Re: econometric model

[Peter Dorman <dormanp@xxxxxxxxxxxxx>]

Matt,

As you know, there's a literature out there that says that interest rate
movements have little effect on investment decisions, although I don't know
if that works equally at the economy-wide and firm levels.

Some thoughts.  I think it matters a lot that she's using short rather than
long rates, although, from a monetary policy perspective, it's logical to
focus on a particular Fed lever.  (But what if changes in FF are correlated
with changes in open market policy?)  I have a real problem with final
sales; that should be changes in final sales, perhaps averaged over a longer
time period.  Changes in profits should also be in there, by itself or with
the profit level.

More generally, I think she should read a bit of the r-I literature to
justify the inclusion or exclusion of particular variables.  (G is a bit
mysterious to me.)

Finally, since this is a teaching exercise, I'd encourage her to:

1. Try out different specifications to get a feel for the effect this has,
and
2. Examine her error plot closely.  Look for possible explanations for
outliers, based on her historical knowledge.

This is all off the top of my head...

Peter

"Forstater, Mathew" wrote:

> List econometricians:
>
> I have a student working on a paper to try to test the hypothesis that
> private investment is more sensitive to interest rate hikes than cuts.
> He has proposed a model:
>
> I = alpha + FFRin + FFRdc + CP + FS + G + eps
>
> I is gross private domestic investment
>
> FFRin takes the value of zero when there is a fed funds rate decrease or
> no change in the value of the FFR from one quarter to the next, the
> positive value (of the change in the FFR) when there is an increase in
> the FFR from one quarter to the next.
>
> FFRdc takes the value of zero when there is a FFR increase or no change
> in the value of the FFR from quarter to quarter, a neg value (of the
> change in the FFR) when there is a decrease in the FFR from one quarter
> to the next.
>
> CP is corporate profits.
>
> FS is final sales.
>
> G is federal non defense gross investment.
>
> Comments?  Does it matter that he is using the FFR and not the discount
> rate, prime rate or other rates?  Should he be including G or not? Any
> problems anyone sees (with solutions hopefully).  Obviously it won't
> 'prove' anything inconclusively, but is it a good shot at testing the
> 'elasticity' of investment?
>
> Thanks,
>
> Mat