More Reverse Causation

There is a common misconception that zero nominal interest rates make monetary policy ineffective.  In fact, the reverse is true.  Because Paul Krugman is so respected by left-of-center economists, I had assumed that his “expectations trap” was now the accepted explanation of policy ineffectiveness, and that it was generally understood that Keynes’ original liquidity trap argument was faulty.  No doubt most macroeconomists are aware of Krugman’s argument, but as this link shows, much of the debate still revolves around the earlier Keynesian model, as formalized by Hicks (1937.)  To be fair to Brad Delong, I am sure that he understands the distinction between liquidity traps and expectations traps.  And he is merely trying to refute crude monetarist models that also lack rational expectations.  So maybe his exercise is defensible.  I am not a monetarist, but I imagine they might favor having the central bank do unconventional QE in that situation, i.e. buy interest-bearing assets.  But it is clear from his comment section that many of his readers do actually think that Keynes’ liquidity trap still shows monetary expansion to be ineffective at the zero bound.

Let’s start by reviewing the intuition behind Krugman’s expectations trap.  Krugman noted that a permanent increase in the money supply should be highly effective, even if rates are zero.  Why?  His explanation is that liquidity traps don’t last forever, and thus a permanent increase in the money supply should raise the expected future price level.  This will raise the expected inflation rate, and thus reduce real interest rates when nominal rates are stuck at zero.  Krugman then developed a new theory of policy ineffectiveness.  (Actually not completely new, I published the basic idea in 1993.  But his 1998 paper was far better.)  The basic intuition is as follows.  Suppose nominal rates are zero and the Wicksellian real interest rate is slightly negative.  Then suppose the government temporarily increased the money supply from $100 billion to $200 billion, with the expectation that all the extra money would be pulled from circulation one year later.  The price level cannot rise in year one, because if it did then there would be expected deflation between year one and year two.  But expected deflation would push the real return from holding cash above zero, i.e. above the rate earned on alternative assets.  Thus people (and banks) would simply hoard all the extra money, and prices wouldn’t rise at all.  So far, so good.  But I also think Krugman’s explanation is a bit misleading, for several reasons.

First, it applies the logic of Ricardian equivilence, so it is much less in the spirit of Keynes than he implies.  Krugman clearly admires Keynes, and I think that clouds his judgment.  Obviously Ricardian equivilence, if true, undermines a key part of the Keynesian model–the expansionary impact of tax cuts.  Second, Krugman suggests that this model is a good explanation for why the BOJ was unable to create inflation.  But as I’ve mentioned ad nauseum, the BOJ clearly didn’t want to create inflation, or even stable prices, they wanted deflation.  So there is no actual example of an expectations trap in all of world history–at least if “trap” means a sincere central bank being unable to boost prices.

Most importantly, I don’t think his argument has much to do with “liquidity traps” as the term is commonly understood.  The problem might just as well occur in an environment where nominal interest rates were 5%, before the Fed tried to inflate.  Consider the previous thought experiment, but now start from a scenario where interest rates were initially 5%, not zero percent.  Continue to assume the Wicksellian equilibrium real rate is slightly negative.  Also assume that the inflation embedded in nominal rates is explained by previous money growth.  We will assume that velocity and real output had been stable, just for simplicity.  Before Keynesians jump all over me, velocity will fall in half when I double the money supply, so my simplifying initial conditions are not what explain my results.

Now let’s suddenly double the money supply from $100 billion to $200 billion, just as in the previous case.  Just as before, we will assume the monetary injection to be temporary; we will assume the money supply is expected to fall back to $100 billion a year later.  What will happen?  We know that the expected rate of deflation cannot exceed the equilibrium real interest rate, or else the public would hoard the entire monetary injection.  But since there is no permanent increase in the money supply, the future price level cannot rise.  In that case the current price level cannot rise.  Inflation comes to a screeching halt.  Without inflation nominal interest rates immediately fall to zero.  All the extra money is hoarded.  BTW, if we made the thought experiment more realistic, long run steady-state velocity would fall as we went from 5% interest rates to zero percent.  In that case the monetary injection would actually cause a sudden and discontinuous drop in the price level.  It would be even more ineffective than in the previous example that started with zero interest rates.

Of course even in the 5% interest rate example, rates do fall to zero after the temporary currency injection.  But let’s be very clear about a couple points.  Whether monetary policy is ineffective has nothing to do with whether interest rates are at zero before you attempt a monetary stimulus.  Zero interest rates are neither a necessary nor sufficient condition for policy ineffectiveness.  There is a word for factors that are neither necessary nor sufficient—irrelevant.  Instead, it is the temporary nature of the injection that matters, indeed that is all that matters.  Second, it should be very clear from the example that started with 5% interest rates that the causation does not run from zero rates to monetary policy ineffectiveness, rather the causation runs from inept monetary policy to zero rates.  This is one of the points I have been trying to make for a long time, without much success.  Almost everyone seems to assume that the very low rates in the U.S. today are a sign that Bernanke and the Fed have given it “the old college try.”  In fact, the zero rates are nothing more than a sign of the ineptness of recent monetary policy.  It’s not just a matter of the markets saying the current policy stance is too tight, they are saying the Fed needs an entirely different operating procedure.  As Krugman pointed out, you need a procedure that is capable of creating positive inflation expectations.

I think a lot of people on the left have a general idea that there are some weakness in the General Theory, but take comfort from the fact that Nobel-prize winning economist Paul Krugman has looked at the question and found that Keynes’ basic intuition is right.  Krugman does make that claim in his famous “It’s Baaack” paper.  But I am afraid the model he came up with does not even come close to validating Keynes’ intuition, nor does it provide a plausible explanation for any real world liquidity trap.  Both Keynes and Krugman seem to have reversed the direction of causation.

I plan another post soon on the situation in Europe, where many seemed to assume a liquidity trap when rates were still far above zero.  At first I thought they must be crazy.  A more charitable explanation is that they subconsciously understood the argument in this post, they understood that policy ineffectiveness has nothing to do with zero rates.

Postscript:  After I completed this I thought the “necessary/sufficient conditions” point was too categorical.  Obviously zero rates have something to do with policy ineffectiveness.  I still think that zero rates are best thought of as a symptom of tight money.  But if markets think the Fed has no idea how to operate in a zero rate environment, then even though in a technical sense monetary policy is just as effective at zero rates as at 5% interest rates, the Fed may have more difficulty in the zero rate environment, simply because it is harder for them to establish credibility.  The solution is easy—start implementing policies that are effective (like interest penalties on excess reserves and explicit nominal target paths), and the credibility will soon follow.



24 Responses to “More Reverse Causation”

  1. Gravatar of Brad DeLong Brad DeLong
    9. April 2009 at 14:11

    Say, rather (I think: this stuff is confusing) that if one is starting from IS-LM with the nominal interest rate on the vertical axis, then quantitative easing is an IS shock…

  2. Gravatar of ssumner ssumner
    9. April 2009 at 15:01

    Brad, Thanks for the comment. I presume you mean by “QE” a policy of buying something other than T-bills. You may be right. I have to confess that I don’t understand the IS-LM model very well, as I never us it in teaching. Since starting this blog I have found that everyone looks at money issues differently. Even people I have a lot in common with (like Bob Lucas) seem to have reached radically different policy conclusions from me. (He thinks monetary policy is doing fine, I think it is the cause of the recession’s intensification last fall.) I had an earlier post called “Economic Babel” about how we all speak different languages. While I’ve got your attention, let me throw out an alternative view of multipliers. To me the relevant ceteris paribus isn’t holding monetary policy fixed during a fiscal expansion, but assuming monetary policy changes according to the preferences of the central bank after the fiscal action. That means the fiscal multiplier is zero during normal times when the Fed targets inflation, but even at zero rates it is a big question mark. It depends how hard the Fed would try to do QE, and how effective it would have been, in the absence of fiscal expansion. That’s also why Christy Romer’s view of gold inflows from Europe is so iffy. We already had deflation between 1937-40, probably at the limit of what FDR would accept. Without the gold inflows pushing up the base, he would have pressured the Fed to lower the gold ratio, or he would have devalued again. Everything seems very questionable, except my single-minded obsession that the Fed can and should have done much more. And then there’s the ECB.

  3. Gravatar of JimP JimP
    9. April 2009 at 17:03


    Convince Brad. He can call Krugman – who knows Bernanke real well and certainly knows his phone number – and we can find out why Bernanke is aiming at deflation. We can just ask him why he doesn’t charge interest on reserves and set a price level target. And this endless and horrible nightmare would be over.

    What do you say Brad? Give Paul a call. And I am not kidding.

    It all really does have to do with the vice grip of incorrect ideas.

    I remain haunted by something Scott said in an early post on here – that the Great Depression was all really just an error in inflation targeting! A stunner.

    And we are getting the same error again.

  4. Gravatar of JimP JimP
    9. April 2009 at 17:05

    I mean – it might not work. But what exactly do we have to lose?

  5. Gravatar of Discussão macroeconômica « De Gustibus Non Est Disputandum Discussão macroeconômica « De Gustibus Non Est Disputandum
    10. April 2009 at 00:59

    […] Discussão macroeconômica Abril 10, 2009 Posted by claudio in Uncategorized. Tags: armadilha da liquidez, armadilha das expectativas, Keynes, macroeconomia, Paul Krugman trackback Sumners comenta Krugman em uma discussão bastante relevante para se entender a situação econômica atual. Atenção para esta frase: Krugman clearly admires Keynes, and I think that clouds his judgment.  […]

  6. Gravatar of Bill Woolsey Bill Woolsey
    10. April 2009 at 02:45

    I think you misread Krugman. See his paper as suggesting that even using a new classical methodology (which they are quite arrogant in claiming is the only legitmate approach,) he can show that monetary policy can be ineffective. You are assuming that this means Krugman “believes” that these models are valuable.

    Krugman has continued to argue that simple IS-LM and AD-AS models are more useful under current conditions.

    The LM curve is generated by a relationship between the nominal interest rate and real income. Generally, the expected inflation rate is subracted off the nominal interest rate, to get a relationship between real income and the real interest rate. The IS relationship is just the relationship between the real interest rate and the various elements of real expenditure all summed up. To put this together with LM, the assumption is that real output adjusts to real expenditure, (because it is real output and real income that is impacting money demand in the LM.) So, it is partial analysis. It is showing shifts in AD, which put with AS completes the picture.

    Anyway, suppose we stop at the intermediate step with LM. And leave it in nominal income and real income “space.” Then, rather than adjust the LM cureve, adjust the IS curve. Add the expected inflation rate to the real interst rate. So that you get a relationship between the nominal interest rate and real expenditure (and real income.)

    An increase in the expected inflation rate shifts the IS curve “up.”

    Add potential real income as a vertical line. That intersects with adjusted IS at a negative nominal interest rate. The LM curve is horizontal at very low interest rates (negative based on the storage costs of currency, I think, but forget that for a moment. Make it zero.)

    If expected inflation rises. The IS curve shifts up. The nominal interest rate consistent with a given real interest rate is higher. That makes IS cross LM at a point closer to potential income.

    So, DeLong’s point above, that there is an IS shock is correct, but only if we adjust IS so that it relates nominal interest rates to real income.

    My view remains that the goal is not to increase the inflation rate beyond the long run target. It is rather to expand the money supply so that it offsets any decrease in velocity resulting from an excess demand for bonds at a zero nominal interest rate. (Really, this is a sort of spillover in to zero interest money.)

    If the price level falls enough, then a committment to get it back up to target temporarly lowers real interst rates. Though, when it reaches target, and if the real natural interest rate remains more negative than the targeted inflaton rate, and you keep to target, you are back where you started.

    But that says that we must have a recession serious enough to get prices down. And, I think, you also have the problem pointed to by Krugman’s model, though not exactly the same. There is no deflation. It is rather than extra high inflation stops and we are back to any targeted inflation.

    Anyway, this entire approach is wrong today because “the” interest rate is not at zero. Some interest rates are near zero.

    It just so happens that it is the interest rates that the Fed traditionally focuses upon that are near zero.

    And so, the implication is that they must throw out the operating procedure of using open market operations with T-bills to target the Federal funds rate. They have to buy other kinds of assets and other nominal interest rates will fall.

    As you well know, this compression of risk premia and term structure with the low end fixed at zero is something that bothers me a bit, and I think the answer is for the short term low risk instruments to have negative nominal yields. Adn that zero interest currency is the stumbling block. Don’t base the financial system on it. But, also, recognize that meeting the demand for zero interest currency in this environment is going to create a zero bound and reauire the issuer to take both interest rate risk and credit risk.

    We can imagine a situation represented by the IS LM above “the” nominal natural interest rate is negative. But we aren’t there now.

    My view is that if we were really in that state, it is time to suspend currency redemptions and go with negative nominal interest rates. But, we aren’t there and so quantitative easing can work by lowering nominal and real interest rates in the ordinary fashion.

    One other point. I think you need to at least consider the possibility that the “Wicksellian” real interest rate is more negative that your preferred inflation target. It is true, of course, that expectations of depression should lower that and so, realistically, a credible monetary policy will raise it, but if it still remains more negative than the inflation target, there is a problem. If that were the case, then your theory that interest rates will rise with success isn’t right. The Wiksellian nominal natural interest rate rises, but if that is currently negative, then it just rises towards the acutal zero market nominal interest rate.

    Perhaps you can explain to me why real economic growth requires a positive market clearing interest rate. I guess I never understood that.

  7. Gravatar of Alex Alex
    10. April 2009 at 06:26


    The problem is that the Fed cannot commit not to withdraw the additional cash in the future because of the way that money is introduced into the economy, i.e. buying bonds with newly printed cash which means that the Fed can always sell back the bonds and withdraw the cash. It all would be much easier if we could perform the classical helicopter drop. But even if we can’t the government has a way to commit to inflation in the future. And that is exactly what Obama is doing with the stimulus plan. Have the government get into deficit spending issuing vasts amounts of debt in the process and then have the Fed buy those bonds. Since we know that the Treasury will not be able to increase real revenues to pay back the debt the only way out for the government is to permanently increase the money supply and inflate the debt away. Friedman was half right when he said ‘inflation is always and everywhere a monetary phenomenon’ but he missed the other half, i.e. why do governments print so much money and that is in the end a fiscal problem.


  8. Gravatar of Bill Woolsey Bill Woolsey
    10. April 2009 at 07:25

    In case anyone cares, there was an especially bad typo in my past post.

    Do IS-LM with the nominal interest rate on the vertical axis and real income on the horizontal axis.

    An increase in the expected inflation rate shifts the IS curve “up,” because it raises the nominal interest rate associated with a given real interest rate.

    Even if the LM curve is horizontal at a zero nominal interest rate over the relevant range and a rightward shift in the LM curve simply slides the horizontal porttion along the horizontal axis, the impact on expected inflation shifts up the IS curve, causing them to intersect at a higher level of real income.

    Come to think about it, if IS LM is done in the usual way, then the horizontal portion of the LM curve is not along the horizontal axis but above it by the expected deflation rate or below it by the expected inflation rate.

    And so, expansionary monetary policy creates higher inflation expectations, the LM curve shifts down. And it crosses the IS curve further to the right.

    I guess this is all obvious

  9. Gravatar of ssumner ssumner
    10. April 2009 at 15:13

    JimP, Thanks, I’m trying to win friends and influence people, but not doing a very good job.

    Bill, There is a lot there, so let me take things one step at a time.

    1. Krugman is certainly very clear that he favors his expectations trap over the Keynesian liquidity trap. He may not have much fondness for the new classical methodology that he used (nor do I) but the central assumption behind his work of the late 1990s was that the liquidity trap requires monetary injections be temporary. And it is pretty clear that he still believes that. (Maybe it is because I reached the same conclusion independently, but I don’t see how anyone reading his 1998 paper could disagree with that central insight, regardless of their views of the new classical methodology.) When commenters recently asked him why the Fed couldn’t create inflation like the Zimbabwe central bank, he did not use the IS-LM model (Thank God!) but instead argued that a Fed attempt to create hyperinflation would lack credibility, any currency injections would be expected to be withdrawn in the future. Imagine trying to refute that thought experiment using a IS-LM model and permanent monetary injections! Even Keynes said his liquidity trap didn’t apply to “flights from the currency”. Krugman has repeated this neo-Ricardian Equivalence argument many times in his blog. And this is in responding to ordinary people, not Ratex economists. Plus he often links to his 1999 article on Japan. In that article he strongly argued that the BOJ failure was one of credibility–that the monetary injection was expected to be temporary. I will grant you one point, he does argue that once you assume currency injections are temporary, and thus nominal rates are zero, the IS-LM model is a good “heuristic” device (his term.) So yes, at the zero rate, temporary currency injections don’t move real or nominal rates, and you get no monetarist excess cash balance mechanism. I agree. But it seems clear to me that he does believe the reason is fundamentally the temporary nature of the injections. He also spoke very kindly of Eggertsson’s 2008 AER piece arguing that FDR turned things around in 1933 by shifting expectations, making the monetary injections seem permanent by devaluing the dollar. If he doesn’t believe what he writes about Japanese policy, or FDR in the Depression, that would be very weird.

    To summarize, I may have given the wrong impression about Krugman’s attitude toward Keynes. He definitely still thinks Keynes had a great insight, and that the IS-LM shows an important aspect of that insight. But the temporary nature of currency injections is a necessary condition. On the other hand, most non-economists who use IS-LM-type arguments don’t even understand that distinction.

    This is why I don’t worry as much about the issues that you mention, such as currency hoarding or the need for a negative interest rate on cash. There has never been a central bank that tried to inflate and failed. So if they spell out an explicit nominal target path, the public will probably believe them, contrary to what Krugman asserts. It’s not that they didn’t believe the BOJ, the BOJ never promised inflation. You are right that real growth and negative real interest rates could coincide. But I find it very, very unlikely that you could have a negative Wicksellian nominal natural rate at the same time as a 5% NGDP growth expectation. So that’s not a big concern to me. And if I am wrong, that’s just another argument for “level targeting”, which means if you only get 3% one year, you go for 7% the next. But again, I think that is extremely unlikely in practice. All the zero rates in history have been either in low NGDP growth environments, or deep in depression when level targeting would have suggested a higher growth target.

    Regarding the IS-LM model itself, I am influenced by a 1994 article by Robert King, who showed that as soon as you added expectations, expansionary monetary policy can raise nominal (and even real!) interest rates within that model. This is something I see all the time in my empirical work. It was obvious to me that interest rates fell to zero last year because Fed policy was highly contractionary. If IS-LM theorists tell me that’s because the IS curve shifted, fine. I have no interest in a model if most of the people who use it in the real world get everything important wrong–that is if they think Fed easing and tightening is nothing more than lower or higher interest rates. That’s not true of everyone who uses it, but it sometimes seems that way. How many IS-LM types correctly saw Fed policy last year as highly contractionary?

    Alex, Before giving up, before saying the Fed can’t credibly inflate, don’t you think it might make sense to have the Fed try? Has the Fed issued a promise to target the CPI growth trajectory, to make sure that the CPI is at least 6% higher in three years? And has the Fed injected enough money to raise inflation expectations? No. So I am not willing to give up, until they actually try. And I haven’t even mentioned all the contractionary things they are doing. They removed hundreds of billions of base money from circulation in the first few months of the new year–not the best move if you want to convince the public that you won’t do so in the future. And they keep paying interest on reserves. Let’s at least try to inflate before we give up and go to fiscal stimulus.

  10. Gravatar of Alex Alex
    10. April 2009 at 15:45


    There are many ways to do what you call fiscal stimulus. One way is to spend (i.e. increase G) another way is to send everybody a check for X dollars. The second one is the closest thing to a helicopter drop I can think of.


  11. Gravatar of Jon Jon
    10. April 2009 at 16:40

    “This will raise the expected inflation rate, and thus reduce real interest rates when nominal rates are stuck at zero. ”

    And yet you need to contend with corporate bond rates being +10%.

    I wonder why a Bank with excess reserves does not make purchases of corporate bonds (perhaps even the really “good” stuff that’s only yielding around 6%). Sure, loan volume can drop because of deleveraging by households and businesses but bonds are freely traded. So why don’t banks bite?

    They must suspect that the CB is going to drain the money supply before the bonds come due. Especially if the reserves arise from short-term auction facilities–the banks are on the hook directly. Whereas if the Fed uses OMO to drain the liquidity, its the public that’s exposed.

    Have we ever seen a liquidity trap?

    What were Japanese corporate bond yields?

  12. Gravatar of Bill Woolsey Bill Woolsey
    11. April 2009 at 03:02

    Much macroeconomic modeling and reasoning is of the single interest rate variety. I think it is a useful simplification. Generally, in macreconomics, we are thinking about market forces that impact all interest rates in the same way. If one is thinking about that, there isn’t really any harm in thinking about the federal funds rate being “the” interest rate.

    However, when we begin to think about “the” interest rate being zero, and when in reality, it is only some interest rates are zero, then single interest reasoning is not applicable.

    Implementing monetary policy by adjusting a target for the Federal Funds rate won’t work. Perhaps T-bills (with their near zero interest rates) are not appropriate for open market operations.

    But that isn’t the same thing as monetary policy is ineffective because “the” interest rate is zero. “The” interest rate is not zero. It is just that the interest rates most directly related to conventional Federal Reserve practices are near zero. It is those practices that must change.

    Anyway, corporate bond rates aren’t 10%. AAA were about 5.5% a year ago, last fall they went up to 6.5% and now are back to about 5.5%. BAA were about 7.5% a year ago. They went up to 9.5% (I guess that rounds up to 10%) and they are now at 8.5%.

    So, AAA corporate bonds are back to normal. BAA are still higher than they were a year ago. It is apparent that risk premia are higher.

    In my view, some of what happened last fall should be understood in terms of factors that impact all interest rates, however, there was a major change in risk premia. Different interest rates were impacted in different ways.

    In particular, the move into T-bills, FDIC insured deposits, and deposits (by banks) in accounts at the Fed, all extremely safe assets, made a monetary policy based on creating shifts between reserve balances at the fed and T-bills ineffective.

    That is why quantitative easing became necesary. However, the Fed uses the term quantitative easing to mean targetting credit into specific credit markets. This is as opposed to efforts to create an excess supply of base money at less than target levels of nominal income.

  13. Gravatar of Bill Woolsey Bill Woolsey
    11. April 2009 at 04:13


    You have worked very hard to refute Krugman’s paper. It has had little impact. Why? Perhaps because it isn’t as central to his thinking as you assume. Or, perhaps, that part of his thinking isn’t all that important to his “followers.”

    It is obvious that all interest rates aren’t zero. It is just that the interest rates on T-bills and the Federal Funds rate are near zero. Even if you believe that monetary policy can only impact nominal income by lowering nominal interest rates, then there is plenty of room for the Fed to institute that. Buy assets that still have greater than zero nominal yields. Don’t sterilize buy selling something else.

    If you want to increase the role of government in society, then “solving” this crises by massive government spending and borrowing is the right solution. Because this should raise the natural interest rate, then ordinary monetary policy can work. More explicitly, sell enough T-bills, and their equilibrium rate will rise. Open market operations to manipulate the Federal Funds rate will work again. If you prefer, say that you would be creating money to finance the additional govenrment spending. In the long run, taxes will need to be raised to support the larger goverment. Presumably, the problems with banking, housing, and the like will be worked out. And, we can live like the Swedes. Make America like Sweden.

    However, there are guys like Barro and Lucas who claim that this is uncessary. The “market” can solve this crisis. Or monetary policy can solve this crisis. But they (and their followers at top universities) use new-classical models.

    Well, Krugman has a response for them. Standard operating procedure for Keynesians has been to throw up special cases where “orthodox” policy would break down. Accept that it is just a bunch of debating points. Make America like Sweden, and the value of any argument is whether or not it builds political support for that goal.

    Well, keep on trying. Good luck.

    However, to me, your efforts to refute Krugman are counterproductive. It is a single interest rate model with that nominal interest rate at zero and monetay policy reducing that real interest rate through inflation. And so you are constantly talking about how the Fed can cause inflation. Even worse, you are doing thought experiments where the Fed causes massive inflation. (We double the quantity of money, double the price level.)

    There are two issues, one relevant to long run policy, and the other to current conditions.

    The long run policy issue is can changes in the quantity of money keep nominal income on target despite decreases in velocity? Suppose the decrease in velocity is due to a substitution of government bonds for FDIC insured deposits or balances at the Federal Reserve because the yields on T-bills have reached the zero lower bound? (I think the answer is yes, but lowering the target for the Federal Funds rate and buying T-bills to get the effective rate there will probably not work.) Krugman’s paper tells us nothing about this. If monetary policy can keep nominal income on target, the inflation rate stays at the long term trend the whole time. The lower bound on the real interest rate is the trend inflation rate. (Which I think should be zero.)

    The immediate problem at hand is can the Fed use monetary policy to get nominal income back up to its previous growth path. It seems likely that moving to that growth path will be associated with higher than trend inflation. However, once we get back to that growth path, then inflation will be back to the long run trend.

    Now, I guess this immediate problem is related to Krugman’s paper. Of course, it isn’t that the price level goes up and then goes down. It is that the real interest rate turns negative and then positive. It is that the price level rises more quickly and then more slowly. But the relevance is that the real interest rate (with the assumed zero nominal interest rate) will become very negative, but once we get back to our responsible target, it will be less negative.

    While I think that moving to a 3% nominal GDP growth should wait for the banking system to be reorganized, and so I am on board with the 5% growth in nominal income and 2% inflation for now, I think an analysis of how this works when the real interest rate is negative until we get back to the growth path and then zero from there on would be helpful.

    In general, how can monetary policy get us back to target when it has fallen below when the target growth path for nominal income implies a zero inflation rate and so a zero real interest rate if the nominal interest rate is at the zero nominal bound? If we are below target, then moving back up to target temporarily results in negative real interest rates, but when we are back to target, the real interest rate is again zero.

    To me, focusing up the issue of whether or not a central bank can engineer a permanent increase in the price level is a distraction.

  14. Gravatar of anon1 anon1
    11. April 2009 at 04:23

    “That is why quantitative easing became necesary. However, the Fed uses the term quantitative easing to mean targetting credit into specific credit markets. This is as opposed to efforts to create an excess supply of base money at less than target levels of nominal income.”


    that’s exactly what they’re doing; although they don’t use the term QE at all

  15. Gravatar of TDemeter TDemeter
    11. April 2009 at 10:51

    One stupid question:
    How about paying back(the best parts of the us) income tax?
    8,5% 0f US GDP(of 14.3 trillion) for a whole year!
    Wouldn´t that come close to perfect a helcopter drop? My point is that neither consumer nor banks expect things to improve in the short run.

  16. Gravatar of ssumner ssumner
    11. April 2009 at 18:08

    Alex, I agree that a helicopter drop would probably work. But the same monetary injection could work just as well without the bad side effects (either ballooning the deficit or creating hyperinflation) if they simply bought bonds. It’s not like the Fed and ECB don’t know how to debase their currencies, they just don’t want to do it. If they wanted inflation they’d set a target and start inflating. I don’t know anything they don’t. If I know about ideas like negative rates on excess reserves, you can be sure they do as well.

    Jon, I don’t see why you seem to think those examples contradict my argument. Obviously if banks are holding T-bills instead of corporate bonds yielding 6% or 10%, its because they expect the rate of return on T-bills (adjusted for risk) will be higher until the cash is pulled out of circulation. My example was not intended to reduce rates on long term bonds, but to show that if T-bill yields stayed near zero, the real rate would have to fall if inflation expectations rose. I would hope long term rates on T-bonds will rise, as if investors began to expect 5% NGDP growth, long risk-free rates should be higher. Regarding corporate rates, it’s harder to say. The risk of default will drop with recovery, but the real, risk-free rate and inflation expectations will rise.

    Bill, I am not trying to refute Krugman’s paper, I am trying to get his supporters to understand his argument, which I believe is the only theoretically sound argument for monetary policy ineffectiveness. Krugman showed (conclusively in my view) that one could only get monetary policy ineffectiveness if the currency injection was believed to be temporary. I’m not trying to refute that view, I am trying to get everyone else to accept it. Because if and when I do, then it will be easy (in my view) to make the case for monetary stimulus. Once people understand that zero rates do not in any way limit the effectiveness of monetary policy, then much of the irrational fear of monetary impotence will vanish. Getting people to perceive monetary injections as permanent seems like a much more manageable problem that overcoming this imaginary “pushing on a string” bogeyman. Maybe I shouldn’t have used 100% increases to make my point, it would be equally true for 2% increases in the money supply, just harder to see because then you need to account for changes in factors like real GDP, etc. Also remember that my example is not showing you can get lots of inflation from increasing the money supply 100%, rather I am showing that you don’t get any inflation at all if it is temporary. So the 100% increase drives home the point of how important the temporary vs. permanent distinction is. In fact, even if interest rates are 5%, not 0%, and you have a temporary 100% increase in the money supply, you don’t get any inflation. On the other hand if rates are zero percent, and you have a permanent 2% increase in the money supply, you then have a permanent 2% increase in the price level.
    I found over on Brad Delong’s blog his commenters were unaware of the fact that this was Krugman’s argument. I don’t blame them, because he often “forgets” to mention the crucial significance of the temporary vs. permanent distinction. I get criticized for trying to read his mind, so I’ll let others draw their own conclusions about why he favors massive fiscal stimulus, and especially whether it is in any way influenced by his well-publicized advocacy of a Northern European-style welfare state. All I know is that when someone knowledgeable presses him on the point, he goes right back to the temporary vs. permanent distinction, as I assume he knows that any sound argument for a liquidity trap must rely on that assumption.

    I agree with you that it is useful to often simplify things with a single interest rate. But when rates get to zero I don’t like to say that “now we have to look at other rates”, rather I would say that interest rates are a meaningless indicator of monetary policy at 100% nominal rates, 20% nominal rates, 10% nominal rates, 5% nominal rates, and 0% nominal rates. They are always meaningless, not just at zero rates. All that matters is expected growth in the nominal aggregates, and any variable that is a useful indicator of that expected growth. Since 0% rates are generally a sign of very low expected NGDP growth, if anything they are an indicator of very, very tight money. And in my view that’s all there is to be said about zero rates. They certainly don’t represent a good faith effort by the central bank to inflate, which is the impression you get from the press, and even many economics blogs.

    Of course I agree with most of your points. What makes the liquidity trap so diabolical is that it must be addressed on a number of fronts at once. My advocacy if a penalty rate on reserves isn’t so much due to the fact that I think ordinary QE won’t work, but that it will reduce the fear of hyperinflation, the fear of overshooting, because with a penalty rate on excess reserves you wouldn’t need much more base money, perhaps even less than we have today. The same thing applies to the explicit nominal target path. You can get stimulus without it, but it makes the Fed’s job far, far easier. This is a problem that must be attacked on a number of fronts at once. And when I just focus on one issue (like the temporary vs. permanent distinction) it seems like I am ignoring other important issues. But I agree that an effective policy must attack them all at once. I would even make this recommendation. It would be very helpful if Bernanke gave a speech explaining that the BOJ never tried to inflate, and that he was bound and determined to inflate. I’ll bet that would make the euro soar even more than his QE decision. What are the chances of such a speech? Near zero.

    anon1, Yes, I agree that that’s what they are doing. And the results are visible all around us. 600,000 jobs lost every month.

    TDememter, Yes it would probably work, if by “work” you mean creating hyperinflation. If you pull back the money later to prevent hyperinflation, then the national debt soars. Ordinary OMOs are better.

  17. Gravatar of Kevin Donoghue Kevin Donoghue
    11. April 2009 at 23:19

    “I found over on Brad Delong’s blog his commenters were unaware of the fact that this was Krugman’s argument.”

    I’m one of them so presumably you established to your own satisfaction that I was unfamiliar with Krugman’s views. So be it. Some of my teachers gave me crappy marks too. I will readily acknowledge that I don’t know as much about his thinking as I would like to. For example, I don’t know why he says the Fed can’t “credibly commit to inflation at rates higher than the 2-ish percent target it’s already believed to have”, because this is not a realistic option right now. Presumably he has in mind some political constraints, since it is hardly controversial that a sovereign power can debase the currency all too easily.

    “It would be very helpful if Bernanke gave a speech explaining that the BOJ never tried to inflate, and that he was bound and determined to inflate. I’ll bet that would make the euro soar even more than his QE decision.”

    Perhaps it’s because I live in the Eurozone, but I don’t think that word “helpful” means quite what you think. To me that sounds like a return to the bad old beggar-thy-neighbour policies of the 1930s. And when you say that the chances of such a speech are near zero, aren’t you in Krugman’s camp, tacitly acknowledging that the Fed is debarred from taking such drastic steps?

  18. Gravatar of anon1 anon1
    12. April 2009 at 02:47

    What level of interest rate charge (i.e. negative interest rate) do you think would be appropriate right now on excess reserves?

    And if the Fed charges interest on reserves won’t that drive the fed funds rate negative as well? (Banks will sell excess reserves to each other at a negative rate up to the negative rate on reserves.)

    Similarly, won’t the entire short term rate structure move in sympathy?

    And won’t the effect ripple through the yield curve as well?

    So won’t pension funds with long term fixed rate liabilities be in even more trouble from a valuation perspective?

  19. Gravatar of ssumner ssumner
    12. April 2009 at 05:23

    Kevin, Krugman may well be right that the Fed cannot commit to greater than 2% inflation for political reasons. But my response would be “so what.” If they could credibly commit to 2% inflation, monetary policy would be far more effective than fiscal policy. We could drop the wasteful fiscal stimulus and just go with 2% inflation. Isn’t that what both Krugman and I want? So I don’t see how that would represent monetary policy failure? The current expected rate is about 0.8% over the next 5 years. Two percent expected inflation would require much faster AD growth than we are likely to get under current policy.
    My most recent post (Mass Suicide) is on the “bad old beggar-thy-neighbor policies of the 1930s.” The leading expert on these policies, Barry Eichengreen just called for a return to those polices. And for good reason. The Great Depression was caused by countries refusing to devalue, and the recovery in each country only started when they began to devalue. Ideally all countries would devalue at once–of course they would fail to devalue against each other, but would succeed against goods and services. I think the ECB would be better off with a more expansionary policy. But if they are content with the current level of AD, then they can decide to stand pat.
    Finally, I think you misunderstood my speech example. Bernanke could certainly give such a speech. The problem is that he has given no indication that he is so inclined.

    Let me summarize my view on credibility. Central banks have no problem make policies credible if they are THE POLICIES THE CENTRAL BANK ACTUALLY WANTS. (That’s my dispute with Krugman) So if the Fed sincerely wants 2% inflation, they should have no trouble making such a policy credible. If they don’t want 5% (and I don’t think they do) then I agree with Krugman that they would have trouble making that policy credible. My only dispute with Krugman is that he seems to assume that the BOJ really wanted to escape deflation, but was held back by a lack of credibility. But there is no evidence for this and lots of evidence against. They are still actively pursuing deflationary policies 15 years later. What does that tell you about their goals?
    BTW, my comment wasn’t specifically aimed at you, as you are clearly more knowledgeable than the average commenter. And I just picked Delong’s blog at random. One sees the same thing everywhere.

    anon, I have no idea exactly what the penalty rate should be, but I think 1 or 2% would be plenty. T-bill rates probably wouldn’t fall much below zero, as people have the alternative of cash. But if they did, isn’t that what the Keynesians say we need? Don’t they say lower rates are needed but we are held back by a liquidity trap?
    I think such a policy would help pension funds, because it would promote rapid recovery and rates would quickly rise to their normal level. It would also sharply boost equity prices, which should also help pensions. My idea is not a “low interest rate policy” its a “let’s generate 5% NGDP growth policy.”

  20. Gravatar of anon1 anon1
    12. April 2009 at 05:42

    Have you considered the possibility that banks can easily exchange excess reserves for currency and hoard currency inventories if they are penalized on excess reserves? Or are you saying they would also be penalized on currency holdings if they did this?

  21. Gravatar of david glasner david glasner
    12. April 2009 at 10:36

    Scott, This comment may be more in the way of a theoretical embellishment than a substantive disagreement, but I do think that there is something misleading in the argument that an increase in the money supply has to be permanent to be effective. What is misleading is that the argument leaves the demand for money entirely unspecified. So, if the current deflation is the result of an increased demand for money, what we want is a permanent increase in the supply of money relative to the demand for money, not a permanent increase in the money supply in absolute terms. So if the precautionary or speculative demand for money that triggered the deflation goes away in the future, then we would want a corresponding reduction in supply of money in the future. Otherwise, future inflation will overshoot its trend. I assume that the proviso is implicit in your discussion, but I also think that it is worth making the point explicit.

    I would also be interested in the basis for your assertions that the Bank of Japan has been deliberately aiming at deflation? I don’t disagree, I am just interested in why you feel that it is clear what their objectives have been.

    Finally, I agree with anon1 that imposing a penalty rate on (legally) excess reserves could cause banks to cash in reserves for currency which would carry significant deflationary implications that would undercut or even sabotage the inflationary objective that we both would like to achieve.

  22. Gravatar of ssumner ssumner
    13. April 2009 at 11:06

    anon1, I discussed that in an earlier post. They could apply the penalty to vault cash as well; it is a part of bank reserves. Or they could let banks hold a modest amount of cash with no penalty, to simplify the information processing necessary. Either approach would work.

    David, Regarding your final point, see my response to anon1. BTW, Robert Hall endorsed my idea today–he is the world’s leading expert on bank interest on reserves.

    You are completely correct about your main point, it is the future path of the money supply relative to demand that matters. I do mention that qualification sometimes, but in simple thought experiments I often leave it ought. I should remember to always add that point in the future.

    The Japanese case is pretty clear to me. Deflation has averaged about 1% a year for quite a long time (I use the GDP deflator, which is more meaningful than the import price-distorted CPI). How did the BOJ react to this ongoing deflation? They tightened money in 2000, by raising their target rate. A mistake some might say? They did it again in 2006. Fool me once . . . They seem to be using discretionary policy to preservea bout 1% defaltion. If they aren’t trying to do that, they are pretty inept. They also let the yen gradually appreciate over time, whereas if they truly wanted inflation they would have depreciated the yen. Svensson calls currency depreciation a “foolproof” solution to the liquidity trap. Of course it is possible that they didn’t actually want deflation, but wanted some other goal (i.e. a strong yen) that implied deflation, but that wouldn’t change my argument. At best they didn’t care about deflation or inflation. It would be like the gold standard in 1932. The relatively stable rate of deflation in their GDP deflator in recent years is striking. I think it finally reached zero in 2008, only to turn negative again. They still have a higher policy rate today than in the late 1990s. (unless they cut it recently.) Why?

  23. Gravatar of abhay abhay
    14. April 2009 at 11:36

    Just wondering how come the results of these models do not depend on “how the extra liquidity is used”.
    I could think of many ways which should lead to different outcomes.
    1) Firms invest these. Increased capacity. Employment Gains. Increase in Real Potential Output.
    2) Money keeps transferring. e.g. Banks to Consumers (who repay their debts) and back to Banks. It might increase the value of output (especially the Financial sector). But most of the output rise would be due to increased (asset) Prices rather then increased quantity.

    Will introducing Trade change the results of the model? Given that most of consumer products are made in China which in turn buys US assets.

  24. Gravatar of ssumner ssumner
    15. April 2009 at 06:38

    abhay, monetary policy works by changing the value of money. The method of injection is not very important, at least relative to the effects on the price level and nominal GDP. The monetary base is very small relative to NGDP, and even smaller compared to total financial assets. So whether the Fed buys T-bonds, foreign bonds, or corporate bonds is much less important than how much they buy. If more money causes 20% inflation, that will occur regardless of what they buy. Then the 20% inflation will have massive effects on the economy, that go far beyond the effects the Fed has had in the asset market it operated in.

    Trade doesn’t change my results at all. And most consumer products are not made in China. We import a few hundred billion worth from China, and not all are consumer products. Consumption is nearly 10 trillion.

Leave a Reply