Tuesday, December 11, 2012

Fed Update

The FOMC is meeting today and tomorrow. What is on its collective mind? There are two issues which are likely to be on the agenda relating, respectively, to the two legs of the Fed's current unconventional policy actions: quantitative easing and forward guidance.

Quantitative Easing
The Fed's "operation twist" will end at the end of this month. Recall that this asset swap program began in September 2011, and was extended in June of this year. The program involves sales of Treasury securities with remaining maturities of 3 years and less in exchange for Treasury securities with remaining maturities 6 years or more. Those swaps have been proceeding at a rate of $45 billion per month. Even if the Fed wanted to continue that program, it would not be feasible, as the short-term Treasury securities on the Fed's balance sheet are all but depleted.

Since September, the Fed has been purchasing $40 billion in mortgage-backed securities (MBS) per month - the "QE3" program. This is also an asset swap, but in this case a swap of reserves for MBS. QE3 is an open-ended program, which will continue under conditions stated, for example, in the last FOMC statement:
If the outlook for the labor market does not improve substantially, the Committee will continue its purchases of agency mortgage-backed securities, undertake additional asset purchases, and employ its other policy tools as appropriate until such improvement is achieved in a context of price stability.
My best guess is that the committee will not think of the new information received since the last meeting as indicating "substantial" improvement in the labor market, and they will be looking to keep their policy stance the "same" as it was at the last meeting. But they can't do that, as it's impossible to continue the twist asset swap. What's the next best thing? Well, if you believe that QE works to actually ease something, as the FOMC certainly does, then you should also think that there is little difference between swapping short Treasuries for long Treasuries and swapping reserves for long Treasuries. What you should expect to see is a QE4 program involving purchases of $40 billion in MBS per month and $45 billion in long Treasuries per month. In terms of total purchases, that's a little larger than QE2, which involved purchases of about $75 billion per month (all Treasuries). The Fed could of course buy more than $40 billion in MBS per month, but that would signal a change, and they're not likely to do it (I'm not sure about the feasibility either - how big is that market?).

Some concerns:
1)The Treasury and the Fed are clearly thinking about debt management in completely different ways. As for example James Hamilton and David Beckworth have pointed out, the Treasury has been systematically increasing the average maturity of the outstanding government debt in the hands of the public, while the Fed is systematically reducing it. The Treasury might be thinking that it can save a huge amount on debt service in the future by, for example, locking in a 30-year borrowing rate of 2.84%. The Treasury seems to think it is looking after us by lengthening the average maturity of government debt, lowering borrowing costs, and presumably lowering our future tax burden. But the Fed thinks that we get more real economic activity (temporarily, permanently?) if the average maturity of government debt is lower. The Fed also thinks it is looking out for us. Maybe Ben Bernanke should walk down the street and try to sort this out with Tim Geithner (or his successor).
2)Short of a theory of QE - or more generally a serious theory of the term structure of interest rates - no one has a clue what the effects are, if any. Until someone suggests something better, the best guess is that QE is irrelevant. Any effects you think you are seeing are either coming from somewhere else, or have to do with what QE signals for the future policy rate. The good news is that, if it's irrelevant, it doesn't do any harm. But if the FOMC thinks it works when it doesn't, that could be a problem, in that negative QE does not tighten, just as positive QE does not ease.

Forward Guidance
This is where the big change in policy is likely to occur. In public statements, various Fed presidents have been honing a policy rule that involves quantitative triggers. Until now, the FOMC's forward guidance statements have included a calendar date for "liftoff" - the date at which the Fed's policy rate (the interest rate on reserves currently, given the large stock of reserves outstanding) rises above 0.25%. The last FOMC statement says that date is "likely" to be mid-2015.

After living with calendar dates in the forward guidance language since August 2011, FOMC members now appear to think they are a bad idea. Why? The Fed generally likes the idea of forward guidance, as it is another tool the Fed thinks it can use when it is up against the zero lower bound. Support for the idea comes from New Keynesians - Woodford et al. - and New Keynesian models. But Woodford is on record as thinking that a calendar date is a bad idea. One may think that extending the liftoff date will be more accommodative, as this increases anticipated inflation and lowers the real rate of interest, but extending the liftoff date also conveys pessimism.

The triggers for liftoff typically take the following form. The policy rate should stay at 0.25% until one of two things happen: (i) the inflation rate rises above x%; (ii) the unemployment rate falls below y%. Most of the public debate currently seems to be over what x and y should be. x is typically in the range 2.5 to 3.0, and y is typically 5.5 to 7.0. The argument for triggers is that a calendar date can lead to policy errors, or negates the intent of the policy. If the Fed commits to the calendar date, it risks waiting too long to tighten, or it tightens too soon. But if the Fed appears willing to move the calendar date in response to new information, the forward guidance becomes meaningless. With triggers, the FOMC can state the policy once, commit to it, and move forward.

Here are the problems with triggers:
(1)To be well-understood, the triggers need to be specified in a very simple form. As such it seems as likely that the Fed will make a policy error if it commits to a trigger as if it commits to a calendar date. The unemployment rate seems as good a variable as any to capture what is going on in the real economy, but as such it's pretty bad. It's hardly a sufficient statistic for everything the Fed should be concerned with.
(2)This is a bad precedent to set, for two reasons. First, the Fed should not be setting numerical targets for anything related to the real side of the dual mandate. As is well-known, the effect of monetary policy on real economic activity is transient, and the transmission process poorly understood. It would be foolish to pretend that we know what the level of aggregate economic activity should be, or that the Fed knows how to get there. Second, once you convince people that triggers are a good idea in this "unusual" circumstance, those same people will wonder what makes other circumstances "normal." Why not just write down a Taylor rule for the Fed, and send the FOMC home? Again, our knowledge of how the economy works, and what future contingencies await us, is so bad that it seems optimal, at least to me, that the Fed make it up as it goes along.

My overriding concern is that the Fed's unconventional policy moves - one on top of the other - are digging a deep hole that it will find it difficult to get out of. Of course, Ben Bernanke seems likely to leave at the end of his term in about a year's time, so it won't be his problem.


  1. Stephen,

    Great summary. A few points:

    Theory and evidence points towards QE being basically useless. But I wouldn't overstate the worry that if the Fed makes the mistake of believing in it, they might take too long to control inflation. For one, if QE is useless, it can be undone at essentially any rate they choose. Also, so long as they are paying IOR, there is nothing stopping them from using short rate policy to tighten even before undoing the balance sheet. They can sell it off later, whenever they want.

    "the Fed should not be setting numerical targets for anything related to the real side of the dual mandate"

    Sure, but it's whichever comes first: 2.5% inflation *or* (not *and*) 7% unemployment. It's not *that* crazy to tolerate higher than target inflation if unemployment, for example, is at 10%. No, we don't know where the LRAS lies, but we don't *think* it's there. Assuming divine coincidence fails, you don't have a choice but to have some kind of real mental target. Also, the gains from credible forward guidance are potentially very big, so it doesn't make sense to refuse to take any risk whatsoever in order to try it. I don't see how you can do that without articulating your mental real target.

    So by all means, get rid of the QE nonsense. But steps towards determining more fully the contingent path of the policy rate are a positive development. 2.5% is pretty conservative. Really, it wouldn't be that disastrous to hit 3 or 4%. The cost of getting back to 2% is pretty small compared to this perpetual output gap. Even 1981 wasn't this painful (yes I remember it), and required disinflation was a lot bigger then.

  2. "Also, so long as they are paying IOR, there is nothing stopping them from using short rate policy to tighten even before undoing the balance sheet."

    There's nothing to stop them, but I don't think they will do it.

    "The cost of getting back to 2% is pretty small compared to this perpetual output gap."

    I think that's the key problem with what you're saying. (i) I think the Fed is thinking that cost is large. (ii) How do you know it's a perpetual output gap, and not an efficient, but lower, growth path?

    1. Another thought: Read James Tobin's paper, "How Dead is Keynes?" referenced in my previous post. He was writing in June 1977. The unemployment rate is 7.2%, the cpi inflation rate is 6.7%, and he's complaining because he thinks the unemployment rate is disastrously high. He wants more accommodation. Today, I think we understand the reasons that the unemployment rate was high at the time, and we certainly don't think that monetary policy was too tight in mid-1977, particularly as inflation was about to take off into the double-digit range. Today, I don't think the labor market conditions we are looking at are the result of sticky price/wage inefficiencies, or any other problem that monetary policy can correct.

  3. What about duration of the Fed's balance sheet? If the Fed wants to keep duration where it currently is, then $45 billion wouldn't be the correct number for the new monthly purchases of long-maturity Treasury securities.


    1. No, they can buy $45 billion a month and adjust the maturity structure of what they buy to hold duration constant, if that is what they want to do.

  4. Stephen,

    Liked the Tobin paper. Very funny. The key evidence that he was wrong, as you point out, was not the level of inflation, but rather the subsequent reflation.

    I agree with you that the Fed seems to think the cost of above target inflation is high. I'd argue the following:

    1) 3-4% is really no big deal either empirically or in any model I'm aware of
    2) Because of 1) we can afford to test the output gap by seeing if we get a bit of reflation, and especially wage reflation. (That's how I'll know it the output gap is gone.) We did not have that luxury in '77.
    3) After five years of voodoo monetary policy that somehow resulted in a relatively stable inflation rate, I fear the Fed is beginning to engage in magical thinking about their powers. *Don't touch inflation*, or you'll jinx it and after 5 years of moving it magically with our minds we have no confidence that we have the power to move it with the actual instruments at our disposal. If the people lose *confidence* the magic will stop working!

    Much better, in my opinion, that inflation moves around *a bit* and they get back to their knitting (spot short rate policy).

    1. "3-4% is really no big deal either empirically or in any model I'm aware of"

      There's certainly nothing magical about 2%. There is no science that tells us this is optimal. We've been in the neighborhood of 2% for a long time, and that seems inoffensive, so we stick with it. That said, if you start thinking "3 or 4 isn't so bad," then when the inflation rate rises to 3 or 4, you start thinking 5 or 6 isn't so bad. Soon you have a problem on your hands. I think we understand how that works.

  5. Suppose we converge into a regime in which the fed just keeps thinking it need to keep interest rates for a very long time, practically the indefinite future. Then the only equilibrium with finite inflation I see is one where by the Fisher relation inflation is actually very low, in fact we get deflation on average? Kind of like Japan. Then we can start arguing again about whether or not monetary policy can change outcomes (long run monetary neutrality says it wouldn't matter anymore in response to how current shocks have played, but I can bet for sure some people will see permanent output gaps to cure).

    1. It's possible to have a zero nominal interest rate forever with positive inflation. I can write down models where that happens.

    2. You mean an economy where the long run interest rate is negative? Without uncertainty, that would mean we can stop worrying about the intertemporal budget constraint of the government since it can run a ponzi scheme. Sure we can write such a model, but I thought there was a a quasi consensus outside some post keynesian circles that we weren't living in that sort of world. Sure in a world with uncertainty you can observe long periods with an average negative real interest rate, but they come with a tail risk that the no ponzi game condition hits you in the end and the government does have to end up running surpluses (Papers by Bohn and a paper by Blanchard and Weil come to mind as good references on this). Maybe the model from Kocherlakota on bursting bubbles has this property though I can't remember. Even among models with financing frictions and incomplete markets, I think most will end up delivering positive long run risk free real interest rates (e.g Bewley/Aiyagari models and their extensions to collateral constraints tied to durables or capital, or the popular models based on Kiyotaki).

  6. With the Cleveland Fed index of inflationary expectations hitting record lows, and with global sovereign debt interest rates trending towards zero bound---are we really worried about inflation?

    Has no one ever thought about Japan?

    I would be more comfortable if the Fed adopted a 2 percent floor on inflation, not a ceiling.

    Better yet, explicitly targeted 8 percent NGDP growth for eight straight quarters.

    Why zero bound is like economic quicksand is an interesting question, and I don't know the answer. Perhaps sheer dithering and timidity on the part of central bankers who do not like QE. It is monetizing debt after all, one of the huge taboos of economics.

    Although in the USA we have monetized $2 trillion and inflation is at record lows. Japan tried QE 2001-6, and see the horrible inflation they have. Not.

    But getting an economy up after zero bound seems to be like telling Grandma to get up off the ice after falling down. Mixing metaphors I know.

    Can the economics profession adjust? Is it telling horror stories about inflation when the real threat is Japanitis?

    The Pentagon is still set up to fight the Cold War.

    The Fed?

  7. "With the Cleveland Fed index of inflationary expectations hitting record lows..."

    That index has never made any sense to me.

  8. Steve, I'm impressed. You absolutely nailed it here!