Simon in [2141] replied:
---------------------
You say first lx and vm (value of money), then py. I am no longer
sure this is right. In part my view is structured by the demands of
empirical investigation:
1. First py.
2. Then s/v from e = s/v = P/W
3. Then vlp from e=(1-vlp)/vlp
4. Then vm from vlp/vm=w
5. Then lx from py=lx/vm.
I imagine I will be accused of confusing measurement with determination. But I am interested in constructing a time trend for vm
and using it to help me think about it theoretically. That's about the
best I can do at the present.
Chai-on:
--------
As far as the measurement alone matters, your procedure from (1) to
(5) has no problem. But you have to explain theoretically why s/v
is capable of being derived from P/W, why lx is capable of being
derived from py=lx/vm, and so on. Becuase we know in theory that
lx/vm determines py, we derive lx from py=lx/vm. Because we know
in theory s/v determines P/W, we derive s/v from P/W.
I have no disagreement with your procedure in the above. But I am
strongly unhappy when you said in the above "You say first lx and
vm (value of money), then py. I am no longer sure this is right.". If
you are not sure theoretically, then you can no longer measure the
lx from py=lx/vm. Without theory, you have no right to do so.
In solidarity,
Chai-on Lee