Friday, May 6, 2011

Liquidity Traps Are Very Unlikely, Even at the Zero Bound: A Rejoinder to Matt Rognlie

Matt Rognlie takes to task Tyler Cowen for claiming there is no liquidity trap. He makes his case for the liquidity trap and then closes with this statement:
This doesn’t mean that all hope is lost: the Fed can still make a difference by shaping expectations of the future trajectory of nominal interest rates, or by making unconventional bond purchases so large that they trigger portfolio balance effects and drive down interest rates on longer-maturity assets...Just don’t go around claiming that 0% isn’t a barrier—because sadly, it is.
Ironically, the above bold phrase is exactly why in most circumstances there is no liquidity trap at the 0% barrier.  To see this, first recall that a liquidity trap is a situation in which the demand for money is perfectly elastic.  That is, no matter what the central bank does it cannot cause money demand to budge.  Monetary policy, therefore, is unable to address the problems created from excess money demand in a liquidity trap.  

Now Rognlie claims this happens once the central bank's targeted short-term interest rate hits the 0% barrier.  This assumes that money demand is only affected by the targeted short-term interest rate.  Milton Friedman argued, however, that money demand is affected by a spectrum of interest rates and that there is still much the Fed can do when the short term interest rate hits 0%. The Fed can still go after long-term treasury yields and corporate bond yields which Friedman believed were also important determinants of money demand.  If so, money demand can still be influenced by the Fed and thus there is no liquidity trap.

The portfolio balance channel mentioned by Rognlie above is implicitly making this very point.  In that channel, the Fed through its purchases of longer-term securities can drive down longer-term yields.  This causes a rebalancing of portfolios that will lead to purchases of normally riskier assets like stocks and capital.  These developments imply a decline in money demand.  Thus, to claim there is a portfolio balance channel is to claim that there is no liquidity trap at the 0% bound of the short-term interest rate.  Rognlie's endorsement of the portfolio balance channel, then, implies Tyler Cowen's skepticism of the liquidity trap is well founded.

So is there any evidence that money demand can be influenced by a spectrum of interest rates?  Some older studies do show this, but for now here are three figures that suggest that answer is yes.  They all show MZM velocity (i.e. GDP/MZM) to be systematically related to various interest rates.  (The R2 for these relationships is always above 70%.)

In general, the 0% bound on short-term interest rates is not a sufficient condition for a liquidity trap.  It requires individuals to become satiated with money balances which implies there must be deflation.  Even then, Peter Ireland has shown using the Krugman liquidity trap model that if there is population growth then there will be distributional effects of money growth that will eliminate the liquidity trap.

Currently, we are far from a liquidity trap though money demand remains elevated and a drag on the economy.  That being the case, there is much that monetary policy could do to address the excess money demand problem and get nominal spending going again.  Here is one suggestion.

Update:  See Josh Hendrickson's reply.
Update II: See Scott Sumner's earlier comments on the liquidity trap.


  1. DB I sense that even you are starting to have doubts about the problem being a monetary one. It was a monetary one for a few weeks in the Fall of 2008, but that crisis long passed and almost 3 years later the long and variable lags have dissolved to leave us with the raw truth that the US real economy is apparently fatally wounded.

  2. This comment has been removed by the author.

  3. This comment has been removed by the author.

  4. Dear David,

    Thanks for taking the time to respond so carefully to my post.

    First, I want to emphasize that I'm not interested in defending a phrase; if you take the term "liquidity trap" to mean that the Fed has absolutely no tools at its disposal that could conceivably affect interest rates, then I don't believe in a liquidity trap either. For the sake of avoiding a purely semantic argument, let's take my claim to instead be the following: monetary stimulus becomes much, much harder at the zero lower bound, and fundamentally different in character from monetary policy above the zero lower bonud.

    Second, I am a little puzzled about the focus on "money demand". At the zero lower bound, this doesn't seem to be a relevant quantity: reserves and T-bills are essentially perfect substitutes. We should really be talking about "demand for short-term liquid assets" instead. And this demand isn't an end in itself: rather, the key variables are interest rates. That said, I agree that portfolio balance effects may, at least in principle, provide a way for the Fed to bring down long-term interest rates and stimulate the economy---even when it's at the zero lower bound.

    Unfortunately, it is not obvious that the portfolio balance effects from QE are of macroeconomically relevant magnitude. Eggertsson and Woodford (2003) even proved an irrelevance result stating that, under certain conditions, portfolio balance effects do not exist at all. In practice I don’t think that the assumptions necessary for this result quite hold, but they might not be so far from reality, and the intuition underlying the theorem is very important. (It's essentially a variation on Ricardian equivalence: since consumers are ultimately liable for the risk on the government's balance sheet as well, shifting the ownership of securities from the private sector to the government does nothing.) The extent to which it fails in practice is governed by the intricacies of portfolio theory, not by the kind of monetary economics you'd find in any textbook, and that is why I've been so skeptical of simple monetarist models; they don't even seem to recognize the subtleties identified by economists like Eggertsson and Woodford.

  5. Part 2 of comment:

    As I'm sure you know, Steve Williamson recently wrote a provocative post about how "QE is irrelevant". I don't agree with his precise argument---it seems likely that the Fed could get away with a smaller haircut on its repo funding than the private sector---but he seems to be thinking in the right way. When the Fed borrows short and lends long, it is effectively acting as just another financial intermediary. It has some special characteristics as a financial intermediary---namely, its size and guarantee of solvency---that may allow it to be more effective in its role, but it is still performing risk and maturity transformation that could in principle be done by the private sector. To understand why Fed intervention might be useful, we need to understand exactly what abilities it has that private banks lack. Are you using a model (even an implicit one) that clearly addresses these issues?

    In fact, I don't know whether this even really deserves to be called "monetary policy". As far as portfolio balance effects are concerned, QE2 is precisely equivalent to a change in maturity structure by the Treasury. In fact, this is true more generally: if not for political complications, QE1 could just as easily have been a Treasury operation, with a trillion in extra 3-month T-bills issued to buy MBS and other assets. Here's my question: if QE had been entirely a Treasury operation, would it have been so obvious to you that QE has real effects? If you think that portfolio balances are a key part of the Fed's influence on the economy, then the answer has to be "yes". (On the other hand, since I think that the effect is mainly from signaling instead, I view Fed and Treasury intervention as different.)

    I am afraid that I don't understand what your plots are supposed to establish. "Velocity" is higher when short-term interest rates are higher; I think we all agree with this. In the US time series, both the 10-year Treasury yields and the AAA corporate bond yields you've included here are highly correlated with 3-month rate. It is therefore no surprise that they also bear some similarity to the velocity time series; this is possible even in a world where velocity depends only on the short rate.

  6. Part 3 of comment:

    "In general, the 0% bound on short-term interest rates is not a sufficient condition for a liquidity trap. It requires individuals to become satiated with money balances which implies there must be deflation."

    I don't follow. "Satiation with money balances", as I interpret the phrase, implies that at the margin money offers no liquidity services that bonds do not. This certainly seems to be the case right now: the 3-month Treasury yield is actually lower that the rate paid on reserves (this is impossible to even rationalize in most monetary models, and presumably it's due to some microstructure issues that macroeconomists generally ignore). But satiation with money balances doesn't mean that portfolio balance effects are ruled out: they don't really have anything to do with money at all. So I'm confused---you say that I don't really believe in a "liquidity trap" because I acknowledge the possibility of a portfolio balance channel, but then you use a "satiated with money balances" definition for liquidity traps that it's actually consistent with an operative portfolio balance channel. Are portfolio balance effects possible in a liquidity trap, or are they not?

    I also don't see why satiation with money balances implies deflation. Surely we can imagine the following universe: the Fed trillions of dollars on its balance sheet forever. Consumers are satiated with money balances. Therefore the gap between the policy rate and the rate paid on reserves must be (approximately) zero, and the Fed uses IOR to control the policy rate and implement a Taylor rule with an inflation target of 2%. What is impossible about this universe? Or are you using a different definition of "satiation with money balances"?

  7. Matt:

    Thanks for the thoughtful comments. Let me attempt to answer them in turn.

    (1) My emphasis on money demand is based on the understanding that it is excess money demand that causes recessions, not excess demand for safe assets. Money is the only assets on every market and thus the only one can cause a contraction in nominal spending. The demand for safe assets only matters to the extent that it spills over to the safe asset money. I am convinced there is still an excess money demand problem.

    Let me throw the question back at you: why do you assume money demand is only affected by the short-term interest rate?

    (2) Regarding the portfolio channel, my view is that it is always there. What is unique now is that because tbills and money have become near perfect substitutes there is no portfolio channel effect by simply swapping the tbills and money. Thus, the Fed has to go after longer term securities.

    Regarding the potential magnitudes of the portfolio channel, there are several empirical studies out in the past year that show it is large. For example, see this paper by Gagnon, et al.

    (3) Regarding Steven Williamson's claim, let me simply refer to Adam P's reply:

    (4) One of the reasons QE2 has not been as effective as it otherwise would have been is because the U.S. Treasury is actually shortening the average duration. Despite this setback, QE2 has done some good via the changing of inflation expectations. Ultimately, then, QE2 is both about the portfolio balance channel and signaling. It failed on the former and did okay on the latter.

    (5) QE2 would have been far more successful on the signalling front had the Fed set an explicit level target. (Obviously, I would have liked a NGDP level target). As Ryan Avent put it, QE2 changed direction of monetary policy but didn't establish the destination.

    (6) The plots show that velocity (a proxy for money demand) appears to be systematically related to several interest rates. It doesn't prove anything, but given there are good theoretical reasons to believe money demand is affected by a spectrum of interest rates is shows evidence consistent with this view.

  8. Matt:

    (7)Regarding satiation with money balances, as I outlined above I don't believe there is satiation now because money demand is a function of a spectrum of interest rates.

    (8) One of problems with the Eggertson and Woodford model you cited is that there is no population growth, no distributional effects, and thus no real balances channel. See Josh Hendrickson's reply to your comments for more on this point:

    (9)On a different but related note, let me repeat a question I posed to you on your blog. Imagine the Fed targeted the price level such that there was enough inflation to lower the path of real interest rates such that full employment was restored. Why would the short-term nominal interest rate hitting zero matter?

    (10) Finally, we have good reason to believe that monetary policy can pack a punch when the short-term interest rate hits 0%. FDR's 1933-1936 monetary policies showed us this. See here: