1. I've discussed at length why I think QE is irrelevant under the current circumstances, most recently here. If the Fed could tax reserves, that would certainly matter, as Sumner points out, but that's not permitted in the United States, and so is not a practical option. Sumner has no special claim to be rooted in the "real world" here. I'm looking at the same real world he is. However, I think I'm more willing to think about the available theory. Apparently he doesn't think that's useful. The key problem under the current circumstances is that you can't just announce an arbitrary NGDP target and hit it with wishful thinking. The Fed needs some tools, and in spite of what Ben Bernanke says, it doesn't have them.
2. Sumner is correct about the change in views of central banks that occurred post-1970s. However, Volcker needed courage to change course. He faced a lot of opposition from within the Fed and without. I'm more cynical about the way central banks adopted New Keynesian economics later on. Rightly or wrongly, this wasn't stuff that was challenging what they were doing - more like an exercise in reverse engineering. How do we write down a framework that justifies the status quo?
3. Sumner says:
...RGDP fluctuations at cyclical frequencies are assumed to be suboptimal, and are assumed to generate large welfare losses—particularly when generated by huge NGDP shocks, as in the early 1930s.My point in looking at seasonally adjusted nominal GDP was to point out that fluctuations in nominal GDP can't be intrinsically bad. I think we all recognize that seasonal variation in NGDP is something that policy need not be doing anything to eliminate. So how do we know that we want to eliminate this variation at business cycle frequencies? In contrast to what Sumner states, it is widely recognized that some of the business cycle variability in RGDP we observe is in fact not suboptimal. Most of what we spend our time discussing (or fighting about) is the nature and quantitative significance of the suboptimalities. Sumner seems to think (like old-fashioned quantity theorists), that there is a sufficient statistic for subomptimality - in this case NGDP. I don't see it.
4. Friedman rule: My conjecture for nominal interest rate smoothing I guess reflects a view that the intertemporal Friedman rule distortions could be more important than the sticky price/wage distortions. Further, it's not clear an optimal policy in a sticky price/wage model is going to imply a lot of variability in the nominal interest rate.
I’m kind of perplexed as to why Williamson calls himself a “New Monetarist.”Randy Wright talked me into it. It's as good a name as any. We like some things about Milton Friedman, but some things we don't like - 100% reserve requirement, his need to separate assets into money and not-money. We're a lot more interested in theory too.