Sunday, May 15, 2011

My Reply to Paul Krugman

Paul Krugman appreciates my efforts against the hard money advocates.  He questions, however, my and other quasi-monetarists' belief that monetary policy can still pack a punch when short-term interest rates hit the zero bound:
[T[hey want to keep that policy action narrowly technocratic, limited to open-market operations by the central bank.  As I’ve argued before, this doctrine has failed the reality test: liquidity traps are real, and blithe assertions that central banks can easily pump up demand even in the face of zero short-term rates have not proved correct.
It is true that us quasi-monetarists believe that the efficacy of monetary policy is not limited by the zero bound, but we have never said the that all it takes is further open-market operations.  Rather, we have said that monetary policy can be highly effective regardless of circumstance if the following steps were taken by the Fed:

(1) Set an explicit nominal GDP level target so that expectations are appropriately shaped.  If such a rule were adopted expectations of current and future nominal spending would be anchored around the level target and make it less likely there would be aggregate demand crashes like the one in late 2008, early 2009. Even if a spending crash did occur the catch-up growth needed to return nominal spending to its level target would most likely imply an expected path of short-term real interest rates consistent with restoring full employment. (See here, here, and here for more on a nominal GDP rule.)

(2) Purchase assets other than t-bills as needed to make sure the nominal GDP level target is maintained.  Thus, if the monetary base and t-bills became perfect substitutes because the 0% bound is reached the Fed should buy longer-term treasuries or foreign exchange.  Nowhere have we said simple open market operations in t-bills would always suffice. A big difference, though, is that where Krugman and others see the move from t-bills to other assets as a discrete jump from conventional to unconventional monetary policy, quasi-monetarists see it as simply moving down the list of assets that can affect money demand.  The 0% bond for us really is not a big deal, but simply an artifact of monetary policy using a short-term interest rate as the targeted instrument.

In practice, this understanding is not that different operationally than a New Keynesian invoking a higher inflation target to lower the expected path of real interest rates or the portfolio channel to drive down the term premium on long-term bonds.  We just do it with a lot less angst.  Maybe Andy Harless with his modified Taylor Rule can bridge the gap between us.

Finally, because of these views we believe the Fed could have done much more in late 2008, early 2009.  Its failure to do so amounted to a passive tightening of monetary policy then.  Even now monetary policy is not all that loose given that money demand continues to be elevated.  Don't believe me? Then just look at domestic spending per capita, it is still below its pre-crisis peak. 

Update:  Fellow quasi-monetarist Bill Woolsey also responds to Krugman.


  1. A big difference, though, is that where Krugman and others see the move from t-bills to other assets as a discrete jump from conventional to unconventional monetary policy, quasi-monetarists see it as simply moving down the list of assets that can affect money demand. The 0% bond for us really is not a big deal, but simply an artifact of monetary policy using a short-term interest rate as the targeted instrument.

    This seems to be the heart of the disagreement.

    I don't see why the zero lower bound is merely an "artifact" of using the short-term interest rate as a policy instrument. Once you hit the zero lower bound, money becomes just another riskless short-term asset. Above the zero lower bound, the Fed's monopoly power in producing base money means that it can have a massive effect with comparatively small changes in its balance sheet.

    At the zero lower bound, on the other hand, it's just another producer of liquid, riskfree assets---it's entering into a far, far, far larger market dominated by the Treasury, the GSEs, and private banks. There is clearly a discrete jump in difficulty here: the Fed is no longer a monopoly supplier of the asset in question, and it somehow has to make an impact in a market much larger than its balance sheet has even been. Under some assumptions, like those in the Eggertsson and Woodford 2003 BPEA paper, it can't have any impact at all. I don't think that these assumptions hold perfectly in practice, but the intuition behind the result offers additional reason to doubt that it is easy for the Fed to make a difference through asset purchases at the zero lower bound.

    I want to emphasize that the impact of QE is extremely context-dependent. At the height of the financial crisis, I suspect that the Fed's unconventional asset purchases were extremely powerful. The key problem at that stage was a spike in the liquidity premium; this meant that even if the policy rate fell to zero, the effective rates facing consumers and businesses were still very high. Since the liquidity premium presumably resulted from a sudden shortage of safe assets (as the AAA tranches of MBS were no longer thought to be invulnerable), it was conceivable that the Fed could step in to alleviate the shortage and bring down effective interest rates.

    At this point, however, I don't think that an acute shortage of safe assets is a problem to nearly the extent it was in the fall of 2008. It might still matter a little---and that's why additional asset purchases would be better than nothing---but it's not clear that the Fed can have much of an impact now that the market for safe, liquid assets has calmed down.

  2. From Krugman's post on Friedman:

    "So, after 2000 the Bank of Japan engineered a huge increase in the monetary base; this was the original quantitative easing. And it didn’t even translate into a surge in the money supply! This is why I’m so skeptical of people who say that all the Fed has to do is target higher nominal GDP growth — in liquidity trap conditions, the Fed doesn’t even control money, so how can you blithely assume that it controls GDP?"

    How would you respond to that, David?

  3. From the “artifact” post you reference:

    “Just let that time path for the nominal magnitude grow over time at a fast enough rate and the equilibrium overnight rate will rise above zero. If monetary policy were framed as the central bank setting some nominal variable, expectations of monetary policy could then coalesce around that variable, monetary policy would be loosened, and the overnight rate, as some endogenous response to monetary policy, would rise above zero.”

    This is an example of the type of thing that Krugman is objecting to, I think.

    “Letting something happen” means you must have the option of stopping it from happening. How does the central bank “let” this time path happen? The active setting of the interest rate seems to be prohibited. So I assume it’s the active setting of the base, as per Sumner’s proposal, for example. And so on to Krugman.

  4. I would add that the NGDP targeting should be LEVEL targeting. If you do that then there will be no liquidity traps.

    I partly agree with Matt and partly disagree. Buying longer term bonds doesn't, by itself, address the objections raised by Eggertsson/Woodford/Krugman. Indeed if the expectations hypothesis holds, the expected return on long bonds (held over the next 12 months) is approximately zero.

    The reason why QE2 "worked" (i.e. raised inflation expectations) is because the markets correctly saw it as the Fed signaling a slightly higher implicit inflation target. Unfortunately, inflation targeting is the wrong approach, and Bernanke ran into the bad luck of a supply shock (oil prices) right after the program was started. That's one reason we need a NGDP target, not an inflation target.

    I disagree on the crisis vs. non-crisis part of Matt's comment. QE1 was not effective because of the program of interest on reserves. Indeed the Fed paid interest on reserves in late 2008 precisely in order to prevent interest rates from falling below 1%, i.e because the Fed didn't want the new money to boost inflation, rather they wanted to rescue the banking system. Unlike QE1, QE2 did raise inflation expectations, precisely because the markets correctly understood that the Fed had decided 0.6% core inflation was intolerably low. That's why core inflation has since risen, and that's why the economy avoided the double dip that loomed in August 2010, when most economists forecast no reduction in the unemployment rate during 2011.

    JKH, The Bank of Japan is acting exactly like a central bank that doesn't want any inflation. They tighten monetary policy whenever the rate of inflation rises to zero percent (as in 2000 and 2006. The QE of the early 2000s in Japan was correctly understood to be temporary, and hence didn't have much effect. That's a prediction both of Krugman circa 1998, and of quasi-monetarism (including a paper I wrote in 1993), which argued that temporary currency injections won't raise prices. The BOJ presents no problems for quasi-monetarism.

  5. JKH,
    Just an addenda to what Scott said about Japan:

    1) During 1993-2002 RGDP growth averaged 0.9%, unemployment rose almost consistently every year from 2.2% in 1992 to 5.4% in 2002. CPI fell from 2.5% in 1992 to -0.7%in 2002.
    2) The Japanese announced their plan of ryōteki kin’yū kanwa (QE) on March 19, 2001 and maintained it through March 9, 2006.
    3) RGDP growth averaged 2.1% during 2003-2007. Unemployment fell every year until it reached 3.9% in 2007. Deflation slowed down to -0.4% by 2007.

    So the real growth rate more than doubled under QE, disinflation ceased and unemployment fell for the first time in a decade. Ultimately the success of QE should not be measured by money supply (as Krugman did) but by NGDP.

  6. Krugman is just using you to score points against Republicans; he is being characteristically insincere.

    His accusation that you advocate technocratic policy action, and his suggestion that you "blithely" dismiss liquidity traps, belie his veiled contempt. You have consistently explained why you believe monetary policy can be effective when some interest rates approach zero, and you have regularly criticised the concept of central banking.

    Krugman is fanning the flames by suggesting otherwise. He doesn't want people like you to find a place in the Republican party: the more kooks and quacks advocating hard money the better.

    Okay, maybe I am being a wee bit too cynical, but Krugman always seems to be playing an angle.

  7. Just to be clear, I completely agree with Scott that QE2 worked primarily through the "signaling channel"; shaping expectations about the future trajectory of the policy rate can be a very powerful tool indeed, as I think all sides of this discussion (both David and Paul Krugman) acknowledge.

    I just think that conducting monetary policy without somehow affecting the policy rate (either by changing it today or shaping future expectations) is, while possible, far harder than conventional monetary policy---since you need to have an impact in a vast, vast market where the Fed is no longer a monopolist.

  8. Money is never just another safe and liquid asset, because unlike those assets money is the medium of exchange. Disequilibria for safe and liquid assets instigate price and quantity adjustments in their respective markets. But money does not have a price of its own, and its quantity does not rise or fall when people wish to change their aggregate holdings. Just because an asset is a close substitute for money from the perspective of the individual doesn't mean it has similar macroeconomic consequences.

    If an excess demand for safe and liquid assets spills over into money (and base money), then the Fed must act to satisfy the increasing money demand or nominal spending will fall.

    If the asset the Fed normally purchases to alter the supply of money becomes a near perfect substitute for money, then it simply needs to start buying something else until desired money holdings equal supply.

  9. This comment has been removed by the author.

  10. I just say the Fed needs to flood so much money into the system we start to have growth, then inflation.

    That makes sense to me, given we are dancing around deflation, and the GDP is depressed and spare capacity is everywhere.

    I will leave it to the academics to put fancy pants explanations on it.

    For me, just print money until the plates melt.

  11. Matt:

    Your concern about the zero lower bound causing the Fed to become just another producer of safe assets in a far bigger market dominated by the Treasury, GSEs, and banks is only an issue from the store of value perspective of money. Yes, from this perspective investors are indifferent. And from this perspective Steve Williamson's claim that QE2 really isn't all that special makes some sense.

    However, as Lee mentions, money is more than a safe, liquid asset. It is the medium of exchange, the only asset on every other good, service, and asset market. Thus, it is the only asset directly capable of disrupting all other markets. And here, the Fed is still a monopolist--it is only entity creating the medium of exchange.

    From this medium of exchange perspective of money, when the Fed is buying up other non-tbill assets it is more than just supplying more safe assets, it is trying to address a medium of exchange problem (i.e. excess money demand). That is why when I think of the portfolio channel I see more than just lowering yields and portfolios adjusting. I see a money demand problem being addressed. It also why I think the shaping of nominal expectations--the signalling channel--is ultimately an exercise in addressing money demand problems.

    As an aside, I would note that this understanding implies the paradox of thrift and balance sheet recessions are better seen as an excess money demand problem as I explain here.

  12. Matt:

    One other thing. If unconventional monetary policy really is so hard then why was FDR's unconventional monetary policy so darn effective in 1933-1936 as I show in this post? (See Christina Romer and Gauti Eggertson for more formal treatments of this episode.)

  13. JKH:

    If I understand your second question, the answer for how to set the policy path is for the Fed to clearly and aggressively communicate to the public that it is going to hit some nominal level target no matter the costs. The expectation of such actions by itself would cause the market to do most of the heavy lifting.

    FDR, for example, with his fireside chats and other speeches made a point to say he was aiming to return the price level to its pre-crisis level. He backed that talk up by devaluing the gold content of the dollar and ordering the Treasury not to sterilize incoming gold flows (it had been doing that). FDR's rhetoric and his actions scared many people about higher future inflation and created exactly the kind of shock needed to reduce the excess money demand problem and get nominal spending going again.

    What instrument the Fed should use--the monetary base or a short-term interest rate--should be irrelevant if the policy is believed. Bennet McCallum, for example, has a famous monetary policy rule (popular long before the Taylor Rule) that uses the monetary base as the targeted instrument. One could easily use a short-term interest rate with his rule. As noted above, though, I believe targeting a short-term interest rate creates confusion when the zero bound is hit.

  14. All:

    In case you missed it, fellow quasi monetarist Bill Woolsey, I mean Mayor Bill Woolsey, has replied to my exchange with Krugman.