RE: Utility Maximisation

[CHU@xxxxxxxx]

Sat, 4 Oct 1997 12:53:09 +0200
Per Gunnar Berglund (pgb@xxxxxxxxxxxxxxxxxx)

Arnold Chu wrote (in part):
>> A concrete example, excellent! Here's how I understand the statistical
procedure: Collect many instances of aggregate income changes for one group
of people. Plot the aggregate consumption changes against the agg income
changes. Try to find a functional form that fits the data well. <<

>> We're done. If the relationship is stable and the data good, then the
fitted curve will be useful. Why do I need to drag in a 'utility function'
and a 'maximization' procedure into this? <<

Per:
:We don't, and that is the point I have been making all the time.

Terrific.  We do agree here.

:The standard UM model will serve us _only_ as a model for measurement of
:utility -- never as a model for "explanation" (prediction) of actual
:behaviour.

Now I'm really confused.  If UM is not necessary, has no explanatory
power and has no predictive capability, then why use the concepts?

Arnold:
>> Take another example. Let say an ecologist wants to study the impact of
habitat loss on the mating frequency of a specie of birds. If she can
construct a stable statistical relationship and several
plausible causal explanations. She's happy. I doubt that she would feel
impelled to say that the birds are 'maximizing their utility'. So why do
some economists feel impelled to do that to people? <<

Per:
:The ecologist is probably not very interested in assessing the well-being of
:the birds, but economists should definitely be interested in assessing the
:well-being of humans. Insofar as they are not, they don't deserve to be
:called "economists".

Here you seem to be saying that Utility or 'Desire' reflects the well-being
of humans.  If you combine this with a 'revealed preference' stance,
that means humans always behave to maximally enhance their well-being.
This seems to doom us to a Panglossian conclusion.
If you assert that the well-being of everyone can be further enhanced
if say the resources are allocated a differen way, then you are
using the inferred Utility/Desire to predict what would happen under
a different situation.  If the UM formalism has no predictive power,
your assertion would not be supported.

I happen to think that there may be situation where th UM formalism does
have predictive power, but that there are even more situaions where
it does not.  Basically I concur with Bruce that it seems like an
unnecessary restriction to modeling self-interested or desire expressive
behaviors.  I have even given examples of other formalisms.
Perhaps it would be more productive to discuss what might be important
attributes for such formal systems.

Regards,

Arnold Chu


--
*******************************************************************************
Bear Stearns is not responsible for any recommendation, solicitation, offer or
agreement or any information about any transaction, customer account or account
activity contained in this communication.
*******************************************************************************