Woodford on Optimal Monetary Policy Rules

I have supported more aggressive action from the Fed throughout the crisis, though perhaps with a bit less confidence that such action would have big effects than some others, and I have supported nominal GDP targeting. It's a way for the the Fed to be more aggressive without running into political opposition. After all, the Fed is simply stabilizing the dollar value GDP, i.e. stabilizing people's incomes, and the real/nominal distinction will be lost on most people.

But I haven't jumped fully onto the NGDP bandwagon (and I get more hate mail over that fact than any other issue, by some margin). The reason is simple. The theoretical underpinnings of this particular rule are not clear. I have asked supporters to answer a question, more than once, "in what class or classes of models is NGDP targeting the optimal policy rule?," but there was no response. It's as though they think this particular targeting scheme is optimal in every conceivable model, with every conceivable transmission mechanism, etc., but that can't be true -- that's obvious. So where are the limits? When is this the best policy rule?

Fortunately, in an interview that anyone interested in monetary policy should read, Michael Woodford notes that he has looked at this question, and guess what? It turns out that NGDP has desirable properties, and sometimes it is the optimal policy, but not always. What's important, as anyone familiar with Woodford's book and work realizes, is the inclusion of both growth and level variables in the policy rule (which to include, output growth, output levels, price growth -- we call this inflation -- or price levels depends critically on the nature of the underlying friction in the model, there is no set rule, which leads one to ask about robust rules across various models, and here I think NGDP targeting might do well, but someone needs to check...). This is Woodford, in an interview with Dylan Mathews, describing some of his work in this area:

... I didn’t specifically prefer to use that formalism of a nominal GDP target, which is why I’m not particularly associated with that specific proposal. What I’ve been writing about for quite a while, and there’s a lot of discussion of this in my 2003 book Interest and Prices and a lot of papers as well, is the desirability of committing to rules where there’s a nominal level variable rather than purely referring to the rate of growth of a nominal variable. The idea was that if the nominal growth in the economy undershoots or overshoots, either one, there should be a commitment to getting back to the target path, so we’ll target the rate of growth going forward. The idea that purely forward looking approaches are undesirable is something I’ve been emphasizing in various papers since the 1990s, and I’ve criticized various proposals that didn’t have that property.

I emphasized then that this was important if you got in a position where you couldn’t lower nominal interest rates below a certain rate. ... One of the issues ... was that you were more likely to have this problem of hitting that lower limit, and if you did hit it, you would particularly want to commit to have higher inflation for a while to get back to this nominal level target. I wrote a paper (pdf) in 2003 for Brookings that was specifically about that mechanism.

We proposed what we called an “output gap adjusted price level target”. The idea was to talk about a price level, as opposed to the inflation rate, but a corrected price level target where you add to it some multiple of the real output gap. In various models you could show there were ideal properties to this kind of proposal. But in only relatively special circumstances would that coincide with nominal GDP. I thought there was a strong case for having the nominal level variable, but not just a price index, to also take into account the level of real economic activity. But it’s not just real economic activity. It has to be corrected for potential growth. There are a lot of things to discuss about the ideal level of the target variable, but I was mostly arguing for the desirability of schemes that involved a level variable. ...

Some people seemed to think Woodford was saying in his Jackson Hole paper that  NGDP targeting is optimal, but he wasn't. Sometimes it is, but more generally he was saying that it's a step in the right direction, but there are often better ways to conduct policy. However, it may be the only politically feasible proposal:

the most practical version of such a proposal that you can imagine the Fed adopting is an NGDP target. That is a compromise relative to the theoretical ideal. It’s worth asking what you could imagine the Fed actually going for, which is not going to be what, in an ideal world where everyone understood economics perfectly, you would want to do.

He also talked about other rules that are a step in the right direction such as the one proposed by Charles Evans, president of the Chicago Fed. Evans has been advocating a policy rule that links the target interest rate to the state of the economy (e.g. buy a certain quantity of bonds each week until inflation or unemployment crosses a predetermined threshold). After reading Woodford's paper, I pushed the Evans proposal over NGDP targeting (which brought some very critical responses from the blogosphere, oh well, all I'm asking is to ground proposal in theory) because I thought it was (1) also a step in the right direction, as Woodford notes, and (2) politically feasible in ways NGDP targeting is not, at least not yet. And in fact the Fed did adopt a loose version of this policy at its last meeting (where the thresholds are left vague instead of being specified in advance).

The Fed has not performed as well as I'd hoped (which is not to say it has performed poorly, only that there is room for improvement), and I am very much in favor of investigating alternative monetary policy rules. NGDP targeting -- when things like model uncertainty and robustness to model specification are considered -- may be the optimal rule, or at least among the rules with very desirable properties. My view is that the Fed needs to be nudged in this direction in, to use Fed terminology, "measured steps," so I will continue to push at the margin rather than for grand reworking of existing policy approaches. But please understand we are all basically on the same side of this question, and working toward the same goals.