1 commento (espandi tutti)

Re: I fu PIGS

crav 24/4/2010 - 08:26

mi permetto di citare:

In the old days people talked of "extrapolative expectations" arguing that - when forecasting the future in the absence of an understanding of the structural model driving the system - we look at trends in past data and extrapolate those trends over the relevant horizon. This happens, though, only when we have become convinced that they are, indeed, permanent trends and not just small and irrelevant blips. When evidence suggests that the trend has changed or reverted back to old patterns, we accept it only after a while but, once accepted, we tend to extrapolate it into the indefinite future. The problem, obviously, is how long is the "while" and how reasonable it is to assume that people extrapolate trends that cannot be sustained forever, such as the one we just noted to exist in the capital share of corporate value added during the last two decades. This is a hard question for which we do not have a good answer and that the learning literature has really never addressed. We will try, nevertheless, to make some practical progresses along these lines.

e raccomandare questo bellissimo paper.