Tuesday, August 7, 2012

Models, reality, and bankers (conservative central banker/time inconsistency stuff again)

This line from Ken Rogoff (quoted by Mark Thoma) caught my eye:
"It is nonsense to argue that central banks are impotent and completely unable to raise inflation expectations, no matter how hard they try. In the extreme, governments can appoint central bank leaders who have a long-standing record of stating a tolerance for moderate inflation – an exact parallel to the idea of appointing “conservative” central bankers as a means of combating high inflation."
At first blush, this looks like a goofy statement. The conservative central banker model isn't about the fact that politicians hire central bankers, after all. The model is about why it would be good to hire a banker whose objective function is more conservative than the public's. You can't just flip the whole formula and cite the conservative central banker model!

But I don't think that's quite right, and I think Rogoff's statement here is astute.

Models are always abstractions from reality that illustrate a mechanism that's important for explaining what we observe in the world. So what is the mechanism that the conservative central banking model highlights?

Time inconsistency. We've been over this ground recently.

But this is a broad mechanism that applies in lots of situations. So let's get more specific. The conservative central banking model is at its heart not a model about central banking. The guy just maximizes an objective function. Any old consumer could do that. Usually these models completely abstract from the policy instrument as well. So it's clearly not about central banking. What it's about is one story of how the macroeconomy works: the expectations augmented Phillip's Curve (specifically how it reacts to inflationary monetary policy).

There was a time when that mechanism was worth talking a lot about. But is that how we should think about monetary policy and output today?

Paul Krugman has another story about inflation expectations and the macroeconomy that he put out in the 1990s when Japan was struggling, and it said that Japan needed to generate credible inflation expectations.

If - as I assert - the really important part of the conservative central banking model was the expectations augmented Phillip's Curve part, and not the "conservative central banker" part, then Rogoff's point makes a lot more sense. One could just as easily imagine a time inconsistency problem where normal behavior will generate less inflation than is optimal, precisely because of the formation of inflation expectations in periods of low inflation. In that situation, by the same logic of the conservative central banker model, we need a liberal central banker and a seemingly-too-inflationary policy rule.

This is also unsettling from the perspective of rules (particularly constitutional ones, as Steve Horwitz has been discussing). If the result of the conservative central banking model is only as good as its underlying macroeconomic mechanism (the expectations augmented Phillip's Curve), then a rule for one economic environment may be entirely inappropriate for another environment (say, when "low inflation traps" are the order of the day).

I don't know where this gets us exactly, because again we're just talking in terms of competing abstractions. But I suspect it means that policy making and constitution writing aren't nearly as simple (pick a rule and constrain policymakers by it) as it's often purported to be.


  1. turnkey rentalSlowly, but surely, the foreclosure crisis seems to be abating rental properties .

  2. If we are using an expectations augmented Phillip's Curve, then having a recognizable "liberal" central banker would be disastrous. For inflation to increase AD, it must be unanticipated. If the public expects it, we just get DWL. Do you mean that when inflation is out of control, we need a "liberal" central banker so that the market form appropriate expectation?

    1. It would be disastrous under normal circumstances. My point is that it's an open question whether a standard expectations augmented Phillip's curve where only unanticipated inflation contributes to AD really makes sense right now. There's a good case that it doesn't.

      In that case, you really need a "liberal central banker".

      Or better yet, a central banker that knows when to be liberal and when to be conservative while always being more or less predictable and credible.

      Not an easy task, of course.


All anonymous comments will be deleted. Consistent pseudonyms are fine.