Friday, February 8, 2013

The Train Has Already Left the Station: A Reply to Paul Krugman

Paul Krugman says now is not the time to cut government spending. Why? Because the Fed is out of ammunition and cannot possibly provide any offset to the fiscal drag such spending cuts would make:
Today, by contrast, we’re still living in the aftermath of the worst financial crisis since the Great Depression, and the Fed, in its effort to fight the slump, has already cut interest rates as far as it can — basically to zero. So the Fed can’t blunt the job-destroying effects of spending cuts, which would hit with full force.

The point, again, is that now is very much not the time to act; fiscal austerity should wait until the economy has recovered, and the Fed can once again cushion the impact. 
There are two big problems with this analysis. First, fiscal retrenchment has already started and has been happening since mid-2010. The fiscal austerity train has already left the station and shocker of shockers, it has not caused the cataclysmic collapse in aggregate demand that Paul Krugman fears. Here is a figure from a earlier post that shows NGDP growth has been relatively stable despite the reduction in total government expenditures:


The same story emerges if one looks at the shrinking budget deficit. Note also that short-term interest rates have been near 0% since early 2009. So somehow the Fed has been able to keep NGDP growth stable despite (1) a fiscal contraction and (2) the zero lower bound on short-term interest rates. According to Paul Krugman this is not possible.

The second problem, then, with Krugman's analysis is his claim that the Fed is out of ammunition at the zero lower bound. He, of all people, should know this is not the case given his seminal work on Japan. He should also know this given all the exchanges he has had with Scott Sumner and me over the past few years. But just to be thorough, let me spell it out once again why the Fed is far from powerless.

To begin, consider how the public would respond if the Fed suddenly announced it was raising its asset purchase by 20% per month until some NGDP level target was hit. That would be the monetary policy equivalent of shock and awe and should catalyze the mother of all portfolio rebalancings. Households and firms would start spending their built up stock of money assets on other riskier assets (like stocks and physical capital) as well as goods and services in expectation of higher nominal income and temporarily higher inflation. Asset prices would increase, household balance sheets would be repaired, and real debt burdens reduced. This would reinforce the recovery in NGDP growth.  

The key to this working is permanently raising the expected size of the monetary base. When this happens, expected nominal incomes will go up and lead to the responses outlined above. Thus, even though open market operations at the zero lower bound might be trading near substitutes now--monetary base for treasuries earning 0%--the belief that they won't stay near substitutes in the future (because of the permanent increase of the monetary base) will trigger the portfolio rebalancing that will lead to higher nominal spending. 

Under QE1 and QE2 this did not happen precisely because the public did not expect the increase in the monetary base to be permanent, a point noted by Michael Woodford. But it has happened before. Back in 1933 FDR effectively took the reigns of monetary policy from the Fed and credibly committed to a higher monetary base level by devaluing the dollar.  As consequence, the U.S. saw one of its sharpest recoveries that year. The same can happen now by adopting an aggressive, but well anchored nominal GDP level target.

13 comments:

  1. David,

    It's incredible how two people can look at the same graph and not at all see the same thing. What *I* see is rapidly recovering NGDP right up until your dotted line when it's suddenly cut short by fiscal retrenchment. Just goes to show you that you can't prove much with pretty pictures (also applies to Krugman). Robust conclusions demand robust econometrics.

    "So somehow the Fed has been able to keep NGDP growth stable despite (1) a fiscal contraction and (2) the zero lower bound on short-term interest rates. According to Paul Krugman this is not possible."

    Simply not true. You are assuming there are no other confounding variables, i.e. ignoring supply side dynamics which may have been raising the natural rate over the same period.

    "Under QE1 and QE2 this did not happen precisely because the public did not expect the increase in the monetary base to be permanent, a point noted by Michael Woodford."

    Woodford said: "The problem is that, for this theory to apply, there must be a permanent increase in the monetary base."

    Also for said theory to apply reserves must pay zero IOR or the CB cannot control the quantity of non-interest-bearing money: the system would simply convert currency to interest bearing reserves. So you are saying that the CB should credibly commit to *irreversably* expanding the base by k%/year. I know you intend that it should somehow be contingent on NGDP. But then, how is it permanent? What if velocity spikes?

    Don't forget that the fed funds rate collapses to zero as soon there is a tiny excess of reserves. Any rule setting excess reserves as a function of NGDP will simply amount to a rule determining when the fed funds rate will be allowed to rise above zero. And until then, the quantity of excess reserves is *irrelevant*. I.e. it's nothing more than a contingent short rate rule that says when we will exit the ZLB. And once FF has left the zero bound you really can't set a reserve quantity rule, because any deficient quantity of reserves would cause extreme spikes in the FF rate. Better just to stick with a rate rule i(NGDP) (not M(NGDP)) since that's what you'll be doing anyways.

    ReplyDelete
    Replies
    1. K, art is in the eye of the beholder, no? :) Seriously, note the figure shows the growth rate of NGDP, not the level. In level form, it increasing at a steady pace as it did before the crisis. (http://research.stlouisfed.org/fred2/graph/?g=fnH) It is not at level it should be, but its rate of growth is similar to its pre-crisis rate. And this is despite the pullback in government expenditures.

      "You are assuming there are no other confounding variables" No, I am not. What I am saying is that despite all that is going on--fiscal retrenchment in particular, but yest there are other shocks(Eurozone crisis, etc)--the Fed is keeping NGDP growth relatively stable. That is a remarkable accomplishment. By keeping NGDP stable, the Fed has effectively offset the effect of these other shocks.

      Not sure what your point is about Woodford, but in multiple places--including his Jackson Hole speech--he has said a permanent increase in the monetary base is needed for monetary policy to pack a punch.

      Delete
  2. David,
    Permanence raises some interesting questions in terms of the mechanics of policy:

    -If the Fed declared the current reserve level permanent, how would it raise the FFR in the future?

    -If it only declared part of reserves permanent, how large should that permanent portion be?

    ReplyDelete
    Replies
    1. Diego, good questions. Clearly, we don't want all of the monetary base increase to be permanent since that would be disastrous. So what we want is just enough to satiate excess money demand and get nominal spending at on an appropriate path. How much that is is impossible to know. That is why a NGDP level target is important-you don't have to know, just aim the for the target and let the endogenous interaction of money demand and money supply figure it out.

      Delete
    2. David,
      Even under NGDP targeting, the Fed would still need to pre-announce how much of the base is permanent in order for "permanence of the base" to affect expectations.

      How can the Fed determine beforehand what the optimal level is for a "permanent" base? It seems to me that it either too low or too high would be quite risky. For instance, too low and expectations don't budge. Moreover, the bulk of QE would have to be reversed, which would likely disrupt the bond market. Too high, and the Fed would be powerless to arrest an inflation expectations spike. It needs to be, "just right".

      Perhaps there is a way to iterate ("trial and error") to determine, "just right", but I'm not seeing it.

      Delete
    3. Diego, I don't see why it would have to announce exactly how much of the base would need to be permanent. FDR didn't, but he did create the impression that it would permanently go up enough to restore some of the decline in the price level. That is how I see it working.

      Delete
    4. David,
      "... impression that it would permanently go up..."

      If the Fed doesn't have to set the base at an initial optimal level, then it doesn't matter whether the base shrinks or expands to hit an NGDP target. Why mention the base at all? I think this is what K is saying below as well.

      Delete
  3. Sorry, but there are several things wrong with your analysis. First off, "cataclysmic" is your word, not Krugman's. It would really be refreshing to see somebody do a KRUGMAN IS WRONG post without injecting some sort of a straw man.

    Second is your willful disdain for nuance. Growth and austerity are relative terms, not a full-on full-off absolute binary pair. You should realize that Krugman has been saying for years we are doing worse than we could be, and that further austerity will make things worse yet.

    Third, let's look at real GDP and at spending changes without a distorting denominator. Here is YoY % change in both, from FRED.

    http://research.stlouisfed.org/fred2/graph/?g=fnE

    To my point 2: Spending growth in actual dollar terms only went scarcely below 0.0% and only for a short span. Meanwhile real GDP growth is stagnating at the lowest non-recessionary level of the post WW II period.

    This is mild austerity coupled with anemic growth, and is fully consistent with what Krugman has been telling us for a long time.

    Last - and I've told you this more than once already - you have to be very careful about drawing broad conclusions from a short time series.

    Take the long view and it becomes pretty clear that Krugman has pegged things pretty accurately.

    Cheers!
    JzB

    ReplyDelete
  4. Jazzbumpa, the denominator is not meant to be tricky but to show the declining contribution to overall aggregate demand (i.e. NGDP) coming from government expenditures. Whether one shows it absolute dollars or scaled as I do, it has been declining since 2010. Yes, it hasn't be a large drop but by Krugman's own definition it should have led to a slowdown in NGDP growth. That has not happened.

    ReplyDelete
  5. I do understand that you are not playing a denominator game. But without it, the picture is not quite the same.

    And - ok - let's look at GDP FRED series GDPC1 - this is NGDP -right? Again, YoY % change.

    http://research.stlouisfed.org/fred2/graph/?g=fnR

    Not a precipitous decline, to be sure, but the downward slope is undeniable - actually all the way back to 1980, if you expand the graph.

    Cheers!
    JzB

    ReplyDelete
  6. "the Fed is keeping NGDP growth relatively stable. That is a remarkable accomplishment."

    With the Fed going all out at the ZLB, NGDP growth happens to come out coincidentally flat. There is nothing special about flat NGDP growth and it certainly doesn't mean NGDP growth wouldn't have been higher with more fiscal stimulus. Or lower. Flat NGDP growth isn't evidence of anything. Even if NGDP was growing slower we could be above capacity. And even if it was back to pre-crisis trend, we might be below capacity. It all depends what happened to supply. None of which contradicts Krugman in the least. All he said is that AD isn't where it should be. There is simply nothing wrong with his statement.

    "Not sure what your point is about Woodford, but in multiple places--including his Jackson Hole speech--he has said a permanent increase in the monetary base is needed for monetary policy to pack a punch."

    I agree with you (and Woodford). My point (and it's right) is that a quantity rule, i.e. a time dependent function M(t, NGDP) is equivalent to a rate rule i(t, NGDP), since there is a relationship between M and i at any given moment in time. BUT... the value of M is irrelevant when i=0, so under those circumstances it's better to ignore M and just say specify that i(t,NGDP)=0 for those values of (t,NGDP).

    It's really important to understand *why* Woodford says that the increase has to be permanent. It's *because* it has to be big enough to cause the policy rate to be zero beyond the day when the natural nominal rate rises above zero. So it doesn't actually have to be permanent, but permanent will do. (Assuming a future state of the world in which the nominal natural rate is actually above zero - otherwise there is no monetary escape from the liquidity trap even in principle). And note that Woodford *never* says that you can get something more from a monetary expansion than you can get from interest rate policy. That's because you can't. A quantity rule is just a cryptic way to specify the short rate. (And as I said, it's essentially impossible above the ZLB).

    I don't get why Market Monetarists keep quoting this "permanent expansion" stuff from Woodford (and Krugman '98) like it offers support for MM theories for some kind of special escape from the liquidity trap, when those same authors in fact demonstrated the exact opposite. All it is, is forward guidance and in theory you can choose to state it in terms of interest rates or quantity of money. But forward M has no more advantage over forward i, than does current M over current i. And we've known that current M is a crappy instrument for over 30 years now.

    ReplyDelete
    Replies
    1. K,
      "With the Fed going all out at the ZLB, NGDP growth happens to come out coincidentally flat. There is nothing special about flat NGDP growth"

      I disagree. The Great Moderation was a period of relatively flat NGDP growth and many observers chalk it up to successful monetary policy. To the extent the Fed did play a role in this development (and I think it did), it indicates the Fed successfully offset shocks (whether real or nominal) to nominal spending. Since 2010, NGDP growth has similarly kept stable, though at a slightly slower 4% growth rate. There needed to be more rapid catch-up NGDP growth, but the fact that the ongoing Eurozone crisis , the debt limit talks in 2011, the fiscal cliff, and threats of a China slowdown did not lead to a falling growth NGDP growth rate is my view an accomplishment.

      Yes, there are identification issues and we are below capacity, but even in this context a fiscal slowdown according to Krugman in various pieces should not lead to stable NGDP growth (i.e. consistently increasing NGDP $ level), especially with all the other economic shocks outlined above. Maybe these are not his exact words, but this is impression you get from his article and blog posts.


      "And note that Woodford *never* says that you can get something more from a monetary expansion than you can get from interest rate policy."

      Yes, he has said they are equivalent and given they way money is generally specified in most macro models he is correct. However, I think where Market Monetarists part ways is in how these models are set up in the first place. Money in the utility function or cash in advance constraints really doesn't do justice to the transaction-cost reducing role of money. It also doesn't capture the Monetarist notion that monetary policy can influence the economy more than just through the expected path of the policy interest rate. In my view, some of the New Monetarist models are an improvement on this front. They don't fix all the problems, but they are making progress.

      P.S. It should be evident from the post, but I don't believe the Fed is "going all out at the ZLB." It is doing better and enough to keep NGDP growth stable, but going all out in my book would be making some catch-up NGDP growth.

      Delete
  7. David,

    "Maybe these are not his exact words, but this is impression you get from his article and blog posts."

    Not my impression. Those are your impressions, and I think they are skewed by forcing his words through your MM framework rather than looking for what he really means. From my, generally NK, perspective most of what he says seems quite defensible (even if I don't weigh risks the same as him).

    There is a trend of arguing with what people think Krugman means, rather than with what he actually said (see e.g. this deranged individual). Occasionally he makes real, technical errors like in his debate with Scott Fullwiler over how banks create money. But mostly he is pretty careful not to be technically wrong. Those who persist in portraying him as saying what they think he *meant*, are liable to end up with a severe case of Krugman Derangement Syndrome.

    "However, I think where Market Monetarists part ways is in how these models are set up in the first place."

    How do Market Monetarists set up the models?

    "Money in the utility function or cash in advance constraints really doesn't do justice to the transaction-cost reducing role of money."

    There has to be a *cost* to holding liquidity in order for there to be any sense in the QTM. If reserves yield the same as t-bills that means the non-pecuniary value of reserves is zero. I.e. there is so much of them they only serve a role as a risk-free store of value and have no additional MOE value because everyone has vastly more than they need to satisfy any conceivable liquidity needs. So the value derives exclusively from the interest rate just like the t-bills. But then, the only economic impact is via the interest rate and portfolio balance. If we pay IOR and keep FF at IOR then reserves will never have non-pecuniary value again, and then we can't even make a case for future impact of the quantity of money. But the short rate, present and future, *still* matters because the real rate matters.

    ReplyDelete