At 02:03 PM 7/14/97 -0700, you wrote:
>Jim Devine writes,
>> But models and econometrics can give one a greater understanding of what
>> may be true.
And Eric writes
>
>As any formal model and any econometric study depend on
>untestable and/or artificial ancillary assumptions I'm not sure how
>one can distinguish between models/empirical results that are
>"true" and those that are "false." Even some well-known
>mainstream econometricians have claimed that econometric
>modelling is merely the production of "metaphors" which might
>have little to do with "reality".
When I do econometrics i am not thinking about truth or falsehood. I am only
thinking about getting a tentatively adequate representation of the data
processes via coherent and transparent methods which are better than other
representations currently available (adjudicated by currently accepted model
selection criteria) and which are not inconsistent with the economic theory
which i may believe in as a matter of faith.
when i do econometric tests i do not test economic theory. the tests are
of hypotheses about the econometric model which is a quite distinct entity
to the theoretical economic model. clearly, the latter conditions the
construction of the former but it is difficult to map the former into the
latter
via testing.
the economic model is a matter of faith (as joan robinson puts it - a branch
of theology).
the econometric model is a construct that may reflect the economic model. To
draw any conclusions about the economic model from the econometric model
is also a matter of faith.
however, econometric modelling is one important aspect of knowledge
accumulation
and in a quantitative discipline like economics i fail to see how we can
conduct
ourselves unless we do get empirical.
the question of truth and whether there is an absolute truth is another
question.
the old econometrics (dirty stuff) used to claim it was finding true models.
read - the ones they believed in at the time.
even if there is a concept of absolute truth, and the next regression i run
reflects it....i have no way of knowing that i have found it. for practical
purposes it is better to aim at adequate representations than to search for
truth.
>
> . . . Poetry, etc., is basically "about" human emotions.
>
>In my mind, poetry involves heightened attention to
>words, sounds, and meanings. This is not really an aside,
>as it relates to how I see econometrics, but time constraints
>don't permit me time to elaborate.
>
Econometrics is a highly sophisticated art form and has an aesthetic
aspect that takes some beating. the beauty of the data and the models
and the dynamic interactions and covariances and the wonderful
patterns of long and short run models, and the sheer intellectual depth of
error correction formations makes it like poetry.
modelling is a very emotional (human) experience.
kind regards
bill
#### ## William F. Mitchell
####### #### Head of Economics Department
################# University of Newcastle
#################### New South Wales, Australia
###################* E-mail: [EMAIL PROTECTED]
################### Phone: +61 49 215065
##### ## ### Fax: +61 49 216919
Mobile: 0419 422 410
##
WWW Home Page: http://econ-www.newcastle.edu.au/~bill/billyhp.html
