Wednesday, December 12, 2012

Why We Shouldn't Feel Well-Guided

Today's FOMC statement was as expected on the quantitative easing (QE) side of policy. The Fed will continue to purchase $40 billion in mortgage-backed securities (MBS) per month, and will be purchasing $45 billion per month in long Treasury securities outright, rather than swapping short Treasuries for long ones.

There were some surprises (for me) in the change in forward guidance. Let's see what the statement says, so we can parse it:
the Committee decided to keep the target range for the federal funds rate at 0 to 1/4 percent and currently anticipates that this exceptionally low range for the federal funds rate will be appropriate at least as long as the unemployment rate remains above 6-1/2 percent, inflation between one and two years ahead is projected to be no more than a half percentage point above the Committee’s 2 percent longer-run goal, and longer-term inflation expectations continue to be well anchored. The Committee views these thresholds as consistent with its earlier date-based guidance. In determining how long to maintain a highly accommodative stance of monetary policy, the Committee will also consider other information, including additional measures of labor market conditions, indicators of inflation pressures and inflation expectations, and readings on financial developments. When the Committee decides to begin to remove policy accommodation, it will take a balanced approach consistent with its longer-run goals of maximum employment and inflation of 2 percent.
The first part of this, is a trigger at at a 6.5% unemployment rate. Actually, it's not a trigger, as the 6.5% unemployment rate is just a necessary condition for tightening. The second part - the inflation trigger - is pretty weird. A second necessary condition for tightening is an inflation forecast -one to two years ahead - that exceeds 2.5%. Note the following:

(i) The FOMC is going to ignore actual inflation. Apparently that's irrelevant.
(ii) Whose forecast is this? You know whose. It's the Fed's own forecast. If you're paying attention to the Fed's forecasts, you'll understand that they basically make it up so that it's consistent with their own policy.

We're also told that inflation expectations becoming unanchored would be grounds for tightening. What is that supposed to mean? Then we're told that, of course, the Fed will look at everything, just as it always does.

If the goal was to provide a more precise statement about what will trigger a tightening of policy in the future, the FOMC has failed dismally. This statement is more vague than the last one, in October, which contained a calendar date.

What about policy after liftoff? We're told that the committee"...will take a balanced approach consistent with its longer-run goals of maximum employment and inflation of 2 percent." I'm worried that this is balanced like Fox News. The last reference to "balanced approach" I saw was in a speech by Janet Yellen. The balanced approach, as far as I can tell, represents a marked change in monetary policy, toward an activist approach rooted in the belief that short-run non-neutralities of money are a very big deal. The Fed has just told us that they care a lot less about inflation. They're losing sight of what their job is.


  1. Stephen,

    It's not only a question of the size of short-run non-neutralities. It's also question of the weighting of the Fed's *three* mandates in their objective function assuming whatever kind of short-run non-neutralities. It sounds like you are permitting literally *zero* latitude in accommodating the other two mandates (inflation is *below* target, and it would be really good to get away from the ZLB). Even if you think the non-neutralities are small (obviously greatly disputed), 2.5% is *really* small too so it's a tiny trade off in favor of the other objectives. By crying wolf at the slightest sign of dovishness I think you increase the risk of getting eaten later when it actually counts. 2.5% is really, really not a cause for panic. It's better not to pretend that it is.

  2. The Fed has a lot of flexibility in how it deals with its Congressional mandate. The mandate is so vague that you could drive a truck through it. Greenspan was quite content to avoid speaking directly to the real side of the mandate, and that did not get him in trouble with Congress, though of course he had the good fortune to live in untroubled times. Bernanke now goes out of his way to speak directly to the real part of the mandate, and I think that's a bad idea. As for crying wolf, I guess we'll see. I just call it as I see it.

    1. "2.5% is really, really not a cause for panic."

      1. Who is panicking?
      2. That's 2.5% in the Fed's one-year-ahead and two-year-ahead forecasts. That tells me they are willing to tolerate a lot more in terms of actual inflation. And that's the thin edge of the wedge. Suppose that we have persistent inflation of 4% or 5% at some time in the future. There is no way that these people will want to reduce it.

    2. 4% or 5% inflation doesn't seem like a cause for panic either. If it improves the labor market, we can live with it for a while. If it fails to improve the labor market, the Fed has the option of tightening policy to reduce inflation.

  3. "We're also told that inflation expectations becoming unanchored would be grounds for tightening. What is that supposed to mean?"

    I think they are promising to look at forward forecasts and the breakeven spread on TIPS/Treasuries, and thus addressing this concern of eminent monetary economists:

    "1. All of the necessary conditions are there for a substantial inflation. I'm thinking of something on the order of 5% to 10%, and once it gets going it will be costly to stop it. The total stock of reserves will rise to about $1.6 trillion by the end of June, and that represents an accident waiting to happen. Right now, as Bernanke points out, financial markets are taking the Fed at its word. The margin between TIPS yields and nominal Treasury yields is not very large, reflecting modest anticipated inflation. But if people start to anticipate that the Fed will not be reducing the size of its balance sheet any time soon, and start to anticipate higher inflation, then nominal rates of return on assets will rise, making reserves undesirable to hold. The Fed can make reserves more desirable to hold by increasing the interest rate on reserves, thus cutting off the incipient inflation, but it will have a hard time doing that."

    Did you ever formally take on inflation-forecast targeting a la Svensson and Woodford?

    I couldn't find any posts, but seems like you really, really don't like it.

  4. "I think they are promising to look at forward forecasts and the breakeven spread on TIPS/Treasuries..."

    You think. So you're guessing. We shouldn't have to guess what they mean.

    "I couldn't find any posts, but seems like you really, really don't like it."

    No. As applied science it's ludicrous. The university wants me to teach and do research. So they base my salary increases on my teaching evaluations and what I've published. I have a better idea. I think the university should rely on my teaching evaluations and what I forecast my publications over the next two years will be.

    What Svensson and Woodford papers are you referring to?

    1. "You think. So you're guessing. We shouldn't have to guess what they mean."

      Aren't we always guessing? For instance, they never told us what Taylor Rule they were using (or if they ever were using one). Under Greenspan, wasn't even hinting at the rate of inflation the Fed saw as consistent with price stability considered really radical transparency?

      "What Svensson and Woodford papers are you referring to?"

      Sorry, I put a link, but I guess formatted it wrong.

      I just wanted to point to a place where it was presumably put into a strong form. You'd know better than me the best form of the argument.

      "I think the university should rely on my teaching evaluations and what I forecast my publications over the next two years will be."

      If there was a betting market on your evals for the next semester and they tanked or soared, wouldn't that alter your best guess for your next semester evals? Wouldn't it alter the school's forecast?

      If the "consensus forecast" of your latest working paper, shot up after an NBER presentation would that change your priors about where it would end up?

      Being a grad student in finance, the idea of just throwing out TIPS just does not compute with me. I have no love for forecasters, but microstructure/liquidity concerns, hedging concerns might make TIPS an imperfect forecast and forecasters could add something. Ang et al. says surveys do better than TIPS (and I'm not happy about it).

      Anyway, I'm sure you have a well thought out argument. I'm just saying I think it would make an interesting post.

    2. "Aren't we always guessing?"

      Exactly. The basic idea is that consistency and transparency for a central bank is a good thing, but it's just too complicated to write down the central bank's policy rule, in part because you can't foresee all the future contingencies. The Fed says it is trying to get more leverage from its policy by being more explicit. But in its struggle to do that, it is not increasing the information content at all; if anything it is telling us less.

      On forecasting inflation, the monetary circumstances are so unusual now that there is a huge amount of risk in anyone's inflation forecast, including what you can infer from market prices. So risk is a problem, and its also a problem that the Fed gets to choose the inflation forecast that determines its policy actions.

  5. The Fed has made a practice (in the past) of obscurantism, even mysticism. Greenspan actually boasted that he mumbled incoherently.

    This is a step in the right direction.

    We are a democracy. As citizens, ee are entitled to accountability and transparency from our most important macroeconomic policy making (and executing) agency.

    The next step is to move the Fed into the Treasury Department, so that it functions like a line agency, and a new Fed chief is brought in a with a new President.

    There is not an public agency in the USA, or a set of "experts" who do not believe their speciality should be exempt from public oversight. The Fed is no different.

    An important question: Is the Fed effectively doubling its QE program with this move?

  6. "We are a democracy. As citizens, ee are entitled to accountability and transparency from our most important macroeconomic policy making (and executing) agency."

    I agree. The Fed should be held accountable for the things it can actually control, and its policies should be transparent. This change in policy is a move away from the ideal, in both respects.

  7. There are plenty of ways to keep the Fed honest in its forecast of inflation. For one, the breakevens implied by TIPS prices are a very commonly cited measure of market expectations. There are also countless surveys of forecasters (e.g. the SPF), projections by international organizations, etc. If these other measures are sending a clear message about inflation above 2.5%, yet the Fed continues to project inflation below this number, the Fed will have a heavy burden to explain itself. (Bernanke said as much in his press conference.)

    There are some models where targeting the forecast is a good idea, though I don't think that in practice it's necessarily better than targeting realized inflation. The problem with targeting realized inflation, though, is that you can get swept away by relative price changes that have nothing to do with monetary policy - it's very easy for the annualized CPI increase to vastly exceed 2.5% over a short period of time (or vastly undershoot it) if the price of oil shoots up. (In fact, since it contains a more-or-less perfectly flexible component, if measured at arbitrarily high frequencies the annualized CPI will have arbitrarily large swings -- a simple consequence of the fact that Brownian motion has infinite total variation.)

    The Fed could successfully target realized inflation if it used some measure other than headline CPI -- median, trimmed mean, core, whatever. The problem is that the media would cry foul if realized headline CPI -- but not the Fed's target measure -- went above the threshold. My guess is that the Fed pragmatically decided to target the forecast in order to sidestep these criticisms.

  8. "There are plenty of ways to keep the Fed honest in its forecast of inflation."

    Inflation forecasts are notoriously bad. The private forecasters are presumably forecasting inflation by looking at Fed behavior. I don't think there's much information in anyone's forecast, particularly right now. Why base your policy action on a forecast that is close to useless?

    "The problem with targeting realized inflation, though, is that you can get swept away by relative price changes..."

    Who is getting swept away? As you say, you look at where the price increases are coming from, and make some judgement about what's transitory and what's not. Even then, you can make a good case that you should just focus on headline inflation, with some tolerance for variability. Look through my archive. I've written on this.

    1. Actually, as Morris and Shin point out (, things can be even worse. It could be that the forecasts of the Fed results in agents ignoring their own private information in favour of the Fed's forecast. If the coordination motive is sufficiently strong, this can be welfare reducing.

  9. In watching the monetary policy and fiscal cliff debates, it seems that "balanced approach" is the new term used for do it my way.

  10. Steven, I agree with much of your post. But let me ask you, how do YOU think they should have phrased what they are trying to do? If they had said instead, "until inflation hits 2.5%" then they run into to problems:
    -Inflation rising to 2.5% due to some temporary supply shock and causing them to tighten when they shouldn't.
    -Inflation much exceeding 2.5% because it responds with a lag to monetary policy (so if you start tighten only after it hits 2.5% you will end up with a higher rate).

    So I can see no better way to frame the statement relative to what the Fed wants to do than the way it was framed, making a reference to forecasted inflation (despite the vagueness on how the forecast is made).

    Besides, I do not think this is different than what the Fed does in general. Monetary policy always responds to forecasts rather than actual levels. All they did was to raise their tolerance for inflation by half percentage point.

    Finally, I think this is an improvement over the previous forward guidance. Nobody trusted their committment on calendar years. At least now we have some idea of what they want to do, roughly speaking, accomodate until unemployment falls to 6.5% so long as inflation does not rise above 2.5%. How well they can stay within these parameters is an issue, but it is always an issue, not just now.

  11. 1. I would not have set any numerical thresholds at all. I didn't like specifying the calendar dates either. I thought that was bad policy if it was a commitment, and bad policy if it wasn't a commitment.
    2. Generally, I think there's no substitute for taking a particular action at a point in time, and then carefully articulating why you are doing it. No need to map out the future. Once everyone understands how the state of the world maps into policy actions, we've reached bliss. No point in trying to describe how you map the state of the world into actions, as that will only confuse people, as we see here.
    3. "At least now we have some idea of what they want to do..." No we don't. All we know is that they will raise the policy rate, if every, at some date after we cross the 6.5% threshold. When does QE end? When will the policy rate rise? Who knows?

    1. Hi Stephen,

      This is a very helpful comment. I read the blog, but I don't think I've really understood what you would consider good policy. This comment certainly makes it clearer.

      I think it would make a good post, if you get a chance to flesh it out a little more. For instance, what all should be included in two so we wouldn't still be guessing what the Fed was thinking, and how the Fed should be accountable in a Democracy.

  12. I hear you. But, for financial institutions trying to decide how much to charge for long term loans, having some vague idea about what the Fed will do under a broad set of possible states)might be better than having no idea. Well, if it does make a difference it should be reflected in the risk premium right? I suppose all we have to do is wait and find out.

    1. The spread between short-term and long-term bonds.

    2. The yield curve is typically upward-sloping. It seems you think that reflects a risk premium. What's the risk?

    3. The so-called interest rate risk.

    4. That is, the risk from an unexpected change in future interest rates. This risk rises with maturity because prices of long-term bonds are more sensitive to changes in the interest rate. This (and perhaps a liquidity premium) explain upward slope of the yield curve.

    5. But if you price assets in the usual way, when you incorporate all that you just get a flat yield curve, on average. No term premium.

    6. Hmm, how so? If short-term interest rates rise unexpectedly by 100 basis points the price in the secondary marker of bonds maturing in ten years will drop more than the price of those maturing in five years. Won't buyers demand a higher yield to be compensated for the additional volatility?

    7. I'm just talking about conventional asset pricing theory, where the only risk that matters is non-diversifiable aggregate risk. Working out the implications for the term structure of interest rates is a standard exercise. For example, take Mehra and Prescott's model, and price claims to consumption 1, 2, 3, etc. periods from today. You can get closed-form solutions for an example where aggregate consumption is i.i.d. You'll find that the yield curve (real yields) sometimes slopes up, sometimes down, and on average it's flat.

    8. Because it has been a while since I read Mehra and Prescott and given that things are very hectic with finals, let me make a plea to intuition.

      Suppose you want to save a given amount for 5-years. If the yield of the 5-year bond was exactly the same as the average of 20 expected yields of a sequence of 3-month Treasury bills, would you personally be indifferent between the two? In other words, would you be indifferent between buying the long-term bond vs. buying the short term and rolling it over? Would your preference change if you became more confident about your expectations? I thought that this is the the whole rationale behind the preferred habitat theory.

    9. Preferred habit relies on market segmentation. No one has written down a convincing model of it - the modern theories I have seen simply start with the assumption that people have different preferences over assets of different maturities. Not a good place to start, in my opinion. With preferred habitat, you would need to explain why private financial intermediaries can't arbitrage, for example by borrowing short and lending long.

    10. "Preferred habit relies on market segmentation."

      I think it only requires that different maturities are less than perfect substitutes in people's.

      "why private financial intermediaries can't arbitrage, for example by borrowing short and lending long."

      Fear of a bank-run perhaps? Wasn't that what brought Bear Sterns down, even before any major losses were recorded? Once you are out of business you don't have the option of waiting for the positive shock to offset the negative.

    11. "...less than perfect substitutes in people's"

      preferences? I hope that's not what you meant. Assets are valued for their payoffs and liquidity - some notion of how easy they are to trade. You don't want to model that at the level of preferences. That just assumes the result, and doesn't teach you anything.

    12. Ha ha, yes, I did omit the word preferences, but not because that's how I would model the problem; I was just using the common definition.

      What I have in mind is the following:
      Suppose one is choosing between buying a) a 3-year coupon bond or b) a sequence of 1-year discount bonds. For simplicity assume that the interest rate is the sum of a constant and a Gaussian white noise, so the optimal forecast beyond one year is equal to the constant. Suppose that the shock is realized in the beginning of each year. Finally, assume that if the person buys the 3-year bond, at the end of each year they decide whether to sell it in the secondary market and keep the price and coupon payment or keep it and buy with the coupon payment a sequence of one-year bonds for the remaining of the 3-year period. Under these assumptions the average return is the same under both choices if the person keeps the 3-year bond until maturity. However, if bonds are always fairly priced, the standard deviation of the dollar-value of the person's assets is higher under option a). Moreover, the difference rises as the standard deviation of the interest rate gets bigger. So if the consumer thinks they may need to liquidate the bond prior to maturity not only will they demand a premium for holding the longer-term bond, but the premium will rise the more uncertain the future path of interest rates is.

      Now, if I understand your argument correctly, if this results in higher yields for long term bonds then risk-neutral financial intermediaries will raise funds by selling short-term bonds to these consumers and use the proceeds to buy long-term bonds, so arbitrage will equate the two interest rates. This argument however relies on the assumption that financial intermediation is costless. Otherwise the spread should be positive to cover the cost. However, while the yield curve will be upward sloping, the spread will not depend on the standard deviation of the interest rate. What can change that?

      A) If financial intermediaries have market power. In this case a higher standard deviation of the interest rate may result in a bigger spread if intermediaries respond by charging consumers a higher "price" for providing "insurance".
      B) If the intermediaries are themselves not risk neutral. Why may they not be? Because by selling short and buying long they open themselves to the possibility of a bunk run that rises with the firm's paper losses. In this case a higher standard deviation of the interest rate raises the possibility of a bank run, which may compromise the solvency of the firm.

  13. Stephen:

    If you wish accountability, then think about moving the Fed into the Treasury Department.

    You will then, every four years, have the opportunity to vote your thoughts. It could be hoped that debates about monetary policy would become more robust, and would be participated in by a broader range of people.

    If a Fed Chief wavered, obscured, pompously pettifogged about the dangers of inflation when millions were unemployed and inflation was at historic lows, then the public could, to the best extent practical, vote that Fed Chief out.

    It may be the public would choose a Fed policy not of my choice, or your choice. The same thing happens every year on fefense, agriculture, taxes etc.

    BTW, I think the consequences of a our national defense policy, such as isolationism before WWII, and then jingoism ever since Vietnam, have been horrific. Far more damaging than Fed policy.

    It is called living in a democracy.

    I too wish for a Fed with public-broadcast FOMC meetings