
Let me try to talk about where I think we stand and what the fiscal-policy issues ought to be.
The basic story—at least as many of us see it—is that we had this really, really dramatic shock to private spending. This is the private-sector financial deficit: gross private domestic investment minus gross private domestic saving as a share of potential GDP as estimated by the CBO. This is not the first time in the post-WWII era we have had a big drop, but it is the biggest: 10% of potential GDP. [30:00] The previous ones in the mid-1970s and early-1980s were associated with tight monetary policy, very high interest rates, and collapses in housing investment driven by tight monetary policy—which, of course, sprang back as soon as the Federal Reserve decided that the American economy had suffered enough.
This time is different. This time it came spontaneously. This time it came in spite of drastic cuts in interest rates to essentially zero.
The question is: “What do we do?”
There is an interesting debate: “When did economics go all wrong? When did macroeconomics go all wrong?” Bob Gordon has rather persuasively made the case that it went all wrong about 1978—that we would have done a better job at macro policy if we had met this crisis with the intellectual panoply we had then and had not had the thirty years since. I saw John Quiggin just made the argument that things actually went all wrong about 1958.
If an economist from 1958 had seen what is going on now, he—and back in 1958 it would have been “he”—would have said: “OK. Private sector does not want to spend. The government should spend. This is a powerful case for fiscal stimulus to prevent this from causing a persistent slump.” We have not done that. We had some fiscal stimulus delivered for a brief period of time in 2009. We have had a fair bit of allowing automatic stabilizers to operate. But at the same time we have had quite a lot of policy austerity. We had a worldwide or at least an advanced-world turn to austerity in 2010 inspired to some extent by the lessons that were drawn—I would say mostly wrongly—from the story of Greece but then applied across the board, and also from a reversion to pre-Keynesian modes of thinking about the macroeconomy. Whatever the reasons—and there are a mixture of political-economy reasons and just plain bad-economics reasons—we made a big turn to austerity. Now we debate: “Was that wrong? How wrong was it? Should we really be doing as much fiscal stimulus as the man from 1958 would say?”
Think about the objections to stimulus. I would put them into three categories:
First, perhaps we do not have nearly as much economic slack as people like—well—me say. Perhaps there is something much more structural going on, and we do not have that much room to expand. We have a huge economic failure, but the failure is not for the most part a simple failure of aggregate demand.
Second—you do not hear this story that much, but it is important to set up the third—is that we should not be using fiscal policy but should instead by using monetary policy. That is a more popular argument in the more informal discussion in the econoblogosphere than it is in academia. But there is the question of what you can do.
Third, even though we are at the zero lower bound, fiscal policy is a lot less effective than the man from 1958 would say it is, and that multipliers are quite low even under urgent conditions.
About limited economic slack:
There is a whole literature trying to identify structural issues—what does the shift in the Beveridge Curve mean—that would be an entirely different discussion. I think the most important argument that has the biggest impact is the argument: “If we have all that economic slack, where is the deflation?” When we look at core inflation, it dropped a lot in the crisis, but has been fluctuating in a 1-2%/year range since then and has not been declining. You will see the argument, which is consistent with what most Principles of Economics or Intermediate Macroeconomics textbooks say or would have said before the crisis, that if we really had a large output gap we should be seeing not just low but declining inflation. The stability of the core inflation rate is an indication that there is not a lot of economic slack. The most recent speech by James Bullard makes that case. The San Francisco Fed has a nice updated chart estimating the output gap by backing it out of a linear expectational Phillips Curve and comparing it to the CBO output gap which is a gussied-up trend. The difference is striking. The stability of inflation says that there is hardly any output gap. Comparing us to the pre-crisis trend says that there is still a very large output gap: $900 billion/year of potential non-inflationary production of goods and services is simply not happening.
Brad asked: “What have we changed our views about?” The inflation process is one area in which I have changed my views. It has become much more apparent that downward nominal rigidity—not just stickiness but people don’t like to cut nominal prices and wages—is a very significant factor. When you have a depressed economy in a state of initially low inflation the zero bound not just on interest rates but on wage changes becomes a really big deal. Again, more San Francisco Fed stuff: they have tried to back out how many people are literally getting zero wage change. The answer is: “a lot”. That suggests that we are indeed an economy in its depressed state, and that the reason that average wages continue to rise is that we have truncated the left edge of the distribution, not that we have anything close to full employment.
That is very important, if true. Among other things, it means that the whole basis on which we constructed monetary policy during the Great Moderation, which is that stabilizing inflation and stabilizing output are the same thing, is all wrong: you can have a sustained period of low but not negative inflation consistent with an economy operating far below its potential productive capacity. That is what I believe is happening now. If so, we are failing dismally in responding to this economic crisis. This is in contrast to what some central bankers are saying—that we have done well because inflation has stayed relatively stable.
Monetary policy: When I arrived at Princeton in 2000 there was a group of us—“Japan worriers”. I am the only one still there. Mike Woodford, Lars Svensson, who is now run off to the Riksbank, me, and Ben Bernanke—I wonder what happened to him? All of us were very concerned by what was happening to Japan in the 1990s. Some people looked at it and said: “That just shows how messed up the Japanese are.” Some of us looked at us and said: “Surface differences apart, Japan looks a lot like us: big advanced country, lots of room to maneuver, government officials who might not be the most brilliant but who were not complete idiots, and if they could get trapped in this sort of deflationary stagnation then it could happen to us.” Sure enough, it did.
At the time, all of the discussion was about what you could do by way of monetary policy. Could the central bank by unconventional purchases of non-standard assets move expectations? The simple fact is that dramatic changes in the simplest measures of what central banks are doing—the size of the monetary base—have been invisible in their effect on either inflation or output. I think we have to say that at this point to make the argument that if only the central bank really wanted to we would be doing much better needs to be accompanied by a very clear explanation of how that it is supposed to work and why the effects of monetary policy to date have been so limited. There is in principle the expectations channel. If a central bank can credibly promise that it will allow a higher inflation rate over the medium term then it ought to be able to reduce real interest rates and have a significant expansionary effect on the economy. The problem is how do you in fact make that promise credible. There are multiple hurdles that you have to cross. First, you have to cross the threshold of the political acceptability of the policy of changing the inflation target, which has proved virtually impossible to tackle in part because people do not think that this is a permanent crisis. They may be right. But that means that it is then very very hard to say that we should change the price-level target for five or ten years in the future to deal with a crisis that everybody expects will be over in a year .
Then, how do you make it credible? Why will the people running the central bank five or ten years from now—who are not the people running it now—go through with it? In an unfortunate phrase I used back in 1998 about Japan, they have to credibly promise to be irresponsible. That is the issue. It has turned out, I think, that, as Michael Woodford says, while in principle unorthodox monetary policy can deal with a situation like what we have now, in practice it is really really hard to see how this could work. And that makes you lean on fiscal policy.
Last comes the question about the effectiveness of fiscal policy. Valerie Ramey will present evidence on the size of multipliers. What are multipliers? That is a critical issue. The trouble is that fiscal policy is very hard to assess econometrically from the historical record. The basic rule is that when all is said and done, no matter how much effort we put it and in spite of all the valid work we do, unless you can show clear natural experiments people are not convinced. Even with natural experiments people are often not convinced, but it is your best chance. And convincing natural experiments are hard to come by. The clearly-exogenous changes in government spending are pretty much those associated with wars. This is just the very simple stuff that Bob Hall did just a little while back. They clearly show that expansionary policy is expansionary. They also show that the multiplier is less than one, which is not what an enthusiastic advocate of Keynesian fiscal stimulus would like to see. Again, the IMF tried recently very carefully to tease out the answer, and again found that expansionary policy is expansionary and contractionary policy is contractionary, but once again multipliers are less than one.
The IMF has changed its mind, or at least Oliver has changed his mind. But that’s where we are.
The question then becomes: is this historical evidence relevant for what we face now? The historical evidence incorporates a lot of crowding-out. The question is then: where is this crowding-out coming from? One answer is the old textbook crowding-out: crowding-out via rising interest rates. That is clearly relevant to the historical cases but not relevant now. A second answer is that in wartime other things are happening. I believe that a lot of the literature on this understates the seriousness of this issue. It’s not just that the multiplier is lower at full employment. During World War II there was severe rationing of consumer goods. During World War II—I have not seen this mentioned at all—there was essentially a prohibition on private construction. You look at World War II and say “private spending fell”. What relevance does that have? We are not about to have such controls on private investment. [45:00]
We can look at periods that do not have war complicating the picture, and the problem is that there is not a lot of that. For the U.S., the World War II period before wartime controls come in is about a year and a half, six quarters. If you are going to use VAR time-series methods, you can look at quarters that have both high unemployment and large changes or news of large changes in military spending, the problem is that the impulse response period extends well into the period of wartime controls. It is not at all easy to get past that.
Finally, Ricardian effects. It is really important to understand how many people misunderstand that. There are many people who believe that higher government spending now means higher taxes later and this will crowd-out private spending now. But higher spending now means higher incomes now as well. In the simplest Ricardian setup, if you believe that resources are unemployed and if interest rates are zero, the multiplier is not zero but one. It is very difficult to come up with a story in which the current multiplier would be less than one. Invoking the expectation of future tax increases as a reason for a multiplier less than one is a much more difficult story to tell than people seem to imagine.
Our evidence is not great. The closest thing to a really good natural experiment is what is happening now—the scary policies of recent years. It is not perfect. But look at the euro area countries—we talk about the great mistake of 1937, Roosevelt’s turn to austerity, but his turn to austerity was less than 3% of GDP. Compare that to what is happening to Greece or Ireland now, that is nothing. In Greece, if the whole program is implemented, we are talking about austerity on the order of 16% of GDP. These are enormous shocks. And if you do a simple regression it looks like a multiplier of 1.3.
The immediate objection is that causation is not reversed? This is where the Blanchard-Leigh stuff comes in: They look at forecast errors in output growth and forecast errors in future policy, and find that their forecasts of output growth which assumed a multiplier of 0.5 underestimated the true multiplier by about 1.0, systematically understating economic contraction in countries with larger-than-expected degrees of austerity.
I think their work is good. Of course, it fits what I wanted to believe, so you have to be careful. But very important stuff, if true.
The final point is policy: Are we sure that expansionary fiscal policy is the right thing to be doing and that austerity is a terrible, terrible mistake? No. We are absolutely sure of nothing. But the consequences, if that is the truth, and I think the evidence tilts that way, is that what we are doing right now is absolutely disastrous. And that is where we are right now.