Evan Soltas
Sep 27, 2013

A New Strategy for the Fed

I've been putting a great amount of time and thought into a big project on monetary policy -- about which, more details in a few weeks -- but let me share this one part of it as an essay. What has interested me is the Fed's apparent difficulty communicating the way it anticipates to tighten monetary policy over the next five years conditional on economic performance and outlook.


To be fair, this isn't at all an easy task. A lot goes into it. The Fed has to think about the likely range of economic scenarios. It has to think about the timing of the start of the exit -- that is, tapering its monthly bond purchases. It has to think about the pace of the exit -- that is, do they hike every other meeting in 2015 and 2016, as they did in 2004 and 2005? It has to think about the endpoint -- that is, where does the federal funds rate end up in a world of full employment and unicorns? 

Above all else, it has to think about how it conditions the start, pace, and end of the monetary exit on economic data. For example, if the monthly increase in nonfarm payrolls slows to 50,000 one FOMC meeting into the taper, what do they do? There are several questions wrapped up in this question. How aggressively do they respond to clear changes in economic performance? How quickly do they respond when changes in economic performance are, as always, obscured in the short run by noise? What indicators do they use, implicitly and explicitly, to condition the path of monetary policy?

The Fed has responded with what is, all things considered, a reasonably effective method of forward guidance. The centerpiece of it is the Evans' rule: The federal funds rate is to remain at extraordinarily low levels at least until the Fed's preferred measure of inflation (headline year-over-year change in the personal consumption expenditures price index) surpasses 2.5 percent or the standard unemployment rate falls below 6.5 percent. The Fed has emphasized that these are thresholds, not triggers -- that is, they're not under any obligation to raise rates when unemployment falls, and they are only disallowed from earlier rate increases.

On top of the Evans' rule, you have its forward guidance on tapering bond purchases. There was a 7.0-percent unemployment threshold for the end of tapering, but I do not think this commitment is still valid. It is a tighter policy than I believe the Fed desires and, in any case, also out of line with the "no-obligation, only-disallowance" framework the Fed has established for increases in the policy rate.

There are some salient benefits to this forward-guidance framework. It sends a clear signal about the binding constraints on Fed policy, and in particular on the start of tightening. It explicitly conditions the path of policy on economic performance. But as the last few months have shown, there are also shortcomings. The one in the news concerns the start, pace, and conditionality of tapering, on which the Fed has never crafted clear communications. That matters in itself, but most of its value is tied up as a signal for the start, pace, and conditionality of rate hikes. 

I have been surprised by the movement in the forecasts of start of rate hikes, which I would have thought would be tied down by the Fed's commitment. I have not been surprised, however, by the uncertainty in the forecasted pace of rate hikes once they have begun and the uncertainty surrounding the Fed's reaction function as it gets economic data. The Fed simply has not said much about either.

Though I should take that back somewhat: I have been encouraged by the work Fed vice chair Janet Yellen and Chicago Fed president Charles Evans have done over the last year or so. More about it in a moment, but let me first tell you why. It works from the same intellectual framework as Lars Svensson and Michael Woodford. That framework is easy to explain: Svensson and Woodford both think of the central bank as an agent in the economy who tries to minimize a loss function. The equivalent way to think about this is that the central bank has a set of objectives, and it uses its set of policy instruments to steer its set of target variables closest to the objective values.

What's usually in this loss function? Nothing that would surprise you: It's the sum of squares of deviations in the inflation rate from the inflation target, of the deviations in the unemployment rate from a natural rate of unemployment, and of the meeting-to-meeting change in the policy rate, all weighted with some coefficients and discounted to present value. (If you're wondering why changes in the policy rate are in the loss function, large changes in interest rates impose costs.)

There is an important intertemporal aspect to this. The stance of monetary policy in one period affects the optimal stance of monetary policy in the successive periods. And this framework leads you eventually to what's called "optimal control theory." That is, it's false that optimal monetary policy is the result of optimizing policy for each individual period. Rather, what you have to optimize is the path of monetary policy over many periods, which forces you into the calculus of variations.

This is, very encouragingly, how both Yellen and Evans have been thinking about monetary policy in recent speeches and papers. See this slide from Yellen's big speech last June:


That green line comes right out of the optimal-control analysis, whereas the Taylor rule comes out of single-period optimization. See the difference? They clarify the Fed's exit strategy in a way that the Evans' rule can't.

And then see this slide from Evans' talk today:

This stuff is straight Woodford and Svensson, applied to the Fed's own mandate and macroeconomic model FRB/US. This is not at all how the Fed has communicated in the past, but it is how the two academic economists have long wanted it to. They suggest making the loss function explicit, and it's a different approach from nominal GDP targeting -- where the single-term loss function would be the discounted squared sum of deviations in NGDP growth, naturally -- but the implied path of monetary policy is basically the same.

So I'm thrilled that the intellectual leaders of the Fed's communication policy -- Yellen, who runs the FOMC's subcommittee on it, and Evans, who came up with the rule -- have realized that loss functions and optimal control theory really are how you ought to think about monetary policy. And the sooner this framework begins influencing forward-guidance language, the better.

How would it do that, exactly? Here is where I come in with my recommendations. What the Fed has tried to do with forward guidance over the last year is plot the possible paths of the policy rate conditional on economic scenarios, but the way they've been doing it is kludgey when you think about it. We've been talking about asset purchases and signaling future policy -- and signaling is imperfect! -- so much that we've forgotten that there is a way to do this more directly.

My recommendation is that the Fed should target federal funds rate futures, eurodollar futures, overnight indexed swap (OIS) rates, or an appropriate proxy for the expectation of future interbank lending rates. (Whatever the specific contract form, I'll call them fed funds futures going forward.) The reason why it ought to do this: There's nothing clearer than just talking about the actual course of policy directly.

Here's how this would work. The FOMC concludes its next full meeting in mid-December. In the summary of economic projections, the Fed should report what fed funds futures prices it believes to be warranted given the midpoint of the central-tendency projection over some five-year curve. It should also report the fed funds futures prices warranted conditional upon upward or downward deviations in economic data from the midpoint of the central-tendency projection. In other words: If we get ugly or nice surprises, how does the Fed plan on changing plans?

My sense is that would be enough to send fed funds futures wherever the Fed wanted them. Why? Because the Fed sets the target federal funds rate, so they have effective control over the futures price already. Banks would arbitrage this market heavily if the prices didn't converge. Note that they wouldn't necessarily go to wherever the Fed's central-tendency projection said they should, as the market has its own prior on the likelihood of different economic conditions.

But let's suppose that fed funds futures prices shrug off the Fed. How should the Fed respond to that? It should buy or sell fed funds futures until it pushes the prices where it wants them. That will transmit itself into other interest rates just as the targeting of the federal funds rate does. Another reason why it should do this is because the value of Fed assets would rise if the market doubted its commitment but then the Fed came through.

Suppose, for example, that the Fed thinks fed funds futures contracts for 2016 are undervalued -- as it suggests is pretty much the case right now -- that is, they imply a sharper increase in the federal funds rate than the Fed desires. If, after the Fed states its fed funds futures targets, the market says "that's just, like, your opinion, man" and implies the Fed will renege on its easy-money promise by tightening early, then the Fed should buy as much in fed funds futures as it takes. If the Fed follows through as promised, then it pockets the capital gains and passes them to the U.S. Treasury. If it reneges, it pays the market a de-facto fine.

Let's emphasize, though, that the main policy change is simply that the Fed talk explicitly about the appropriate pricing of its futures contracts.

Another benefit to targeting fed funds futures is that the Fed's current strategy -- buying up long-term Treasury debt -- is suboptimal. In particular, it has inspecific objectives, isn't really in the Fed's control, and faces a severe knowledge problem. Let's take each of those objections one-by-one. Inspecific objectives: Has the Fed told anyone how far it wanted to depress long-term interest rates? What it wanted the yield curve to look like? No, it hasn't. Out of control: The Fed sets monetary policy, not fiscal policy. Other factors also affect the yield on long-term Treasury debt beyond the supply and demand for money. Knowledge problem: Does the Fed believe it knows what the appropriate term premium is? What about the sovereign-risk premium? I'm not hearing answers. It really doesn't have to make any of these assumptions when it comes to the federal funds rate, and the equivalent security to target future policy isn't long-dated Treasuries, it's fed funds futures.

Keep in mind that this strategy is a far more conservative method of forward guidance than prior asset-purchase methods. Conservative is not to say ineffectual, but rather more consistent with the policy regime. It's not that big an intellectual leap to say that if the Fed is targeting an interbank lending rate at the present, it ought to target the same variable's expected future values. Lowering expectations of future fed funds rates would also depress longer-term lending rates, given that long rates can be thought of as the sum of short rates over the loan period.

The Fed is abuzz with communications strategies. Governor Jeremy Stein had what I thought was an interesting but ultimately undesirable recommendation to taper by some fixed amount with every 0.1-percentage-point drop in the unemployment rate. The unemployment rate's measurement error is larger than that, and you don't want to respond to noise, and structurally the unemployment rate is giving a much healthier picture of labor markets than any other measure.

I've put a lot of thought into these issues. Where I end up is that trying to explicitly tie tapering and rate hikes on incremental changes in economic data isn't a good idea. The difference between unemployment in your model and unemployment as measured in statistical reality becomes too significant. Where I end up, also, is that the strategy of buying Treasuries and mortgage-backed securities strategy is pretty imperfect. Where I end up is that there are good and bad ways to be specific in your monetary commitment and forward guidance, but there's no good way to be vague or to not commit. My sense is that targeting fed funds futures would be most consistent with conventional monetary policy -- in fact, it would fit well within the normal regime -- and would be the best way to make the conditional monetary commitments necessary to to ensure a healthy economic recovery.