To respond to Ellen Frank's latest posting (I'm glad someone is
still interested in the AD/AS brouhaha!), I would say first that
what is needed is a graphical framework in which P-dot is on the
vertical axis and not Y but Y/Y* (or U) is on the horizontal axis;
the latter approach sets the stage for the Phillips Curve.
Why then employ the AD/AS framework at all? In my view, it is
simply in order to build up from the basic Keynesian model to
get to the Phillips Curve. I think the AS curve is actually
a useful construct [I see more controversy coming here!] to
distinguish circumstances in which a demand stimulus leads mainly
to an increase in Y and those in which it leads mainly to an
increase in P; I would draw the AS curve with considerable
curvature, close to a backward L. The analysis could be carried
out plausibly with a perfectly vertical AD curve, but for reasons
that I have expounded on ad nauseam I believe that the AD curve
should be drawn with a bit of a downward-to-the-right slope.
One can then move from the AD/AS framework to the Phillips curve
framework by examining what happens when the AD/AS equilibrium is
kept to the left or to the right of Y* ("potential output") for
an extended period of time -- building in adaptive inflationary
expectations.
Cheers, <Tom Weisskopf>
