Wrong question


One thing that has really struck me during the past several years of watching the Fed and watching others watch the Fed is how the debate within the consensus seems always to be framed so that the answer is: raise interest rates.

The debate has been over before it has begun, which is one reason the conclusion has been so unremittingly wrong the one way, at least among the talkers. The forward markets have also been a bit too hawkish, but to a much lesser degree and in a way that I have had more trouble identifying in real time. One interpretation of this is that people with money on the line have seen the same issues I have. On the other hand, and admittedly, my droning on about how forward pricing is not wrong may strike some as pointless!

Let me give three examples of bias among the talkers that will be quite familiar to anyone who has been following this blog.

  • The discussion about how stimulative QE has been (and how that might or might not have to be offset eventually by higher short-term interest rates) seems always to spot the benefit of the doubt to the QE enthusiast. I have been over this in some detail in earlier posts. See in particular this post on confirmation bias, if the subject interests you.  The gist of my complaint here is that the relevant evidence seems be chosen ex-post so that it can support the claim that QE was powerful. By far, my favorite post hoc rationalization is that QE applications drove up inflation expectations in the bond market, although — pointedly — not in the real economy. This was not part of the case originally made for QE, which was, you may recall, supposed to drive DOWN bond yields through the portfolio balance channel.  And higher inflation expectations confined to the bond market were certainly not stimulative, given that they implied upward pressure on the perceived real interest rate in the real economy.  (I return to this issue of inflation expectations below.)
  • Fed officials seem invariably to describe monetary policy as “accommodative.” This is striking because the word does not mean anything in the context in which it is used.  From the Fed leadership’s perspective this might be an advantage, because it allows for plausible deniability when people ask for evidence that policy has actually accommodated.  To wit, there is no conceivable evidence at odds with the idea that policy has been accommodative.  Policy could “accommodate” a boom. Or policy could “accommodate” collapsing inflation expectations. Accommodative policy describes everything and thus nothing, to use the well worn rhetoric. Contrast this with the Fed saying flatly that policy is stimulative or – better yet – is actually stimulatING in some unusual way. They do not typically say that because it is too easily falsified.  Instead, we have the weasel word, chosen to hint at the idea that policy is doing something unusual. (With the approach of full employment, there is a case for the Fed easing off its pursuit of strong demand growth.  The question is the funds  rate path that goes with this thought.  There is a reasonable debate around that, which should not be just assumed away, IMV.)
  • Closely related, people debate endlessly about how quickly monetary policy will have to “renormalize.” They take as given that the funds rate that has been in place for seven years now is necessarily abnormal. Maybe it is. But isn’t that supposed to be what we are actually DEBATING, rather than just assuming?

I don’t think this is just me being nitpicky about process. This approach would seem to help explain why the consensus has been wrong in the same direction now for about seven years. Ask a stupid question, that implies its own answer, etc.

I was reminded of this issue yesterday morning when I read a piece from columnist Greg Ip of the Wall Street Journal, urging us not to “panic” about low inflation breakevens in the bond market. His argument is that even long-dated forward inflation breakevens seem to be too correlated with the the oil price to be believable in some sense. He cites work by Mike Feroli at JPM documenting this, convincingly in my view.

That is an interesting observation.  Why would the expected inflation rate beginning five years from now be so tightly linked to the recent change of the oil price? That does seem weird. And good work by Greg, a friend, for pointing it out.

On the other hand, when inflation breakevens were higher, or when they were moving up with the applications of QE, people seemed much more inclined to take them at face value.  These phenomena showed that inflation expectations were “anchored” or responding favorably to QE, as the case may be.  The cottage industry of nitpicking and second guessing the trend was much less developed, which fits the theme I am pushing here.  Now that inflation expectations are low, the consensus analysis of them has become VERY sophisticated. That seems weird too, eh?

Until recently, the argument was that the low inflation breakeven did not really show lower inflation expectations. Rather, the inflation “risk premium” had fallen.  Once that became implausible, because the breakeven had fallen too far, the argument shifted to the claim that the process setting the breakeven is irrational somehow, so ignore it or downplay it.

My own view is that bond market breakevens are a small part of the puzzle. We ought not take too much comfort when they are stable in the desired range, particularly when we can imagine conditions that might knock them below that. And then when they do fall, we ought not freak out, because inflation expectation in the bond market are probably not actually self-fulfilling, unlike those the real economy. And  as Greg and Mike Feroli point out, there are technical issues in extracting reliable information  from them.  I could be wrong in my take, but one thing I would insist on: we don’t get to change our interpretation of breakevens based on whether they suit our priors or not.

As for Greg’s insistence that we not “panic,” I hope this does not sound rude, because Greg is a friend whose work (and writing ability!) I respect. But to paraphrase Keynes, he sets himself too easy an objective.   I see no evidence at all that people are panicked about low inflation breakevens in the bond market.  The doves just think, quite reasonably, that low inflation expectations in the bond market and real economy are one reason for the Fed to go gingerly – and probably not again immediately. The doves think that this is one of many issues that the consensus Fed watcher, who is too hawkish, is too quick to dismiss

To his credit, IMV, Greg actually accepts that take, but he seems to bury it out of eagerness to get to a hawkish interpretation, that the whole tightening program need not be called off.

To be sure, the fed funds futures market prices my take on this, probably not for the reasons I assert, but more just coincidentally.  So these points are more about the social psychology of Fed watching and backward looking than a call on valuation in the futures strip — or the link from that to the stock market.

I think the consensus take on the Fed in recent years has involved a mass delusion, although I realize that probably sounds cranky, and that monetary policy analysis can turn people into cranks, especially if they do it sitting alone and unemployed in their home office!

The worst example of this delusion, as I see it,  was the uncritical-bordering-on-sycophantic acceptance of the Fed leadership’s claims about how QE works. But there is actually a fairly long list of the consensus anchoring on what former Defense Secretary Rumsfeld calls “unknown unknowns.” I listed some of those unknown unknowns at the top of this note, but was not exhaustive.  Perhaps I am rather exhaustING! I definitely have a perspective here, I know.


 Here is a chart showing inflation expectations in the bond market and real economy.  For some reason WordPress seems to balk when I try to show evidence of inflation expectations, directly and without forcing the reader to hit a link.  Doubtless they are in cahoots with the Fed!  It could not possibly be that I am just technically incompetent.

The data are monthly frequency to February in both cases. Measured daily, breakevens have picked up a bit very recently, but that is not obvious from the chart.