Sunday, June 30, 2013

The Fed's Communication Problem

Since early May, real and nominal long-term bond yields have risen in the United States. The most stark depiction of this is in the following chart, which shows the 10-year TIPS yield - which has risen by roughly 100 basis points since early May - and the breakeven rate (nominal 10-year yield minus the TIPS yield) - which has fallen by about 50 basis points over the same period.
I think it's fair to say that this was not the Fed's intent. The Fed thinks that "accommodation" is what is appropriate, and the way it sees that working is through low real bond yields and high anticipated inflation. But apparently real bond yields have risen, and anticipated inflation has fallen. Further, I think it's also fair to say that the bond price movements since early May have been driven primarily by the interpretation by financial market participants of public statements by Fed officials - principally Ben Bernanke.

On Thursday, Narayana Kocheralakota was interviewed on CNBC, and tried to make sense of this. Narayana thinks that the key problem was that the markets were (are) misinterpreting statements about QE (quantitative easing) as statements about the future path of the policy rate. That's in the right ballpark, but doesn't quite get at the essence of the problem. While the Fed took some pains to to make public statements about how QE was just normal policy (ease by moving long rates down rather than short rates), they have consistently segmented QE from forward guidance (statements about the future path of the fed funds rate), particularly in the FOMC statement. QE and forward guidance are typically described by the Fed as two different tools - like a hammer and a saw. But it should be clear to anyone - and I think it is - that these two elements of policy are somehow related.

But how are QE and forward guidance related? The Fed tells us that purchases of long-maturity government bonds and mortgage-backed securities by the Fed work to reduce long bond yields, and that this will increase some components of aggregate expenditure. Sounds just like standard Fed talk, right? When things are too hot - in the sense of inflation being "too high," and real economic activity being "too high," then the Fed cools things down by "tightening," i.e. the Fed intervenes in an attempt to increase nominal market interest rates. And the reverse if things are "too cold."

So what could the problem be? As economists, we know that things are actually more complicated than that. We disagree about what "too hot" and "too cold" mean, whether the Fed can actually heat and cool in particular circumstances, and whether the heating/cooling/thermostat analogy makes any sense. Sometimes it makes me cringe, and medical analogies are even worse ("the patient is bleeding," "the appendix needs to come out," etc.) But, in terms of what can be understood by the average person on the street, or even by the average bond trader, this may be the best the Fed can do in terms of communication. There are two directions: up and down. And the Fed can communicate whether they are going up or down, and the likelihood of going up and down in the future.

So, the Fed seems to have communicated QE in its usual up/down hot/cold language, but the message isn't getting across. Why?

1. QE is an experiment. As with all experiments, ideas about how to run the experiment have changed as the experiment proceeds. Sometimes the Fed swaps reserves for long Treasuries (QE2 and QE3), sometimes it swaps short Treasuries for long Treasuries (operation twist), and sometimes it swaps reserves for mortgage-backed securities. Why has the Fed done this in different ways? Does it now think that one type of intervention is preferable to another, or that there were different circumstances along the path we have followed since the financial crisis that warrant different approaches? None of that is clear from Fed communications. The only explanation we have is that these are different tools, and that when you have a lot of tools and you're in a predicament, you should use them all. Maybe there's little difference among the effects (if any) of these different tools. If so, the Fed is needlessly confusing us.

2. QE2 and operation twist were announced as asset purchases of specific assets at a specific rate, for a specific period of time (with some provision to change the plans in unusual circumstances). QE3 started that way, but then changed to a contingent plan (move the rate of purchases up or down depending on new information). What's confusing about this is that we have no idea where some of the numbers are coming from. Why does the Fed think that $85 billion per month in asset purchases is the appropriate number to move long bond yields by the amount the Fed wants to move them? If the Fed can't tell us, we have to be suspicious that they don't know. Why doesn't the Fed just announce a target for, say, the 10-year nominal Treasury yield? The fact that they do not makes us suspicious that the Fed thinks it may not be able to move Treasury yields in the way it confidently tells us it can. So, if the Fed is confident on the surface, but we're suspicious that it is actually mired in ignorance and doubt, how are we to think about what the FOMC will be doing at the next meeting, or next year?

3. The Fed has taken pains to be more specific over time about when the date of "liftoff" will occur - the date at which the policy rate (the interest rate on reserves under current conditions) increases above 0.25%. Liftoff will occur after the unemployment rate passes the 6.5% threshold. But, until recently, the Fed was not specific about "tapering" in asset purchases. Whenever the forward guidance language changed, it was clear that this was intended as a change in policy (up/down; heating/cooling) toward more accommodation. So, when Ben Bernanke gives more information about how the tapering will occur, how else should anyone interpret that but as "tightening," even though that was clearly not the Fed's intention?

One last point. Most Fed officials who speak in public argue that, even if we don't understand how QE works, that the empirical evidence demonstrates that it does. That empirical evidence is based on announcement effects - the Fed says something, and asset prices move. For example, the Fed announces some upcoming asset purchases, and bond yields move down. I hope that recent movements in asset prices will call that logic into question. In this case bond yields have moved up in response to something the Fed said when, in terms of how the Fed thinks about policy, either nothing has happened on the policy front, or the news should be interpreted as more accommodation rather than less.


  1. "Why doesn't the Fed just announce a target for, say, the 10-year nominal Treasury yield? The fact that they do not makes us suspicious that the Fed thinks it may not be able to move Treasury yields in the way it confidently tells us it can."

    If the Fed announced a target for the 10 year note yield -- ie. a fixed rate at which they'd purchase the 10 year -- wouldn't the price of the 10 year immediately jump to the Fed's price? QE might be irrelevant, but an all-out peg is on a whole different plane.

  2. "If the Fed announced a target for the 10 year note yield -- ie. a fixed rate at which they'd purchase the 10 year -- wouldn't the price of the 10 year immediately jump to the Fed's price?"

    If the Fed announced a target for the S&P 500 index of 1800, would the index immediately jump to 1800?

    1. I'd hazard a guess of yes, but you'd probably say no because reserves are no longer special and the private sector can undo the operations.

      Ok, but what stumps me then is how did FDR manage to push the gold price up in 1933 and how did the Fed peg bond rates during WWII?

    2. First, I have no doubt that, in normal times (without excess reserves outstanding overnight) that open market operations by the Fed can move all asset prices - short rates, long rates, stock prices, etc. The question is whether the central bank can actually peg anything other than the overnight risk-free nominal interest rate. It is certainly possible to peg an exchange rate, as we know, under some conditions. It is also possible to peg the price of outside money in terms of a precious metal, as under the gold standard, again under some conditions. So, does the same thing apply to long-maturity government debt? What if the Fed promised to buy and sell 10-year Treasuries at a fixed price? Could it peg the price of 10-year Treasuries to anything it wanted? Could it do that for all maturities and fix the whole nominal yield curve? If that's possible, why doesn't the Fed do it?