Believe those who are seeking the truth. Doubt those who find it. Andre Gide

Thursday, March 5, 2015

Lifting Off...Sooner or Later

From Barron's yesterday we have this lovely headline: Two Fed Presidents Contradict Each Other on Same Day.
From the dovish corner, Charles Evans, president of the Chicago Fed, suggested that the Fed should be patient about raising rates and not act until 2016. He said: 
Given uncomfortably low inflation and an uncertain global environment, there are few benefits and significant risks to increasing interest rates prematurely. Let's be confident that we will achieve both dual mandate goals within a reasonable period of time before taking actions that could undermine the very progress we seek.
Weighing in for the Fed hawks, Kansas City Fed president Esther George said she thought the Fed should raise rates mid-year. Her take: 
This balanced approach framework supports taking steps to remove the extraordinary amount of monetary accommodation currently in place. The next phase in this process is to move the federal funds rate off its near-zero setting. While the FOMC has made no decisions about the timing of this action, I continue to support liftoff towards the middle of this year due to improvement in the labor market, expectations of firmer inflation, and the balance of risks over the medium and longer run.
I want to evaluate these two views in the context of a Taylor rule. The Taylor rule is simply a mathematical representation of how the Fed should (or will) set its policy rate in relation to the current state of the economy as measured by inflation gaps (inflation minus target inflation) and output gaps (output minus potential output). Every FOMC member presumably has a Taylor rule in mind if for no other reason than the existence of the Fed's dual mandate (the Congressional mandate that the Fed strive to stabilize inflation and employment around some long-run targets). 
A simple version of the Taylor can be written in this way:
i(t) = r* + p* + A[p(t) - p*] + B[y(t) - y*]
where i(t) is the nominal interest rate (IOER) at date t, p(t) is the inflation rate at date t, and y(t) is the (logged) real GDP at date t. The starred variables are long-run values associated with the real interest rate (r*), the inflation target (p*) and the level of "potential" GDP (y*). The parameters A and B govern how strongly the Fed reacts to deviations in the inflation target [p(t) - p*] and the output gap [y(t) - y*]. 
Let me start with the hawkish view (see also this presentation by Jim Bullard). According to this view, y(t) is below, but very close to y*. So, let's just say that the output gap is zero. PCE inflation is presently around p(t) = 1%. We all know that p* = 2%, so the inflation gap is -1%. Now, we have some leeway here with respect to the parameter A, but let's assume that the Fed responds aggressively to the inflation gap (consist with the Taylor principle) so that A=2. 
Now, if we think of the long-run real rate of interest as r* = 2%, then our Taylor rule delivers i(t) = 2%. Presently, the Fed's policy rate is i(t) = 0.25%. So, if you're OK with these calculations, the Fed should be "lifting off" (raising its policy rate) right now. Oh, and don't call it a "tightening." Instead, call it a "normalization." After all, even with i(t) = 2%, the Fed is still maintaining an accommodative stance on monetary policy because 2% is lower than the long-run target policy rate of r* + p* = 4%. 
What about the doves? Because doves like to emphasize the unemployment rate, the argument of a large negative output gap is now harder for them to make (see also here). But one could reasonably make the case that the output gap--as measured, say, by the employment rate of prime-age males--is still negative, let's say [y(t) - y*] = -1%. Let's be generous and also assume B=1. 
Now, if we continue to assume r*+p* = 4%, our dovish Taylor rule tells us that the policy rate should presently be set at  i(t) = 4% - 2% - 1% = 1%. So the recommended policy rate is lower than the hawkish case, but still significantly above 25 basis points. 
Thus, if we take the historical Taylor rule as a decent policy rule (in the sense that historically, it was associated with good outcomes), then one might say that the hawks have a stronger case than the doves. Both camps should be arguing for lift-off--the only question is how much and how fast. 
On the other hand, something does not seem quite right with the hawk view that things are presently close to normal and that the Fed should therefore normalize its policy rate. All we have to do is look around and observe all sorts of strange things happening. The real interest on U.S. treasuries is significantly negative, for example. Indeed, the nominal interest rate on some sovereigns is significantly negative. This does not look "normal" to a lot of people (including me). And so, maybe this is one way to rescue the dovish position. For example, one might claim that the real interest rate is now lower than it normally was, e.g., r* = 1%. (see this post by James Hamilton). If so, then this might be used to justify delaying liftoff.

Regardless of positions, everyone seems to assume that liftoff will occur sooner or later. But as Jim Bullard observed here in 2010, the promise of low rates off into the indefinite future may mean low rates (and deflation) forever. Few people seem to take this argument seriously except for, gosh, the predictions seems to be playing out (see Noah Smith's post here). For those who hold this position, the question of liftoff becomes more like now or never, rather than sooner or later. 
To conclude, we see that the contradictory views expressed by Evans and George might spring from something as basic as a disagreement on what constitutes the "natural" rate of interest r*. Further disagreement might be based on the appropriate measure of "potential" y* and on the appropriate size of the parameters A and B. There are also other concerns (like "financial stability") that are not captured in the Taylor rule above that might lead Fed presidents to adopt different views on policy. 
In the immortal words of Buffalo Springfield: "There's something happening here, What it is ain't exactly clear." What this something is, its root cause, and what might be done about it seems rather elusive at the moment. And I mean elusive not in the sense that nobody knows. I mean in the sense that everyone seems to have an opinion, most of which are mutually inconsistent. It makes for interesting times, at least. 


  1. Hi David,

    Thanks for this, it helps fix ideas a bit. I have a question though. As you and most others do, I find the negative yields (and ECB negative interest on excess reserves) in the Euro area perplexing. In finance, we long have modeled interest rates with models that expressly will not result in negative yields at any point, so my usual tools aren't much help to understand the current situation.

    So question: Are you aware of any models that either (a) result in negative yields as an outcome or (b) tell us what negative yields mean?

    1. Thanks, Prof J.
      Yes, I am aware of models that permit negative nominal interest rates, though I can't recall specific papers. I saw Randy Wright present such a paper last year at the Chicago Fed money and payments conference. But it is not rocket science. All one has to do is dispense with the assumption that it is costless to store money.

      Evan Soltas has a nice discussion here:

    2. Thanks David. I can see Mr. Soltas' point vis-a-vis checking accounts, but I don't think that can hold water for bond yields. People need to use checking accounts, after all - I don't know of a single legitimate business that doesn't use direct deposit, and electronic payment for goods & services is near-universal. But bonds - that's a discretionary investment. Seems like a different story to me.

    3. I think of it this way. Suppose I want to park a large amount of cash overnight in a bank account. The bank charges me 50bp (so negative nominal interest rate). Alternatively, I could purchase a safe bond (or repo). How much will the seller of that safe bond ask? He will sell it at a premium (negative nominal interest rate). And I will accept because the alternative is a minus 50bp bank account. Yes?

    4. Yes, I can see where that story makes sense.

      On another, but related, topic. I've been seeing lots of estimates of the natural rate of interest lately. These all require some measurement of the output gap, to my knowledge. But what if the real barriers to employment have increased, and so potential output is lower than it might appear from a modeling exercise? Would that not result in an incorrect estimate of the natural rate?

    5. Ah, the "natural" rate. Richard Rogerson once wrote a nice article on this that you might find useful. I blogged about it here:

    6. Thanks, David. That was educational, and I see your point. I have noticed quite a variety of definitions and assumptions about the meaning of the "natural" anything, really.

  2. Isn't this also an implicit argument against discretion?

    1. "This" referring to disagreement? Could you please elaborate? I'm not seeing the connection.

      But for what it's worth, I don't think commitment is a reasonable assumption to make. Back in 2008, for example, Bernanke made very clear that QE was just temporary. But now it seems that the Fed's large balance sheet will be with us for the foreseeable future. Not that this really matters. What matters is that the FOMC as a group (with overlapping generations of members) all appear to appreciate the need to keep inflation low and stable. It may not be a first-best policy, but it has served reasonably well, especially when compared to past policy mistakes and the crazy monetary policy regimes in other countries.

  3. OT but not way OT: Why do people say "the Fed just swapped reserves for bonds" when describing QE?

    The Fed bought $4 trillion in bonds; bank reserves rose by $2.5 trillion. Okay...only $1.5 trillion diff...

    Also, the Fed buys bonds from the 22 primary dealers. The PDs buy from clients or in the market. The sellers of bonds to the PDs get $4 trillion in cash. Does not that cash mean something?

    The reserves are created, of course, when the FedNY credits the commercial bank accounts of the PD when it buys bonds. Not from bond sellers' accounts.

    Side question: Since QE, another $500 billion in cash has been put into circulation, bringing total to $1.34 trillion. Okay, some of that is suit ace money.

    But all of it? And how often does a paper bills circulate? Not M2 velocity---a paper bill.