Thursday, September 13, 2012

The Fed Plays With Its Tools

Well I got most of this right, except for the change in forward guidance. Today the FOMC announced that the Fed will be purchasing more mortgage-backed securities, and expects to maintain the policy rate where it is for a longer period than previously announced.

In its press release, the FOMC first tells us what it is worried about:
The Committee is concerned that, without further policy accommodation, economic growth might not be strong enough to generate sustained improvement in labor market conditions. Furthermore, strains in global financial markets continue to pose significant downside risks to the economic outlook. The Committee also anticipates that inflation over the medium term likely would run at or below its 2 percent objective.
Here, both parts of the dual mandate are addressed. The Committee thinks they are missing on the low side on real economic activity. From their point of view, while things are improving, they are not improving quickly enough. What exactly would an appropriate rate of improvement be? The statement makes that clear - economic growth has to be proceeding at a sufficiently high rate so the there is "sustained improvement in labor market conditions." I think the right interpretation of that is that the Committee wants to see the unemployment rate decreasing and the employment/population ratio increasing.

The second element in the dual mandate is inflation, of course. Note that the FOMC does not speak to the current or past rate of inflation, but to what they anticipate it will be. In some economic models, the central bank should want to control the anticipated rate of inflation, and current and past inflation are bygones, but there are pitfalls in practice. The key problem is that the Fed gets to write its own performance review if it defines success in terms of what it forecasts. In this case, the Fed's forecasts are in this document, and the inflation component of those forecasts is very optimistic. The pce inflation rate is forecast to be lower than 2% out to the end of 2015. I'm not sure what generated that projection, other than wishful thinking. I can't see how it's consistent with the forward guidance in the current FOMC statement, but more on that later.

In the paragraph in the FOMC statement that details "QE3," the latest Fed asset-purchase program, the Committee first tells us what they think QE does:
To support a stronger economic recovery and to help ensure that inflation, over time, is at the rate most consistent with its dual mandate,...
Thus, according to the FOMC, quantitative easing both increases inflation and real economic activity - the Committee thinks more QE will move us up the Phillips curve. The Phillips curve, if you hadn't noticed, seems to be at the core of current Fed religion.

The announced QE program is to be added on top of two previously-announced asset purchase policies. The first of these replaces maturing agency securities and mortgage-backed securities (MBS) that run off (because of prepayments and defaults) with more MBS. In the absence of other asset-purchase programs, that policy would hold constant the size of the Fed's balance sheet (in nominal terms) - i.e. the balance sheet would otherwise be shrinking, though at a fairly slow rate. The second policy is so-called "operation twist," which had been extended to the end of the year, involved swaps of short-maturity Treasury securities for long-maturity Treasuries, at the rate of about $45 billion per month. The newly announced program calls for purchases of MBS at the rate of $40 billion per month.

Given past Fed actions, this move is somewhat aggressive. An operation twist purchase of long-maturity Treasuries is roughly the same as purchasing the same dollar amount in MBS. The fact that the Fed is issuing reserves in the latter case rather than selling short-maturity Treasuries should not make much difference, and it should not matter much if the Fed purchases MBS, rather than Treasury bonds (though I know there are claims to the contrary - see this and this). So, roughly, what the Fed plans is $85 billion in asset purchases per month (as it points out in the statement), which is a little larger than QE2, which proceeded at the rate of about $75 billion per month.*

An interesting feature of QE3 is that it is open-ended. Assuming one thinks that QE is a good idea under the current circumstances (and I don't think it matters at all), this also seems like a good idea. In cases where Fed decisions concern only a target for the overnight rate, it is deemed useful for the FOMC to reconsider what it is doing at each FOMC meeting. Why not do that for asset purchases as well?

Of course, it's important that the Fed adopt a reasonable stopping rule for QE3. What is it?
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.
Consistent with typical central-banker vagueness, that leaves the FOMC a lot of wiggle room, but I take it that they want to see the unemployment rate falling and the employment/population ratio rising in line with what we might see in a "normal" recovery.

The Fed also threw some forward guidance into the mix, for good measure:
the Committee also decided today to keep the target range for the federal funds rate at 0 to 1/4 percent and currently anticipates that exceptionally low levels for the federal funds rate are likely to be warranted at least through mid-2015.
The previous wording said "late 2014." At this point, I think the forward guidance is essentially meaningless. This is not a promise - it's just a forecast - and no one should think the Fed is actually confident about that forecast.

We should also be asking why the change in forward guidance did not go the other way. If the Fed is so confident about the effectiveness of QE, it should be shortening the period until "liftoff," not lengthening it.

*Corrected from an earlier version.


  1. I think you may have misread the size of the asset purchases. It looks to me like the total asset purchases (including operation twist) are $85 billion, with $40 billion of that being the new MBS purchases.

  2. If QE is itself irrelevant, and forward guidance essentially meaningless, what caused the huge jump in equity and commodity markets after the 12:30 announcement?

    1. Suppose: (a)QE-type asset swaps don't matter; and (b) the type of forward guidance messages the Fed is sending do not matter. Then, it can still be the case that this type of announcement moves current asset prices. Why? The Fed thinks it matters, so what it does today can matter for its decisions about future actions that actually do matter. For example, the fact that the Fed thinks QE matters may make it do QE in reverse in the future when it wants to tighten, rather than raising the interest rate on reserves when it wants to tighten. Thus, the misguided Fed delays an action that actually matters, and that matters for future asset prices and therefore, for today's asset prices.

    2. Interesting theory. If the Fed thinks that QE matters but the market thinks that the Fed is being irrational about this, what would justify this large gap in beliefs? Are you implying some degree of market segmentation or a preferred habitat story?

      It would be a strange world indeed in which the more irrational the Fed appears to the outside world, the more effective its stated policy.

    3. Stephen:

      I think this is a very interesting line of thought, but I don't follow your answer. Why would the Fed following a misguided path that will lead it to mess up in the future increase equity prices today? I can see that it may lead to expectations of higher inflation and therefore to higher commodity prices today. But the interaction of inflation and the US tax system are usually thought to be bad for equity prices... Maybe you can expand on your ideas in a longer post?

    4. I'm trying to give you a story that's consistent with everyone being rational. People in the Fed might even think that the policy doesn't work, and still be willing to do it. Of course, it could be that both the Fed and the market participants are deluded. That's hard to think about, let alone to model. Maybe they can learn the hard way, though.

    5. For anonymous 6:41: Maybe the effect on stock prices is just higher expected future inflation. Those stocks are claims to future nominal dividends of course. That's entirely consistent with my story.

  3. $40 billion in new purchases, $25 billion in re-invested MBS principal, and $20 billion in Twist. So, balance sheet to grow with $40 billion a month.


    1. Exactly, the balance sheet grows by $40 billion per month. But it's not the Fed's balance sheet that matters, it's the composition of the outstanding debt of the consolidated Fed/Treasury. In terms of that, Twist and QE3 matter in the same way.

  4. They have no religion, only smokescreens. This is just a continuation of the stealth bailout of zombie banks.

  5. "Assuming one thinks that QE is a good idea under the current circumstances (and I don't think it matters at all)"

    Stephen, here's a big question I have for you: Note the "at all". You use this kind of absolute language all the time about QE. Now, as you well know, I did my big Wallace 1981 AER project. I understand perfectly that in Wallace's model, with Wallace's assumptions, QE does nothing "at all", perfect zero (at least for asset prices and consumption). But since I know the assumptions required to get this result are far from 100% true in the real world, I am not sure at all that the conclusions, of perfectly zero effect, would also be 100% true. How can you be so sure the real world works 100% like models like Wallace's, and the deviations from the assumptions will have absolutely zero effect on making QE have an effect – because I think they would (and I'll soon post on the reasoning why)?

    Joseph Gagnon, a Stanford trained monetary economist, has a brief post on how the deviations from the assumptions result in QE being effective (and I came to the same conclusions in studying Wallace). A great post, if I might suggest, would be for you to tell us why Gagnon is wrong. His post is at:

    1. Even if Gagnon is right that there are exceptions to Wallace's assumptions, he may not know what effect, if any, QE would have.

      For instance, let's say Gagnon is right in claiming taxpayers cannot gauge the contingent tax liability created by QE duration bets. This condition may hold for a time, and then reach a "tipping point" at some level of Fed balance sheet growth. At that point, taxpayers might receive a constant stream of information about that risk, to the point where they irrationally overestimate their liability. At that point, QE would have the detrimental effect of causing taxpayers to pull back on spending.

      Or say that Gagnon is right about preferred habitat, but that post-QE those investors choose to buy commodities instead of stocks. Forcing up commodities prices might reduce middle-income households' real income growth expectations, forcing them to pull back on spending.

      When one leaves the simplified world of models, the alternative is not foresight and precision; its a murky soup of complexity. People like Gagnon are dangerous because they assign such unwarranted certainty to their predicted outcomes.

  6. No one, including Gagnon, has written down a structural model that captures the observed features of the term structure of interest rates and in which QE works as the Fed says it does. Observing some correlations in the data does not count for anything, and that's especially tricky when it comes to asset price movements and Fed behavior.

  7. Correlations are often grossly underestimated, and misunderstood as worthless. It depends. What's the apriori, or related info. If 1,000 people drink from a well and all 1,000 die minutes after, is the smart thing to say, that's only a correlation; it means nothing without a chemical analysis of the water – gimmie a drink!

    If the correlation is very strong, and there's no other explanation that fits that correlation – and apriori and related information in the world – anywhere near as well as the straightforward one, then that's strong evidence.

    I was just taught the straightforward explanation for the term structure of interest rates – longer term (expected real) rates are higher than the product of forecasted shorter term rates because of the risk that your forecast is wrong (or really wrong). You need a pretty big risk premium to lock in an interest rate for 30 years (at least I do), because inflation could end up being a lot higher than you expected, and that could be devastating – There's not perfect forecasting in the real world.

  8. "I was just taught the straightforward explanation for the term structure of interest rates..."

    That "explanation," or any other - e.g. expectations theory, standard Lucas-type asset pricing - captured in a serious model, does not explain the term structure. The models don't fit.