Re: Utility Maximisation

["Bruce R. McFarling" <ecbm@xxxxxxxxxxxxxxxxxxx>]

On Fri, 03 Oct 1997 15:15:45 -0400, "Alan G. Isaac"
<aisaac@xxxxxxxxxxxx> wrote:

>We are closer together than I realized when
>we started chatting.

>Our primary disagreement on u max seems to be
>that I claim it is useful to bundle the formal apparatus
>with substantive characterizations of plausible goals in
>order to work out the implications. By useful I mean
>it helps me understand and predict outcomes.
>(Of course for understanding and prediction it is also
>helpful to know many other things about people.)
>For example, formal analysis sheds light on
>agency problems. You say u max models are not
>explanatory, but if we learn about the world by
>exploring the implications of u max coupled with
>plausible substantive goals, in what sense is it
>not explanatory? Further, you worry about observables
>but as soon as we introduce substantive goals we
>begin to introduce observables. (The goals aren't
>observable, but behavioral implications are.)

	Under a systems definition of goals, system
targets are observables if system behavior can be
observed in a sufficient mix of conditions; system
goals requiring information from within the system.
But the systems we are talking about are people and
organisations of people.  In both cases, some goal
information will be observable.  In the latter case,
interview techniques may be necessary, although it
sometimes seems that some economists get stuck at
the "you can't take an interview response at face
value" and don't take the time to learn what you
can do to gain useful information from an interview
process.  Or perhaps in providing as one of the
apologies for utility maximisation a sequence of
questions with heavy framing bias and unrecognised
order dependent, learning how to recognise and
correct for framing biases and order dependencies
gives rise to too much cognitive dissonance.
	My objection to the utility maxmising formalism
is that it is a *language* for talking about behavior,
and there are limitations on the types of goal-directed
behavior that it is capable of expressing.  In models
such as the innovation of diffusion through social
networks, it is useless baggage.  In problems such
as the study of institutions, it leads to work such
as that by the New Institutionalists that poses
the absence of processes of institutionalisation
that are well-established, and replaces them by
different forms of tactical social contracts.  And
it focuses the attention of large numbers of bright
people on rationalising behavior that they could be
engaged in explaining if they pursued a scientific
approach instead.

>My weaker claim, the PK microeconomists ought
>to characterize individuals as picking from the
>(subjective) best of their available option seems
>to bother you in two ways. First, it's neither whiskey
>nor lemonade. I.e., why not just go with u max at
>that point? My reason is that it can be hard to formalize
>important considerations, and that shd not stop us
>from talking about them.

	You misread me here.  There isn't any reason to
take utility maximising models as the same thing as
formal reasoning.  This is where I'm willing to forgive
the founders of neoclassical economics, because at the
time there were far fewer examples of other styles of
formal scientific reasoning to choose from.  Partly this
is technological.  Although graph theory, for example,
has roots that go into previous centuries, applied work
in graph theory relies on computer processing.  So,
where early work on social networks in the diffusion of
innovations was dependent on analytical approaches,
now we can adopt a social matrix approach.
	My objection is that the characterisation
*frames* the question as the type of situation that
best fits the situations presupposed by utility
maximisation models.  I have no objection to a
characterisation that, when confronted with the
need to make a choice among a group of alternatives,
individuals select the alternative that they perceive
as the best, as long as we bear in mind that the
question of the social contexts in which people
are confronted with these choices are only some of
the contexts in which they make decisions with
important economic consequences, and that internalised
norms of behavior as 'normal' or 'right' in a given
organisational context can be every bit as important
in determining behavior as reaction to external stimulus.


>Second, you worry (do I misread you
>here) that it will lead to some kind of methodological
>individualism in macroeconomics. Far from it.
>I have repeatedly emphasized that my claim concerns
>PK microeconomists. I view one goal of both micro
>and macro economics to be a search for persistent
>structures. It is fairly well recognized by now (outside
>of Austrian circles, at least) that what can be deduced
>from plausible microfoundations is essentially nothing.
>When coupled with the embeddedness of individuals
>in a macro structure, this suggests not only that macro-
>economics may but that it must have a life of its own.
>(Indeed, even neoclassical macro has fully accepted this, with
>its continued reliance on the ad hoc "representative agent"
>as the source of a priori structure in macro models.)

	The fact that the foundations derived from the need
to provide a defense for use of the utility maximising model
does not permit us to deduce anything does *not* imply
that these microfoundations will permit us to express
any model of individual behavior.  I see no reason to
adopt the shackles of the utility maximising model is
(1) it is not an explanatory model, and is not considered
an explanatory model even by its proponents and (2) it
does *not* give rise to a macrostructure that can be
used to independently test the foundational system.  Why
would *anyone* want to adopt a restriction on the
characteristics of an unobservable feature, when it is
not supposed to be an explanatory model, it is not supposed
to be provide an accurate simulcra of the corresponding
concrete system, and it does not predict any higher level
regularities whatsoever at a higher level in the system
hierarchy?
	Pointing out some of the benefits of formal reasoning
simply begs the question of why *that* formalism.  I find
choice among a set of alternatives to be an incomplete
description of the range of contexts of self-interested
behavior, and therefore an inadequate generalisation for
the introduction of self-interested behavior.  I find the
utility maximising model to be an even more restricted
formal language for talking about all behavior under this
informal characterisation.  I think that Post Keynesian
economists are laboring under enough unavoidable handicaps
as it is, and adopting too much of the same trained incapacity
as more mainstream schools would be a strategic miistake.



Virtually,

Bruce McFarling, Newcastle, NSW
ecbm@xxxxxxxxxxxxxxxxxxx