« Unavoidable failure | Main | Is conservatism dead? »

November 08, 2013


Szczepan Stachura

What I think the chart is showing is adaptive inflation expectations vs. basis of its forming. It’s only my opinion: but don’t know, why you consider inflation expectations as a measure of rational expectations, and not adaptive expectations.


Over my head here, but wouldn't the rational person say "I have no idea what inflation will be in the future"? What that does for models I don't know. I'm guessing rules of thumb, but I have no idea. (IIRC, "same as today" is a pretty good method of weather forecasting.)

Ralph Musgrave

Wiki in defining rational expectations says “Rational expectations theory defines this kind of expectations as being identical to the best guess of the future (the optimal forecast) that uses all available information.”

The idea that households and firms use “all available information” is plain bonkers. Only academics, more concerned with getting their models published than with solving the problems of the real world could make that sort of assumption.

I.e. Chris is right to ask: “What’s the empirical evidence here?”

Lars Syll takes Wren-Lewis to task here:


Luis Enrique

I like your ordinary/extraordinary times distinction. I wonder if models of costly information acquisition, or Sim's rational inattention models, imply that people use adaptive expectations during normal times, but during periods in which the newspapers etc. are loudly proclaiming "things are changing!" start to do something that could be reasonably modelled as Rat Ex.

I find these discussions often frustrating. One only has to be told what Rat Ex means to know that it is unrealistic, so, for example, Lars Syll's response to Simon - repeating at greater length that Rat Ex is unrealistic - gets us nowhere. The question is whether - or as you remind us, when - Rat Ex is one of those unrealistic things it makes sense to employ in a model, on a least-bad basis.

Robert Waldmann adds to your attention to the empirical evidence. Here.

I'm sure one could quibble, but let's say he's right. Does that mean that if we use adaptive expectations, in some form that accords with the empirical evidence, those models will be less wrong than Rat Ex models? My guess is that it depends on what the model if for, and if you put adaptive expectations in a model built to ask the question how should we conduct monetary policy, you will get an answer that essentially says we should exploit the fact people only have adaptive expectations and dupe them, which may turn out to be unrealistic if, as you suggest, just at the time you need monetary policy that turns out not to be how people behave.

[p.s. I also like Robert's defence of Rat Ex in the comments to his own blog. I think that carries a lot of weight in academia]

The comments to this entry are closed.

blogs I like

Blog powered by Typepad