Stephen Williamson has an excellent observation about the Fed’s new commitment to hold rates near zero for 2 years:
Further, it seems the outcome the Fed would hope for is one where inflation increases, the interest rate on reserves increases commensurately, and the Fed proceeds to sell off assets so as to normalize the state of its balance sheet, with zero excess reserves. Committing to IROR = 0.25 for two years risks two outcomes that seem equally bad (if we believe that 2% inflation is optimal). One is the too-high-inflation outcome: People anticipate high inflation, reserves start to look much less desirable, and the high inflation is self-fulfilling. The other is too-low-inflation: People anticipate low inflation, the reserves look more desirable, and low inflation is self-fulfilling. The first scenario is something that I have been worried about. The second scenario was a concern of Jim Bullard (pdf), and Narayana Kocherlakota.I think both are possibilities, i.e. there are multiple equilibria.
Fed officials like to talk about “anchoring expectations.” In this circumstance, the kind of FOMC statement that would anchor expectations would be something like: “We anticipate raising the fed funds rate target (actually the IROR target, but what the heck) as observed and anticipated inflation warrants. Currently, we think we are on a path on which inflation will increase.”
George Selgin made some similar criticisms in the comment section of this post and this post (where Nick Rowe also makes comments.). I completely accept their analysis. Interest rate pegging is a really bad idea. But I am not particularly worried that the high inflation outcome will occur, for two reasons:
- Consider Stephen’s comment, “Currently, we think we are on a path on which inflation will increase.” You would think that is the view of the Fed. That would certainly be their view if Lars Svensson (of targeting the forecast fame) were running the place. But alas, Svensson is in Sweden, the land of negative IOR (and the only western European country expected to achieve rapid growth during 2009-12.) The Fed probably expects core inflation to slow a bit over the next year. Why not adjust one of its policy levers? Why not more QE, or lower IOR? Your guess is as good as mine.
- The Fed still has plenty of anti-inflation credibility. And thus it’s widely understood that if inflation exceeded the Fed’s comfort zone, it would slam on the brakes in mid-2013. But what about the period before mid-2013? It turns out that many New Keynesian models suggest that current NGDP is strongly affected by future expected NGDP. Although I’m not exactly a NK economist, my intuition tells me they are correct on this point.
Consider the following analogy. Suppose the Fed said it would let the dollar/euro exchange rate float for two years, and then peg it at 1.40. Also suppose this announcement was credible. In practice, the spot rate would show very little variation over the next two years, even though the Fed was taking no explicit actions to stabilize rates (until mid-2013.) The same intuition applies to NGDP.
Keynes understood this at an intuitive level, which is why he always talked about the importance of business “confidence.” Whenever reading Keynes, just substitute the term “expected NGDP growth” for “confidence” and the meaning will be much clearer. Of course other factors such as regulation can also affect business confidence, but Keynes focused on AD.
So although I agree with the criticism of Williamson and Selgin, as a practical matter I focus on the fact that the Fed’s action is almost certainly going to be too little, not too much. (And obviously the 3 year T-note market agrees.) I hope I’m wrong, as I’d much rather see the Fed trying to rein in 4% inflation in 2013, than trying to boost up 0% inflation in 2013
Selgin and Williamson are also correct that what the Fed really needs to do is commit to a nominal target, and let interest rates move as necessary to hit that target. Because it failed to do so, a two-year interest rate commitment is likely to be meaningless.