A couple of days ago I wrote a post suggesting that Scott Sumner may have had less influence on Thursday’s Fed announcement than many in the blogosphere think.1 Roughly speaking, I had two reasons in the back of my mind for saying this:
- Most of the commentary contained a built-in assumption that the Fed’s action was sort of a mini version of NGDP targeting, which Sumner has been crusading for. But it’s really not. It’s an attempt to use future guidance to affect market expectations. That’s a component of NGDP targeting, but nothing more. If I spend years crusading for more energy efficient houses and you then decide to replace your windows, that doesn’t mean I influenced you just because new windows happen to be a component of my energy efficient housing plan. You probably knew about the benefits of UV-coated windows already. (And that goes double if you’re a world expert in energy efficient housing.)
- A lot of bloggers learned about NGDP targeting (and the expectations channel) from Scott Sumner. I certainly did. But it’s a leap to think that just because we were influenced by Sumner, this means that Ben Bernanke and the Fed’s economists were influenced by Sumner. Maybe they were, maybe they weren’t. I don’t know, and neither does anyone else in the blogosphere. But Michael Woodford, who is indisputably influential on this subject, says point blank that Sumner’s blog didn’t affect him. That’s not surprising: like Ben Bernanke, he’s been studying this stuff all his life. He already knew it.
In my original blog post, I admitted that I felt sort of churlish for pointing this stuff out. But Robert Waldmann, who would probably not object to being labeled a bit churlish now and again, writes that there may be an even bigger problem here. Regardless of who influenced the Fed to try out expectations management, it doesn’t seem to have worked:
We have new relevant evidence on the effectiveness of QEIII as forward guidance about future short term rates. Medium term rates are expected average short term rates plus a risk premium. If there is more certainty that short term rates will be very very low, both terms should decline.
[From a later post]: From Wednesday close (before the announcement) until Friday close (latest data) the 5 year rate changed 0.00%….Now I admit that my approach of focusing on changes over 48 hours and then declaring the matter settled is not the same as the approach of Michael Woodford, the leading advocate of the focus on the expectations channel. In his huge paper, he analysed transaction by transaction data and only looked at the day of the announcement….To claim that the question is still open, however, would require those who stress the importance of the expectations channel to argue that a statistical method is perfectly fine when it gives them results of the sign they like (although always of trivial magnitude) but invalid when it gives the answer they don’t like.
To be as rude as possible, if they don’t admit they were wrong (and I was right about the potential on Wednesday for further future guidance) their reasoning is like the argument that defence spending creates jobs but ARRA spending doesn’t.
By Woodford’s rules, the debate is over and he lost.
Basically, Waldmann is saying that if the expectations channel works, medium-term and long-term interest rates should have declined immediately after the Fed announcement on Thursday. But they didn’t. The forward guidance enthusiasts were wrong.
Personally, I’d be a little more charitable. One of the problems with the Fed’s announcement was that it was extremely weak. Yes, it tried to provide future guidance by promising open-ended action, but (a) the action was fairly modest, (b) the Fed didn’t really make any concrete promises about how long it would continue its QE program, and (c) there’s an election coming up and there’s really no telling how long Ben Bernanke will be in charge of such decisions anyway. It’s hardly any wonder that the bond markets had a muted response.
Of course, this is a fundamental problem with expectations management: it only works (even in theory) if investors are absolutely convinced that the Fed (a) can do what it says and (b) will do what it says. That’s pretty hard to pull off, and it’s one of the reasons that the blogosphere should probably be a little more skeptical than it is on the topic of both NGDP targeting and expectations management in general. For more on this subject, Mark Thoma has some sensible comments here.
1The backstory behind all this is long and complicated, and if you have no idea what I’m talking about here, I apologize. But it’s just too convoluted to explain in a sentence or two. On the bright side, your life will be no poorer if you just skip the whole thing.
UPDATE: Sorry to add even more to this, but Matt Yglesias emailed me a comment that suggests I should clear something up. Michael Woodford thinks that using the expectations channel should immediately reduce long-term interest rates. Market monetarists like Scott Sumner think it should raise long-term interest rates because it raises expectations of future economic growth. I started this post with a comment about Sumner’s influence on the Fed, which might have made it sound like Robert Waldmann was responding to Sumner. He wasn’t. He was responding to Woodford.
So who’s right? I think it’s hard to tell. Waldmann uses the Wednesday closing rate vs. the Friday close for 5-year T-notes. By that measure, nothing changed. Sumner uses the 10-year T-note, which was up about 6%. If, instead of the Wednesday close, you use the rate just before the Fed announcement as the starting point, both rates are up.
So either (a) the market monetarists are right, (b) nobody is right because rates didn’t change, or (c) there’s too much noise to tell. Take your pick.