A Little More on the Fed

Scott Sumner helps me to see that he and I are (even) closer on monetary policy than I'd thought. He agrees with a point I made here, that it would be difficult to adopt a level target for NGDP all at once and right now.

Crook...points out that level targeting of NGDP might lack credibility in the current environment. I think he's partly right, but I don't see it as a problem, because he's actually considering a hypothetical that is both unlikely to occur, and too expansionary from even my perspective. Recall that in 2009 I favored going back to the old pre-2008 NGDP trend line, but more recently have advocated going just a third of the way back. The problem here is that while NGDPLT is completely automatic when fully implemented, discretion is required when deciding where to set the initial trend line. Last year Christy Romer did a NYT column calling for a return to the old trend line. Because we are roughly 10% below that line, this would entail quite rapid NGDP growth (and inflation) over the next few years. Because a lot of time has gone by, and because many wage contracts now reflect the lower NGDP trend line, I no longer favor going all the way back, but rather looking for a pragmatic compromise. I think Crook's right that 100% level targeting right now (Romer's plan) would currently lack credibility, there simply aren't enough economists inside or outside the Fed that favor that degree of stimulus. [My emphases.]

But I would warn readers not to become overly pessimistic about credibility on the basis of this thought experiment... If we can get the consensus of economists to shift to NGDPLT (and so far we market monetarists have had much more success than I expected) then a future Fed may adopt NGDPLT, and it will be highly credible. If economists as a class think level targeting is the right way to go, then a period of "catch-up" will not seem "irresponsible." Even better, level targeting keeps NGDP from wandering so far from trend in the first place.

I agree with nearly all of that. (I hesitate at "economists as a class". The day when "economists as a class" agree about anything will never come.)

I'd say the points to stress are, first, adopting a level target when the gap between the target and where you are is huge is extremely difficult. You have to move to the new regime by degrees. Second, even after you've arrived, there'll often be doubts about whether the earlier path is still correct in the aftermath of a big disturbance. So the central bank can't ever wholly bind itself. Though Michael Woodford would say (too pessimistically) that retaining discretion defeats the main purpose of NGDP targeting, there's no choice but to retain some, at least: That's a political fact of life. On the other hand, third, the difference between level targeting and NGDP-growth targeting matters far less if you're choosing at a position of full employment. Either approach would then keep NGDP "from wandering so far from trend in the first place," which is the whole point.

Moving on, what to make of Narayana Kocherlakota's conversion--if that's what it was? Kocherlakota, previously seen as an inflation hawk among Fed governors, last week advocated a policy-rate lift-off rule that many commentators thought very dovish: Keep interest rates at zero until unemployment falls to 5.5 percent or inflation rises to 2.25 percent. In this he was echoing Charles Evans (and he drew attention to the fact). Evans suggests a lift-off rule of 7 percent unemployment and 3 percent inflation. Question: Who's the more cautious? I'd just written a column supporting QE3 but regretting the Fed's inability to communicate its intentions more clearly, despite (or because of) Bernanke's supposed commitment to greater transparency. I thought the reaction to Kocherlakota's speech proved my point. Commentators were all over the place in judging what it really amounted to.

The correct answer, I think, is rather subtle. Under present conditions a ceiling on inflation of 2.25 percent looks hawkish. From an NGDP point of view, it might cap growth in demand at much too slow a pace. It's certainly hawkish on Kocherlakota's previous view that structural unemployment has increased during the recession, because this implies that inflation would rise, and hence trigger higher rates, long before unemployment fell to 5.5 percent. But Kocherlakota says he's changed his mind about that. He says new evidence has persuaded him that unemployment can fall to 5.5 percent without pushing inflation up. So he expects his lift-off rule to keep rates at zero for an extended period. (Makes you wonder, by the way, if the paper on the labor market that Edward Lazear and James Spletzer did for the Jackson Hole conference, arguing that US unemployment is cyclical not structural, made a bigger difference to policy than the Woodford paper that attracted all the attention.)

So is Kocherlakota now a dove? It's mostly semantics, once you understand his position--but I'd say no, even though his assessment has switched from "rates will need to rise soon" to "rates won't need to rise soon". Put it this way: He has a hawkish rule and a dovish new forecast. As he himself just proved, however, forecasts are easier to change than rules. At the first sign of rising inflation, Kocherlakota wants rates to go up. He says he doesn't expect that to happen for a long while, but if it does he'll strike. In my book, he's a hawk.

All this underscores what I see as the biggest advantage of the NGDP approach. It invites--indeed it instructs--the central bank to remain agnostic about the real short-term effects of increased aggregate demand. Over the course of the next year, an extra 5 percent of nominal income might be 5 percent inflation and no growth, or 5 percent growth in output and no inflation, or something in between. The central bank can't know or influence how the increase in demand will split between the two. So it shouldn't concern itself with that question. It should simply ask whether demand is increasing at its target rate, a rate consistent with satisfactory growth and low inflation.

That's why I think Kocherlakota--whether you call him a hawk or a dove--is reasoning incorrectly. You don't need to know whether the cyclical component of unemployment is 1 percent, 2 percent, 3 percent or 4 percent to know that demand is currently growing too slowly. And if you know that demand is growing too slowly, you maintain or increase the monetary stimulus.