Tinkerbell strikes back: Why Krugman is wrong

Paul Krugman recently responded to my expectations trap post, and most of his comments were either wrong or misleading.

What Scott is suggesting is that all macro policy, both monetary and fiscal, is subject to what we might call the Tinkerbell Principle: you can fly, but only if you believe you can fly.

This one is correct.  But just so there is no misunderstanding, Thoma was the one making the Tinkerbell argument for monetary policy, and in my expectations trap post I thought it exceedingly unlikely that it would be a problem in practice.  I said:

So please don’t take this post the wrong way.  When I say that the expectations trap is equally applicable to fiscal policy, I’m not saying that I think the Fed is likely to go out of its way to sabotage fiscal policy.  But it is even less likely to sabotage monetary policy, indeed the latter hypothesis seems much more far-fetched to me.

For those who don’t know, the Tinkerbell argument (which I believe Krugman developed for monetary policy in 1998) is based on the fact that current AD is heavily influenced by future expected AD (and hence future expected monetary policy.)  If current policy is expansionary and next year’s monetary policy is expected to be highly contractionary, then current AD won’t rise very much because the effect of future expected monetary policy on current AD is much stronger than the effects of current monetary.  And I would argue that the same applies to current fiscal policy.   (We don’t have to worry as much about future expected fiscal policy, because most new Keynesian models assume that you aren’t going to be in a liquidity trap forever, and hence monetary policymakers drive NGDP growth in the long run.)

Suppose the price level is 100, and you want to raise it to 110 in two years.  In principle, this could be done with either monetary or fiscal stimulus.  Of course fiscal stimulus is assumed to boost AD by raising velocity.  I have doubts about the efficacy of fiscal stimulus, but certainly concede that it might have some stimulative effect.  But not if the public expects the Fed to change course in 12 months, and start trying to drive the price level back to 100 in 2 years.  Unless the public expects prices to be at 110 in two years, current attempts to boost AD will be largely ineffective.

I don’t think it is difficult to raise inflation expectations.  The central bank merely needs to lay out an explicit price level or NGDP target, level targeting, and promise to catch up for shortfalls.  If they do that they will be believed.  Why wouldn’t they be believed?  If they aren’t believed, then they need to consider other plans, like currency depreciation, or NGDP futures targeting.  But let’s cross that bridge when we come to it.  There is no example of central banks that were determined to inflate but failed.  Even Krugman criticizes modern central banks for their refusal to set higher explicit inflation targets.  So I don’t worry about the Tinkerbell Principle, at least for monetary policy.  It’s Thoma (and Krugman?) who seem to take it seriously.

[Update:  I should have said “fiat money central banks” or “governments” who wanted to inflate.  There may have been central banks that lacked the authority to inflate due to gold standard or currency board considerations.  HT:  Alex.]

Here’s Krugman again:

But this isn’t even true about monetary policy, unless you’re at the zero lower bound. Under normal circumstances, an open-market operation will reduce short-term interest rates, regardless of what the market believes. It’s only when you’re up against the lower bound that increasing the current monetary base does nothing, so an open-market operation matters only if people believe it signals higher inflation later.

In a 1993 paper I argued that if the Fed doubled the money supply, and was expected to cut the money supply in half a year later (i.e. return it to its original level) then prices would show little or no short term gain, despite the predictions of the simple Quantity Theory of Money.  I argued that prices could not double in the short run, as that would imply they’d be expected to fall in half over the following year.  But that would imply an implausibly high real yield on cash (100%.)  Instead, people would mostly hoard the extra cash.  And this is even true if T-bill yields were 5% when the Fed made the bizarre monetary injection.  Krugman might reasonably argue that the policy would immediately drive short term rates to zero, and I’d agree.  But I don’t think interest rates are the right way of thinking about monetary policy.  What matters is the impact of policy on asset prices; stocks, commodities and real estate.   These are forward-looking markets.  So while a cut in the current fed funds rate might have a tiny effect, it is always true that monetary policy is much more about future expected changes in policy, than the current setting of the fed funds rate.  Even new Keynesians like Woodford and Eggertsson agree on this point.

Krugman also errs in the preceding quotation in assuming that a lower short term interest rate means monetary policy has eased.  There was a severe negative monetary shock in December 2007, when the Fed eased less than expected.  Stocks plunged on the 2:15 announcement.  Krugman’s model would predict that three month T-bill yields should have risen on the news.  Instead they fell, as investors suddenly expected a much weaker economy.  And investors were right, the economy began weakening almost immediately, and the Fed had to scramble with some aggressive make-up calls in January–125 basis points of cuts in 2 weeks.  As Milton Friedman pointed out, ultra-low interest rates usually mean money is very tight.  Indeed over time tight money can so weaken real economic activity that even short term real rates end up lower (as Robert King showed.)

Krugman continues:

And the Tinkerbell principle NEVER applies to fiscal policy. If the government goes out and hires a million people to dig ditches, the direct effect is that a million people have been put to work digging ditches. It doesn’t matter whether people believe it will work.

This is wrong.  If the public doesn’t believe inflation will be higher, then they don’t believe the AD curve will shift to the right.  Why might they believe the AD won’t shift to the right?  Let’s use Krugman’s own assumption that central bankers are scrooge-like reactionaries who have an inordinate fear of inflation.  As soon as Congress passes that stimulus, the Fed will begin plotting to gradually tighten money, so that inflation doesn’t rise.  But you can only prevent inflation from rising if you prevent AD from rising.  And that means you must tighten monetary policy enough so that any gains to AD from fiscal stimulus, are offset by decreases in private sector activity due to monetary tightening.  Again, I am not claiming this would happen, what I am claiming is that it is more likely that the future Fed would sabotage the current Congress, than that the future Fed would try to sabotage an explicit inflation promise made by the current Fed.  I’m not the Tinkerbell!  (Not that there is anything wrong with a grown man styling himself after Tinkerbell.)

[As an aside, it is odd that Krugman would make this argument without at least addressing the reasoning I used for why the Tinkerbell Principle applies to fiscal policy.  Is or is not the Fed the last mover?]

Krugman continues:

Beyond the theoretical confusion, Scott argues that

“Almost everyone agrees that Japan does not face an expectations trap. They can devalue the yen whenever they wish, as much as they wish.”

Someone should tell my current hosts. The Swiss National Bank has been trying hard to prevent an appreciation of the franc: since the start of the crisis Switzerland has added around $100 billion, or 20% of GDP, to its foreign exchange reserves. An equivalent intervention for the US would be $3.5 trillion. Yet the franc has still strengthened from .6 to .7 euro.

The point here is that currency intervention is actually just a form of quantitative easing “” conceptually no different from buying commercial paper or long-term bonds. And the same problem arises “” namely, that you have to engage in a huge expansion of the central bank’s balance sheet to gain traction. You can say that’s OK, and in fact I would; I’d like to see the Fed add several trillion to its assets. But central banks are leery of doing this, for various reasons, including the fact that they’re taking on risk.

So it’s just wrong to suggest that currency intervention offers an easy way out of the expectations trap; there’s nothing special or magical about that particular form of QE “” or rather, nothing special except that it’s a beggar-thy-neighbor policy.

There are all sorts of problems here.  First one quick observation.  If the Fed was concerned that a bloated monetary base was leading them to take on too much risk, they’d have an interest penalty on excess reserves (or at least a zero rate.)  Instead they actual pay banks a quarter point to encourage them to hold on to a trillion dollars in ERs.  Second, there are trillions in T-bills and T-notes that could be bought with modest risk.  But in any case if they set a higher explicit inflation target it would reduce the demand for base money.  I’m not arguing for using QE as the weapon—inflation or NGDP targeting is the weapon. QE merely accommodates whatever demand the public has for base money at that target.

But let’s focus on the main issue here.  A central bank can always depreciate it’s currency by offering to sell unlimited amounts of currency at whatever nominal rate they wish to peg.  That open-ended offer becomes the market price.  So far I am not disagreeing with Krugman, but some of his readers might have misinterpreted what he was saying here.  His real point is that to do this they might have to expand the monetary base by a large amount.  I have two reactions to this:

1.  If so, (and he is probably right about Switzerland) then the Swiss basically have to choose which problem is worse; mild inflation, or a big monetary base.  If the bloated base is mostly due to bank reserves, then the Swiss can put a negative interest rate on reserves, and that will lower the demand for base money.  They could also set a higher inflation target; in that case fewer people might want to hoard Swiss Francs.  It’s their choice.

2.  It is clear from the behavior of the BOJ (which tightened monetary policy in both 2000 and 2006) that they don’t want inflation.  They are behaving exactly like a central bank would behave if it didn’t wish to boost prices.  So naturally there is a bloated demand for Japanese base money, their GDP deflator has been falling at 1% a year for 16 years.  But that’s their choice; if they wanted a higher trend rate of inflation they wouldn’t have tightened monetary policy in 2000 and 2006.  If you have a low rate of trend inflation, you’ll have a lot of demand for your base money.  It’s their choice.

Here’s where Krugman is misleading.  You do not need to inject a lot of base money to accommodate a credible and significantly positive inflation target, because when inflation expectations are significantly positive, people generally don’t want to hold a lot of base money.  Krugman puts the horse before the cart here.  Bloated monetary bases aren’t evidence of good faith inflationary attempts that failed, they are evidence of central banks that refuse to set robust inflation objectives, level targeting.

Krugman’s right that in one respect currency depreciation and open market purchases are two sides of the same coin.  They are both expansionary monetary policies.  But there is an important difference.  Simple OMPs don’t signal future price level intentions very clearly.  On the other hand if the BOJ starts pegging the yen at 150 to the dollar, it would be a pretty clear signal that they wished to sharply raise their future price level.  The reason is complicated, and has to do with interest parity and purchasing power parity.  People like Lars Svensson and Bennett McCallum have worked out “foolproof” models of reflation involving currency depreciation.  The basic idea is that when rates are near zero, interest parity implies that any fall in the spot exchange rate is associated with a fall in the forward rate.  And a lower forward rate (ceteris paribus) implies a higher expected rate of inflation.

What most discouraged me about Krugman’s comment here is the remark about “beggar-thy-neighbor” polices.  I think (but am not certain) that Krugman’s comments above imply that even ordinary OMPs, if they are to work at all, have effects similar to currency depreciation.  But this means that if the Japanese tried to do a more expansionary monetary policy, even without any intervention in the forex markets, it would be expected to depreciate the yen if investors thought it credible.  If we want the Japanese to do this (and I was under the impression that Krugman did) then the last thing we want to do is raise the issue of beggar-thy-neighbor policies any time they yen falls.  Perhaps Krugman is merely saying that the BOJ should buy domestic assets, not foreign assets.  Fair enough.  But let’s be aware that the yen might fall sharply if anything effective is done on the domestic front.  Let’s not even talk about “beggar-thy-neighbor” polices.

BTW, Gauti Eggertsson’s 2008 AER piece discusses how FDR did exactly what I suggested.  In 1932 the Fed’s OMOs failed to produce significant reflation.  They were not credible because of fears that the gold standard constraint would limit the Fed’s ability to carry through with the stimulus.  Gold outflows surged.  In 1933 FDR tried depreciating the dollar to get prices back up to pre-Depression levels (aka level targeting.)  Just as expected, this did raise inflation sharply.  And with little change in the monetary base.  So much for Krugman’s theory that currency depreciation is not much different from open market purchases.  Indeed the WPI rose by over 20% in 12 months, and RGDP started to recover.  And all this occurred at the zero bound.  FDR understood that “all we have to fear is fear itself.”  If we believe we can fly (inflate), and act on those beliefs with explicit nominal targets, level targeting, then we can fly.

PS.  David Pearson raised this point in the comment section to my earlier post:

I think the “expectations trap” is something we agree on: the Fed can avoid it if it shows enough determination. Where we disagree is that you think the cost of doing so is small.

Here’s where I disagree with just about everyone else in the world.  What others see as huge barriers to a higher inflation target, I see as symptoms of the current disinflationary policy.  Change that policy and those barriers won’t look nearly so formidable.

HT Marcus.



11 Responses to “Tinkerbell strikes back: Why Krugman is wrong”

  1. Gravatar of The Ambrosini Critique » Blog Archive » Isn’t or Shouldn’t? The Ambrosini Critique » Blog Archive » Isn’t or Shouldn’t?
    4. June 2010 at 09:37

    […] argues that if Krugman’s claim is true that the Fed is too conservative, that they will do whatever […]

  2. Gravatar of q2 q2
    4. June 2010 at 10:23

    I believe both Thoma and Krugman did not read your entire post, or at least not carefully, as they seemed to miss the point that the Fed can counteract any attempt to raise AD through fiscal policy. As I read it, they think you’re arguing that in order for fiscal policy to work, the *legislator* needs to be credible, which is why they emphasize that all the legislator needs to do is spend the money; they don’t need to set expectations.

  3. Gravatar of Alex Alex
    4. June 2010 at 10:53

    I don’t have much to say on this debate, but isn’t this:

    “There is no example of central banks that were determined to inflate but failed.”

    contradicted by this:

    “In 1932 the Fed’s OMOs failed to produce significant reflation.”


  4. Gravatar of Alex Alex
    4. June 2010 at 11:04

    In fact, this whole paragraph doesn’t make sense to me:

    “In 1932 the Fed’s OMOs failed to produce significant reflation. They were not credible because of fears that the gold standard constraint would limit the Fed’s ability to carry through with the stimulus. Gold outflows surged. In 1933 FDR tried depreciating the dollar to get prices back up to pre-Depression levels (aka level targeting.) Just as expected, this did raise inflation sharply. And with little change in the monetary base. So much for Krugman’s theory that currency depreciation is not much different from open market purchases. Indeed the WPI rose by over 20% in 12 months, and RGDP started to recover. And all this occurred at the zero bound. FDR understood that “all we have to fear is fear itself.” If we believe we can fly (inflate), and act on those beliefs with explicit nominal targets, level targeting, then we can fly.”

    Surely the difference between 1932 and 1933 was that in 1933, the US came off the gold standard?

    And am I right in saying that while th monetary base didn’t increase from 1932 to 1933, it did from 1933, and surely that’s the more important measure?

  5. Gravatar of Richard A. Richard A.
    4. June 2010 at 11:12

    The fact that the Fed is paying any interests on reserves should be prove positive that they don’t want much of the expanded monetary base to be loaned out. The Fed wanted to buy up MBS on a large scale and didn’t want this to expand the money supply. Obviously, they are trying to prop up the value of MBS. If inflation expectations increase, nominal interests will increase and make MBS less valuable.

    Maybe the Fed is more concerned about the value of MBS then they are about the economy.

  6. Gravatar of Indy Indy
    4. June 2010 at 15:41

    The Fed, like all of us, wants two incompatible things:

    1. The ability to make a credible commitment – a promise that be believed and considered rock-solid reliable, trustworthy, as a risk-free basis on which to rely since it is guaranteed to occur.

    2. Discretion, room for maneuver, the power and ability to adapt and improvise and set an adjustable, ad-hoc policy according to what the latest data and latest thinking and latest set of priorities indicate. This also implies the ability to surprise and shock, and to be unpredictable, secretive, and mysterious.

    You can’t have both of these things, at least, not at the same time. You have to make a choice, and which choice you make and how you articulate that choice tell us a lot about any institution and the organization’s understanding of the reality of its mission.

    Number 1 is about “anchoring expectations”. The Fed has chosen option 2, and if it to adjust policy, it has to do so by other means.

  7. Gravatar of Doc Merlin Doc Merlin
    4. June 2010 at 17:21

    1. Actually Krugman has a bit of a point. Not something I say often.

    Q: Who is the largest owner of US treasury bonds and thus would benefit the most from price deflation in the sort term?
    A: Why, the Federal Reserve of course.

    Q: Won’t that hurt the federal reserve’s member banks (aka stockholder)?
    A: Yes, but only if they won’t be bailed out. Really the NY fed is what matters and NY banks have been largely made whole.

    Q: So, the fed paying interest on reserves is deflationary and in the fed’s interest?
    A: Yes, and it also it gives more money to the member banks and protects them against tight monetary policy.

    2. I don’t know why so many of your commenters have been giving so much flack to the Austrians and their view that loose monetary policy in the past necessarily causes tight future policy.
    Now Scott, I wish to play devil’s advocate
    To put it in terms austrians would never use, in a credit dominated system, excessively high AD caused by money expansion, leads to either unsustainable money expansion in the future (hyperinflation) , /or/ a crash in AD (an Austrian business cycle). Attempts to further prop up the system with monetary expansion (that aren’t accelerationistic) will lead to higher levels of indebtedness which will ultimately make a future collapse much worse. What about this do you disagree with?

  8. Gravatar of TheMoneyIllusion » Ambrosini on the fiscal expectations trap TheMoneyIllusion » Ambrosini on the fiscal expectations trap
    4. June 2010 at 19:09

    […] has finally addressed my fiscal expectations trap idea.  Here is Ambrosini: Sumner argues that if Krugman’s claim is true that the Fed is too conservative, that they will do whatever to […]

  9. Gravatar of scott sumner scott sumner
    5. June 2010 at 08:19

    Q2, Thanks, I did a new post.

    Alex, Thanks, I updated the post to indicate central banks under fiat regimes, not constrained by a commodity or currency peg.

    Alex#2, Exactly. FDR went off gold and started targeting the price level. And he got immediate results. We are off the gold standard, and can do the same.

    The monetary base did increase modestly under the 1932 open market purchases. I don’t think it increased significantly during the dollar depreciation of 1933, but don’t have the data in front of me. Prices rose in 1933 because the future expected money supply rose.

    Richard, I’d hate to think that.

    Indy, You may be right, but I think they made the wrong choice.

    Doc Merlin, But even Krugman believes the Fed should not be doing that, so I don’t think he and I disagree on that point. And if they don’t want inflation, they’ll also try to stop the fiscal authorities from generating inflation.

  10. Gravatar of marcus nunes marcus nunes
    5. June 2010 at 09:35

    Joseph Gagnon argues that MP should strive to decrease long term rates:

  11. Gravatar of scott sumner scott sumner
    6. June 2010 at 06:08

    Thanks Marcus.

Leave a Reply