Data Base Management
In an effort to better meet our goals for a timely, high quality delivery of
the version 5 Data Base, Robert McDougall has developed a formal plan for
managing this release. He has also laid out a set of standards to which data
base development work at Purdue must adhere. To the extent that outside
contributors can also follow these standards, this will greatly facilitate
interactions.
From the perspective of the board members, there are a few key changes that
deserve highlighting here. First on everyone's mind is surely the timetable.
We project a pre-release date of December 1, 1999 and a final release date of
June 1, 2000 (the "millennium data set"?).
The second important point has to do with the full automation of the data
base construction process. We are well on the road to this goal already.
However, we envision that by the fall we will be able to easily remake the data
base whenever any new information becomes available. This changes the way we can
interact with consortium members. In fact, Rob has proposed that, after the
initial pre-release, we will continuously update the Data Base whenever bugs are
located or, as new information becomes available. Obviously these "interim" pre-
releases cannot be supported - rather they will be supplied on a "use at your own
risk" basis. (Of course feedback will be most welcome!) However, this process of
continuous improvement will surely be useful for those of you trying to build up
your analysis of critical issues, prior to the final release of version 5. In this
way you can be assured of working with the most recent data. Also, through the use
of several such prerelease updates, the switch to the final release will entail a
minimum of changes to your results/policy prescriptions. This was a major area of
frustration for some of the board members with version 4.
Getting the Data Base documentation out the door has also been a major
challenge in the past. Coordinating dozens of authors (many of who would prefer
to be modeling or working with data instead of writing) is no small chore! What
we propose to do this time parallels what we hope to do on the Data Base itself.
The basic idea is to maintain a web page - much like the one that now contains
the GTAP v. 4 documentation. A full table of contents will be there - together
with links to PDF files for the most recent version of the documentation on each
section. Of course this documentation will often be very rough and it is
important that it be used only for information purposes. It definitely should
not be printed out and reproduced, as this could give a bad image to the
authors, the project and ultimately the entire Data Base. (This was somewhat of
a problem with version 4, after we distributed some preliminary documentation at
the 6th Annual Short Course.) Do you think this will work? Once all of the
chapters are finalized, we will publish the hard copy version of the
documentation. As with version 4, the web version will be subject to continuous
correction, so it is always the preferred source.
GTAP Policy on Value-Added Retailing of the Data Base
In thinking about this new issue, we have sought to apply the principles behind our
existing policy which are:
- free access to highly aggregated versions of the Data Base,
- otherwise, we require data users to make a money contribution to Data Base
development and maintenance,
- the size of the contribution varies with ability to pay,
- we encourage but do no require users to join the GTAP community and
contribute to the Data Base.
For GTAP value-added retailers (VARs) who wish to supply value-added data bases
larger than 10x10, we propose to:
-
require that the end user acquire a GTAP data license,
-
offer flexible pricing for the license, based on the size of the Data Base
supplied to the end user (price is a purchase price not a periodic rent),
-
by default, base the price on our highest standard license price (corporate
license, new user); but allow academic and upgrade discounts as
appropriate,
-
if the VAR supplies the Data Base as part of a size-restricted-aggregation
product, base the license price on the size of the largest Data Base
incorporated in the Data Base, not on the size of the aggregated Data Base,
(*)
-
require the VAR to license the product to the
Center free of charge on most-favored-customer terms,
-
require the VAR to display prominently to end
users:
-
their GTAP license,
-
GTAP contact information,
-
pointers to GTAP data documentation on our web
site,
-
a statement that the CGTA is happy to take
questions about original GTAP data, but cannot help with data modified by
the retailer.
(*) This refers to a somewhat complicated scheme, under which a retailer
sells a product that contains a large aggregated GTAP Data Base, or the
full-size GTAP Data Base, but denies end-users direct access to it. Instead, the
retailers let end-users create their own aggregations of the large Data Base,
subject to some size limitation. For example, the retailer might ship a 20x20
Data Base, but let end-users use only 10x10 or smaller aggregations.
In such a case, we would base our terms to the retailer on the larger, 20x20
Data Base size, not on the smaller 10x10 size.
For VARs who wish to supply value-added Data Bases less than 10x10, we
propose to:
-
not require end-user licensing,
-
require the VAR to license the product to us
free of charge on most-favored-customer terms,
-
require the VAR to make the value-added data
base freely accessible and freely redistributable,
-
require the VAR to allow us to publish the
value-added Data Base, on our web page or otherwise, with due
acknowledgement,
-
note that the requirements for the value-added
Data Base do not extend to any data sets not derived from GTAP that the
retailer may bundle with the GTAP-derived product;
OR, at their option, treat them on the same basis as retailers of > 10x10,
and charge as if for a 10x10 Data Base.
New National Data Bases
Based on the deliberations at the board meeting, there are already several
commitments for new Data Bases in version 5.
The European Commission will be funding work to fully disaggregate the EU-15
member countries. The EC also envisions funding work to disaggregate some of the
countries in what is currently termed the CEEC region (Central and Eastern
Europe). The Japanese EPA envisions contributing a new Data Base built upon the
recently released, 1995 IO table for Japan. At the request of the Japanese EPA,
which is doing work for the Economic Committee of APEC, we will disaggregate
Peru in the version 5 Data Base. This Data Base was contributed too late to be
included in version 4, but it will be incorporated into version 5. There was
also discussion of the possibility of new Data Bases for Korea, the US, and
various economies in Northern and Southern Africa. No firm commitments have yet
been made in this regard.
Tom Rutherford's Proposal: Establishing a Common Baseline Projection
At the bottom of this note is a draft summary of a meeting after the main
Board meeting, on sharing information about baseline scenarios for CGE
projections. Comments and corrections welcome.
This is how the meeting came to be (many of you know all this, but some
missed all or part of it). At the Board meeting, several representatives spoke
about CGE projections work at their agencies. And Tom Rutherford gave a
presentation on developing a baseline for dynamic CGE analysis. On Thursday
afternoon, as we were discussing off-board meetings for Friday, we decided we
should include a meeting on baseline projections. On Friday the meeting duly
occurred, and most of the representatives still in town attended.
From the meeting, the general consensus was that we should assemble baseline
scenario information and share it over the internet. See the attached summary of
the meeting for details. As the summary shows, the consensus doesn't extend to
every detail; where there were differing points of view, I've tried to note them
all without reconciling them. I haven't recorded who was present or who said
what, partly because I don't entirely remember.
How does this relate to Tom Rutherford's proposal? We don't have Tom's views
on this yet, since he left Thursday. Broadly speaking, they appear different but
compatible. Many of the points made at the Friday meeting repeat points Tom made
in his presentation. The Friday proposal takes one part of Tom's proposal and
pushes it further. On the one hand, it doesn't involve a template dynamic model,
as Tom's does; on the other, it goes beyond Tom's proposal for a single
scenario, to cover a collection of scenarios from diverse sources.
My two cents worth: Tom's proposal looks good for providing instructional
materials, and helping new players get into the projections game (perhaps more
specifically, the climate change projections game). The Friday proposal is
suitable for consortium members who have their own specific projections needs
and practices, but feel that they could learn and profit from parallel work
elsewhere. There doesn't seem any reason we can't have both, or a combination of
the two.
We don't have the GTAP development mailing list in place yet; in lieu of
that, I've attached the distribution list I used for this message.
Meeting on baseline scenarios: draft summary
On April 16, following the GTAP board meeting, several of us met to discuss
baseline scenarios for CGE projections. Here is a [draft] summary of points made
in the discussion.
-
Many GTAP users are doing dynamic modeling and
need baseline scenarios.
-
There are economies to be gained from sharing
information.
-
Not everyone has the same needs or wants to use
the same scenarios. We may want several alternative scenarios for the same
variable, e.g., both IPCC and World Bank GDP projections.
-
As far as possible, the shared data should be
atheoretic, so that they are useful to users of different models.
-
For the same reason, it would be desirable to
provide data at the country rather than the GTAP region level.
-
It would be convenient to provide the data on a
web page.
-
We would not need to provide all the data
directly on the web page; we might provide links to other sites.
-
It would be convenient to be able to download
the most widely used series from a single site.
-
Problems might arise if users mixed and matched
incompatible scenarios for different variables, e.g., an optimistic GDP
scenario and a pessimistic investment scenario. A possible solution is to
group the data into complete multi-variable scenarios, e.g. a World Bank
scenario for population, GDP, etc; an IPCC scenario; etc.
-
Also useful would be information on sources,
methods, and citations for the various series.
-
A suitable typical time horizon is 2050. Some
users might want to go out as far as 2100 (and might not be satisfied to
extrapolate trends from before 2050).
-
To be widely useful, a series should run at
least to 2010.
-
A suitable typical starting date would be 1990.
-
Good historical coverage pre-1990 would be nice
to have, some day.
-
Variables of interest include:
- population
- labor force
- labor force, by skill level
- human capital
- physical capital
- GDP
- household consumption
- investment
- global saving
- depreciation
- technological progress
- energy prices
- energy usage
- energy supply
- protection
-
Possible data contributors and sources include:
- ABARE
- CPB
- DoE
- EC
- ERS
- IEA
- IISA
- IPCC
- Project Link
- World Bank
|
1999 Agenda
Photos
