In reply to Jerry:
I would start from the idea that money is supposed to represent labor
time, so that in some sense the world money markets have (on average?) to
equalize labor times in commodities. But it is clear that there is a
whole lot of short term variation in exchange rates and real exchange
rates which probably doesn't reflect changes in underlying real labor
inputs. I suppose the problem is both theoretically and empirically to
specify the mediating factors that produce the short term "noise" in the
system, and to find ways to control for that noise in making the
measurements.
As a general comment, I think methodological and theoretical discussions
are much more productive when they can be related to specific problems.
Without the discipline of a specific research problem, it is too easy for
people to talk past each other (one thinking of one case, or level of
abstraction, and another thinking of another, and thus missing the chance
for a productive dialogue.)
Duncan