I didn't expect that

Potentially a rather important post from Paul Krugman, documenting one of the nasty little secrets of central banking - the great "credibility" doesn't actually matter that much, and that people's expectations are for the most part conditioned by what they see in front of them.

I would actually take it a bit further and say that the time is ripe for someone to do a reassessment of "myopic expectations", which was the main modelling approach in use for forecasting purposes before the Rational Expectations Revolution (and its notoriously awful forecasting models).  In general, we have established that people do not form their expectations based on the correct solution of the correct economic model.  Thanks guys. 

But before everyone reaches for their copy of that Kahneman book, I don't think that the way forward is to start making laundry lists of results from the behavioural psychology literature and "incorporating" them in some way that nobody has ever really said what it means in a macroeconomic context.  There are two big problems with behavioural economics; first, the robustness of their results really degrades quite startlingly quickly once you get out of the lab.  And second, there are so damn many of these "anomalies" and "biases" that the models pile up degrees of freedom until they can fit anything (and therefore fit nothing).  And the only way to prune down the degrees of freedom is to start ruling things out ad hoc, which is pretty transparent model-mining.

On the other hand, I think that there are very good reasons to have another look at myopic expectations on their own merits.  The argument I'd make is that it in a surprisingly large proportion of cases, it doesn't matter so much what you actually, truly in your heart of hearts expect; it matters what you are willing or able to act on economically.  In a world of institutions, this is often constrained by what you can present objective evidence for, which is what you see around you.  Or to put it another way, you can't borrow money based on your expected salary (in most cases).

I've had decent success with myopic-expectations models over the years - particularly the One True Model Of House Prices (which assumes that housebuyers don't form any expectations at all; they just take the current price of housing and credit as given and gear up to the maximum extent possible).  I think they deserve another run out, and that central bankers hung up on their "credibility" ought to be much more embarrassed than they are about the nearly total absense of this "credibility" from discussions on wage and price setting.