The US Model on this Website
The US model is a resource for business forecasters,
government policy
analysts, macroeconomic researchers, teachers, and students.
One can on the site:
- Use the model
to forecast, do policy analysis,
and examine historical episodes. For example, one can change
government policy variables
and examine the estimated effects of the changes.
- Table and graph online and/or download all or part of the
historical data,
forecast data, and data that are creaded by the user.
- Read online and/or download all the documentation, memos, and
papers.
- Download for use on one's own computer the Fair-Parke (FP) program
and the US model.
- Download for use on one's own computer the US model in
EViews.
History
The US
model was developed by Ray Fair in 1974-1976, and it has been used
since then for research, forecasting, policy analysis, and teaching. It has
been available for use on personal computers since 1983 and was the first
such model to be so. (And it was first on the Web!) The current version
contains 25 stochastic equations and about 100 identities.
The data base begins in the first quarter of 1952.
The basic estimation technique is two-stage least squares. The model accounts
for all flow-of-fund and
balance-sheet constraints, which makes it useful for considering
various monetary policy options. A complete description of the
model is in
Macroeconometric
Modeling: 2018.
Comparison to Commercial Models
The main strength of the US model is that it is probably the best
approximation
of the U.S. economy available. It has been extensively tested and analyzed,
and unlike
commercial models, it does not have to be subjectively adjusted to produce
accurate
forecasts. You can have more confidence using the model than using
commercial models that
the results are actually telling you something about how the macroeconomy
works. Commercial models are not even consistently estimated, even though
consistent techniques
have been available for over 50 years, whereas the US model has been
consistently
estimated from its beginning. The past forecasting record of the model is
updated each
quarter, so the user always has a complete record of how the model has done.
The first forecast is dated September 23, 1983.
See
Forecast Record.