Re: GT and microfoundations

["Alan G. Isaac" <aisaac@xxxxxxxxxxxx>]

Paul Davidson wrote:

> By prediction I assume you do not mean stating tomorrow's quantitative
> value plus or minus two standard errors.

Right.


>  By accepting an employment
> contract at the University of Tennessee in 1986, I acted -- but  what did I
> predict? And the only things I was "certain" about was the terms of the
> contract, i.e., how many courses I would teach, receiving tenure, first
> year's salary, etc.  And by certain all I mean is that if the other party
> did not live up to its contractual commitments, I had the right to sue in
> civil court for "damages" i.e., a sum of nominal money that I could collect
> only if the University did not declare bankruptcy in the interim.  If
> that  what you meant by predict?

Yes; that's a start.
The number of presumptions about the future are
breathtaking when you start to list them,
aren't they?


> Except when something goes wrong -- that is why even in the current
> prosperity, the number of bankruptcy filings is  hitting new records.  Will
> the bankruptcy rate continue to go up Alan?  We all await your prediction
> before we enter into another contract.

Since you would put zero weight on my prediction
(i.e., saying ahead of time), it would not help you
with yours. And yet, as your example above
illustrates, you will not act without taking attitudes
toward the future. However much we pretend to
skepticism when we philosophize, when we act we
rely on our representations of the consequences of
our acts. (I.e., our predictions.) There is no escaping
this.

Can we characterize the consequences probabilistically?
Even when narrowly possible (e.g., playing roulette),
most of us do not. Are our predictions any good?
Most of the time they are. (Does the car start when you
turn the key? Is the store there when you arrive?
Was your money accepted for the tuna you bought?
Did the tuna nourish you instead of killing you?
Etc, etc, etc, etc.) Are they sometimes badly wrong?
Yes. Do we rely on models when making such
predictions? Yes; in the broadest sense: we rely on
our simple representatations of a complex world.
Are our models always logically consistent? No.
And they are seldom analyzed for consistency.
Does that make them useless? No; thank goodness.

Is this all relevant to making economic policy?
Yes: it is impossible to propose policy without
discussing the consequences of that policy
(i.e., without *prediction*). There is simply
no escaping this.

And what about econmetric predictions?
These should be seen as kinds of data summaries,
and we should be as open to their utility as we
are to, say, simple theoretical characterizations
of a complex reality. Policy makers must learn
to pragmatically take guidance from both.
And economists too, since we are useless if we
cannot propose way to improve things (i.e., to
make the *future* better than the present), by
*pre*dicting the consequences of policy changes.
We cannot even avoid quantitative predictions
(e.g., how much should we cut taxes or raise
spending this year, how much should the interest
rate be cut this quarter, etc.) if we are interested
in activist policy.

Alan Isaac