“A serious mistake?” Yeah, I’d say so.

Here is a long passage from pp. 33-36 of a November 2009 paper by Woodford and Curdia, which describes a 2003 paper by Woodford and Eggertsson (you’ll need to open the PDF):

Eggertsson and Woodford show that it can be a serious mistake for a central bank to be expected to return immediately to the pursuit of its normal policy target as soon as the zero bound no longer prevents it from hitting that target. For example, Figure 11 (reproduced from their paper) compares the dynamic paths of the policy rate, the inflation rate, and aggregate output under two alternative monetary policies, in the case of a real disturbance
(here interpreted as an exogenous increase in the probability that loans are bad, requiring intermediaries to increase the credit spread by several percentage points) that begins in period zero and lasts for 15 quarters, before real fundamentals permanently return to their original (“normal”) state.

Note:  I wasn’t able to copy the figure 11.  It is on page 61, and is worth looking at.  The dotted line shows a deep and prolonged recession with a policy of inflation rate targeting.  The solid line shows the economy avoiding a recession (and avoiding deflation) with a policy of targeting the price level.  Note that they are proposing an elastic price target, so it is actually quite close to NGDP targeting.  Of course the other difference is that they do not contemplate targeting the forecast, which I think would make it even more likely that a recession could have been avoided.

C&W continue . . .

In the case considered, if the financial disturbance were never to occur, optimal policy would involve maintaining a zero inflation rate, as this would also imply a zero output gap in every period. After the disturbance dissipates, one of the feasible policies is an immediate return to this zero-inflation steady state (under the parameterization assumed in the figure, this involves a nominal interest rate of 4 percent), and this is optimal from the point of view of welfare in all of the periods after the financial disturbance dissipates.  It is not, however, possible to maintain the zero-inflation steady state at all times, because during the financial disturbance this would require the policy rate to equal minus 2 percent, which would violate the zero lower bound.

Note:  They consider a zero inflation target for simplicity.  If you used a more realistic 2% inflation target, then the “bad” policy would be simply returning to a 2% inflation target when the financial crisis/liquidity trap was over.

C&W continue . . .

One of the policies considered in the figure (the dashed lines) is strict (forward-looking) inflation targeting: the central bank uses interest-rate policy to maintain a zero inflation rate whenever it is not prevented by the zero lower bound on the policy rate; and when undershooting the inflation target cannot be avoided, the policy rate is maintained at the lower bound. The other policy (the solid lines) is the optimal Ramsey policy, when the zero lower bound is included among the constraints on the set of possible equilibria.  The forward-looking inflation targeting policy is clearly much worse, as it involves both a much more severe output contraction and much more severe deflation during the period when the zero bound constrains policy. The problem with the forward-looking inflation targeting policy is that because the central bank simply targets zero inflation from the time that it again becomes possible to do so, all of the deflation that occurs while the zero bound binds is fully accommodated by the subsequent policy: the central bank continues to maintain the price level at whatever level it has fallen to. This results in expected deflation during the entire period of the financial disturbance, for deflation will continue as long as the financial disruption continues, while no inflation will be allowed even if the disturbance dissipates; this expected deflation makes the zero bound on nominal interest rates a higher lower bound on the real policy rate, making the contraction and deflation worse, giving people reason to expect more deflation as long as the disruption continues, and so on in a vicious circle.

Note:  Don’t be thrown off by the term “forward-looking.”  In this case it simply means inflation targeting going forward.  The “good” policy can also be forward-looking as long as you target a price level trend line that is rising at 2% a year from the beginning of the liquidity trap.  In our case, that would be October 2008.

C&W continue:

The outcome would be even worse in the case that the central bank were to seek to achieve the target criterion (3.1) each period, as soon as it becomes possible to do so. This is because, once credit spreads contract again, this policy would require the central bank to target negative inflation and/or a negative output gap (even though zero inflation and a zero output gap would now be achievable), simply because there had been a large negative output gap in the recent past (when the zero bound was a binding constraint); but the expectation of such policy would make the contraction while the zero bound constrained policy even more severe (justifying even tighter policy immediately following the “exit” from the “liquidity trap”, and so on).

Under the optimal policy, there is instead a commitment to maintain accommodative conditions for a brief interval, even though the reduction in credit spreads means that this level for the policy rate is now expansionary, leading to a mild boom and temporary inflation above the long-run target level (of zero). The expectation that this will occur during the “exit” from the trap results in much less contraction of economic activity and much less deflation, because it makes the perceived real rate of interest lower at all times while the policy rate is at zero (given that there is in each period some probability that credit spreads will shrink again in the next period, allowing mild inflation to occur). This expectation results in less deflation and higher real activity while the lower bound binds; and the expectation that continuation of the financial stress will have less drastic consequences is itself a substantial factor in making those consequences much less drastic — in a “virtuous circle” that exactly reverses the logic of our analysis above.

While our analysis implies that it is desirable for people to be able to expect that the “exit” from the trap will involve mild inflation, it does not follow that the possibility of occasionally hitting the zero lower bound on the policy rate is a reason to aim for a substantial positive rate of inflation at all times (as proposed by Summers, 1991), simply in order to ensure that a zero nominal interest rate will always mean a sufficiently negative value of the real policy rate. To the extent that a history-dependent inflation target of the kind called for Eggertsson and Woodford can be made credible and understood by the public, it suffices that the central bank be committed to bring about a temporarily higher rate of inflation only on the particular occasions when the zero lower bound has bound in the recent past, and not all of the time.

Note:  That is a big “if.”  I have been arguing that Woolsey’s zero inflation goal (implicit in his 3% NGDP goal) is better than my 2% inflation goal (implicit in my 5% NGDP target) if we have an optimal monetary regime.  If we don’t, then we need the higher expected inflation rate.  Woodford and Eggertsson make the same point.   I would add that this it is not a matter of convincing the public (which is what their last sentence implies), it is a matter of convincing the FOMC.  The public is far more rational than the FOMC.  They have read policy correctly, and foreseen the consequences.  The problem is that we have had a deflationary policy, and the Fed clearly does not intend to follow Woodford and Eggertsson’s advice to make up for any price level undershooting.

C&W continue . . .

This analysis implies that a commitment to maintain policy accommodation can play an important role in mitigating the effects of the zero lower bound on interest rates. One might reasonably ask for what length of time it is sensible to commit to keep rates low, and in particular whether it is really prudent to make any lengthy commitment when it is hard for a central bank to be certain that recovery may not come much sooner than anticipated.  The answer is that the best way to formulate such a commitment is not in terms of a period of time that can be identified with certainty in advance, but rather in terms of targets that must be met in order for the removal of policy accommodation to be appropriate.

Note: I had forgotten that they made this point.  So their views are even closer to mine than I recall.  Do not promise low interest rates as far as the eye can see, promise to keep them low until you hit your target.  And of course in my view policy should be forward-looking, so keep rates low until the expected target is hit.

C&W continue . . .

In fact, Eggertsson and Woodford (2003) show that in the representative household model (and hence similarly in the special case described above), optimal policy can be precisely described, regardless of the nature of the exogenous disturbances, by a target criterion involving only the path of the output-gap-adjusted price level defined in (3.2). Under the optimal rule, the central bank has a target each period for Ëœpt, that depends only on the economy’s history through period t − 1, and must use interest-rate policy to achieve the target, if this is possible without violation of the zero lower bound; if the target is undershot even with a zero policy rate, the policy rate is at any rate reduced to zero “” and the target for Ëœpt+1 is increased in proportion to the degree of undershooting. In periods when the zero bound does not bind, the target for the gap-adjusted price level is not adjusted, and the target criterion is the same as the one discussed in section 3.1.

Actually, the adjustments of the target are not of great importance, even when the zero bound does bind: Eggertsson and Woodford show that almost all of the improvement in stabilization achievable under the optimal policy commitment can be obtained by simply committing to a target criterion of the form (3.2) with a constant target p∗. The crucial feature of the optimal policy is that the target for ˜pt must not be allowed to fall as a result of having undershot the target in past periods. Hence one of the approximate characterizations of optimal policy proposed in section 3.1 continues to provide a good approximation to optimal policy even when the zero lower bound sometimes binds: it is simply important that the commitment be to the level form of the target criterion (3.2) rather than to the growth rate form (3.1).

Translation: Level targeting is essential in a liquidity trap.  But even if you are not in a liquidity trap, it is almost as good as rate targeting.  So if you are going to have a policy that is transparent, and credible, why not do level targeting all the time?

The final paragraph of the paper sums up the key findings:

The main respect in which the appropriate target criterion for interest rate  policy should be modified to take account of the possibility of financial disruptions is by aiming at a target path for the price level (ideally, for an output-gap-adjusted price level), rather than for a target rate of inflation looking forward, as a forward-looking inflation target accommodates a permanent decline in the price level after a period of one-sided target misses due to a binding zero lower bound on interest rates. Our analysis implies that a credible commitment to the right kind of “exit strategy” should substantially improve the ability of monetary policy to deal with the unusual challenges posed by a binding zero lower bound during a deep financial crisis, and to the extent that this is true, the development of an integrated framework for policy deliberations, suitable both for crisis periods and for more normal times, is a matter of considerable urgency for the world’s central banks.

Serious scholars use much more measured language than I do.  So when Woodford and Curdia conclude by saying level targeting “is a matter of considerable urgency for the world’s central banks” they are sounding the alarm.  When I criticized Woodford earlier for suggesting that rates needed to be held at a low level for a long period of time, I had forgotten that they also favored level targeting.  In my view, this sort of policy would make the “real shock” from the financial sector much less severe, so that the nominal interest rate would not have to be held at zero for very long at all.

Recall that in my debate with James Hamilton I argued that NGDP level targeting around a plus 5% trend line would have had a dramatic effect in terms of propping up near-term output, even during the fall and winter of 2008-09.  Hamilton argued that policy lags meant that a policy of level targeting would not have accomplished much in the short run.  Looks like Woodford and Eggertsson agree with me.  BTW, for those who think my ideas on monetary policy are a bid extreme, consider that Woodford wrote the monetary textbook that has become the standard reference at upper level PhD programs.  So maybe we don’t need a “new paradigm,” maybe we just need to implement the best paradigm that we currently have.

HT:  Marcus.

Update:  “123” sent me the pdf for the graph



20 Responses to ““A serious mistake?” Yeah, I’d say so.”

  1. Gravatar of JimP JimP
    13. November 2009 at 11:49

    Extraordinary.It is exactly as if they had read one Scott Sumner.

    I hope Bernanke reads this. And he actually might. Woodford is a friend and a former co-author. They don’t need to follow the radical Sumnar – but rather their own guy. Is there any chance they might do so? It is utterly clear what they should do. Don’t give us Japan. There is no need to do so. Ignore the deflationists. Target the price level.

  2. Gravatar of JimP JimP
    13. November 2009 at 11:52

    Spelling on the second Sumner. Sorry.

  3. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    13. November 2009 at 12:25

    OT, but Steve Landsburg has some fun:


    It’s always impressive to see one person excel in two widely disparate activities: a first-rate mathematician who’s also a world class mountaineer, or a titan of industry who conducts symphony orchestras on the side. But sometimes I think Paul Krugman is out to top them all, by excelling in two activities that are not just disparate but diametrically opposed: economics (for which he was awarded a well-deserved Nobel Prize) and obliviousness to the lessons of economics (for which he’s been awarded a column at the New York Times).

    It’s a dazzling performance. Time after time, Krugman leaves me wide-eyed with wonder at how much economics he has to forget to write those columns.

  4. Gravatar of pushmedia1 pushmedia1
    13. November 2009 at 12:46

    JimP, Bernanke will most certainly read Woodford.

    Professor, maybe I’m too dense, but I don’t see what switching regimes NOW get us? The policy mistake was made; it got us in a mess. Wouldn’t switching regimes now cause more problems by unanchoring expectations?

  5. Gravatar of Scott Sumner Scott Sumner
    13. November 2009 at 12:53

    JimP, That’s what I’m hoping. I think that relative to the bland nature of most academic writing, there is an undertone of concern that the world’s major central banks are making a big mistake. I just hope they don’t wait until it’s too late. There is a tendency of governments, like generals, to be always fighting the last war. Eventually, Woodford’s approach will sink in. But they need to move relatively quickly if we don’t want another year of 10% unemployment.

    Patrick, Thanks, I enjoyed that. Landsburg’s a very smart guy, and is much tougher on Krugman than I am. BTW, countries like France and Germany have averaged around 8-10% unemployment over the past decade, so why would Krugman want to use them as a model?

  6. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    13. November 2009 at 13:36

    It reminds me of the 1992 Presidential campaign. Lester Thurow had just been featured on 60 Minutes promoting the German economic model, so Bill Clinton worked that into his spiel during one of the debates with HW and Perot.

    Of course, the US was a year and a half into what turned out to be a ten year expansion at the time.

  7. Gravatar of 123 123
    13. November 2009 at 14:02

    figure 11 JPG is available here:

    direct link:

  8. Gravatar of Scott Sumner Scott Sumner
    13. November 2009 at 14:18

    pushmedia1, I understand your point. It is similar to the argument against tough love to hold down moral hazard after a banking debacle has occurred. the answer is that it is never too late to start the right policy, as it will help the next time around if the public understand the Fed would let NGDP fall to a lower track permanently. We may not want to return all the way to the previous track, as you say, but we should at least try to return part way, as it would boost the speed of what looks to be a very slow recovery. Perhaps we could split the difference, announce that after falling 8% below target, we would try to return to a new trend line only 4% below the old one. That would keep inflation levels pretty low, but still allow for faster recovery than staying on a trend line 8% below mid-2008.

    But you are right, it would have been 10 times better to do this a year ago.

    Patrick, I’d like to see a candidate push the Singapore model, but unfortunately it is hard to explain to voters (maybe that’s why it took a relatively authoritarian government to enact it.)

  9. Gravatar of pushmedia1 pushmedia1
    13. November 2009 at 15:41

    Its like the issue with moral hazard in bank bailouts except there’s no moral hazard. The best time, economically, to implement regime change at the Fed would be when the change can be least disruptive, i.e. when we’re not just recovering from a severe recession. If we implemented this change a year ago, during the crisis, expectations would have become even more unanchored… the public wouldn’t have a clue about what the Fed’s policy would be.

    Are you taking the Friedman/Emanuel model of policy change?
    Friedman: “Only a crisis, actual or perceived, produces real change.”
    Rahm Emanuel: “You never want a serious crisis to go to waste.”
    Sumner: Exploit this crisis to implement my pet policy! 🙂

    More seriously, how do you think the public sets expectations about Fed policy? I subscribe to a learning theory of expectation setting. By watching the Fed, they learn if its credible or not. They don’t just believe everything the Fed says and so policy announcement don’t automatically translate into expectations. Instead it takes time for the public to learn the new policy is credible. (Isn’t this an explanation for why the Volker disinflation took so many attempts over several years to finally tame inflation?)

    If the Fed dramatically changes policies, expectations are unanchored from Fed policy. This is why changing regimes in the middle of a crisis is a bad idea.

    This is just me shooting from the hip. I don’t know of any relevant research.

  10. Gravatar of Doc Merlin Doc Merlin
    13. November 2009 at 16:11

    Honestly, Scott constant NGDP growth targeting makes me uncomfortable. It suffers from similar problems as Inflation rate targeting, and doesn’t fix what I see as the main problem. Market expectations need some way of being fed back into the supply side of money. The easiest ways to do this is free banking, and/or futures convertibility. Anyway, grats that more people are taking your pet idea seriously, Scott.

  11. Gravatar of JimP JimP
    13. November 2009 at 16:40

    Everyone should really read that paper. I see no big policy difference between the paper and the hopes of this blog. It is a big deal that Woodford endorses this, I think. Maybe this means Bernanke endorses this as well, and is trying to collect a little political support against the deflationists. A good “exit strategy” for the Fed would obviously be to do what the paper suggests.

  12. Gravatar of JimP JimP
    13. November 2009 at 17:06

    And one last thing – I think the paper makes it entirely impossible for Krugman et al to now claim that theory says that monetary policy has no power at the zero bound. And if he does continue to claim that, professional economists well clearly understand that he is what he is – a political hack.

  13. Gravatar of van van
    13. November 2009 at 18:30

    may i ask one (ill-timed) question. I believe Austrians contend that economies need a period where prices are allowed to readjust in order to cleanse the system of malinvestment. scott and others, do u agree with such a contention? and if so, how does an economy adjust prices when Fed policy explicitly targets a general price level in the future? perhaps there is a nuance in ABC theory that i am missing – but it seems the argument is both relative prices and price levels need time to adjust.

  14. Gravatar of malavel malavel
    14. November 2009 at 03:47


    I’m sorry but I think you can throw your plan in the trash can now. Not even the Swedish Pirate Party dared to adopt it since it was deemed too radical. If they adopt a monetary policy suggestion it will probably be something mainstream like fractional reserve banking is fraud and the FED is teh evil. Again, I’m sorry if I got your hopes up earlier.


  15. Gravatar of rob rob
    14. November 2009 at 08:17

    Krugman capitulates! (And admits he’s unwilling to argue for a first best solution.)(As long as he believes he has no influence, why does he bother showing up for work?)

  16. Gravatar of JimP JimP
    14. November 2009 at 08:58

    Yes he does. And here is the column – written in his usual nice and non-insulting way. What a clown the man is.


  17. Gravatar of rob rob
    14. November 2009 at 11:07

    Krugman’s point number 3:

    “(iii) Convince investors, consumers, and firms that you have in fact achieved (i) and (ii).”

    really goes to show he doesn’t believe in the wisdom of the market. But if you don’t assume the market will get it right, then…

  18. Gravatar of Scott Sumner Scott Sumner
    14. November 2009 at 11:15

    Pushmedia1, We did do a regime change last year. We went from positive 5% NGDP growth to sharply negative nominal growth. And the results were very bad, as you indicated. Expectations became completely unanchored, as you indicated, and now nobody has a clue as to where we are going. A 5% NGDP rule would have anchored expectations last year.

    So I think we both agree, the Fed should have stuck to the “Great Moderation” policy, instead of switching to a deflationary policy in the midst of a crisis. Indeed the crisis was caused by regime change.

    Doc Merlin, We agree, I favor NGDP targeting through futures convertibility.

    JimP, Krugman should stop dancing around the issue and come out strongly for a more expansionary monetary policy. What is the worst that could happen from his perspective? Remember that he already favors much more stimulus.

    van, I don’t think prices need just “a time” to readjust, I favor allowing them to always readjust to reflect changing market conditions. I completely oppose price controls that would inhibit this adjustment. Of course relative prices can adjust regardless of the price level, but the adjustment process is easiest when the macro economy avoids unexpected NGDP shocks. That’s why Hayek favored NGDP targeting.

    malavel, Thanks for trying. I understand that pirates are cautious people, who don’t just adopt any new idea that comes along. Perhaps once they’ve had time to mull it over.

    rob and JimP. Thanks, that may be worth a post.

  19. Gravatar of van van
    14. November 2009 at 13:14

    scott, ok then i think i am getting confused on the semantics. i thikn what u r saying is that under NGDP targeting, all prices are allowed to function freely and properly. for whatever reason i was getting hung up on the fact taht if we are talking about “price level” targeting, somehow we would be institutionalizing price controls, and the only direction they could go is up. so the question: is woodford’s price level targeting the same in concept as NGDP targeting?

  20. Gravatar of Scott Sumner Scott Sumner
    15. November 2009 at 08:12

    van, Yyes. it’s the same concept but a different target. Price level targeting still allows complete freedom of the relative price of each good to move up or down against other goods.

Leave a Reply