About GTAP

Getting Started
The Project
The Center
GTAP Consortium
GTAP History
GTAP in the News
GTAP Newsletters
Job Opportunities
Graduate Programs
Contact Us

GTAP at a Glance

About GTAP: GTAP History

A Brief History of GTAP
The following history of GTAP helps to provide an understanding of the core philosophy and motivation behind the project. The timeline and history of the GTAP Data Base are also very helpful in viewing GTAP's and the Center's development over the past decades.

Humble Beginnings: The Birth of GTAP
In the mid-1980s, Dr. Tom Hertel became disillusioned with how CGE modeling was being conducted in Europe and North America. Not only was the data unavailable publicly, there was no way to verify the results that economists were presenting at professional conferences. His frustration led him to consider whether to continue in the field of CGE modeling or move on to another area. After reading papers by Drs. Alan Powell and Peter Dixon, Hertel went to Australia in 1990 on a Fulbright Scholarship to conduct research with the IMPACT Project. In the course of this sabbatical, Powell explained to Hertel the basics behind the IMPACT philosophy: data, training and open-source modeling, supported by econometrics.

Dr. Ken Pearson met the Hertel family and became an instant friend and tacit host. Hertel and Pearson translated a GAMS-based trade model into GEMPACK, and in the process, teamed with Mark Horridge to "mend the family tree" of CGE modeling, bringing together the so-called "Levels" and "Linearized" Schools of CGE modeling. They concluded that each approach had its strengths and its limitations, but that the fundamental answers do not differ if the two are properly implemented.

In addition to working with IMPACT in Melbourne, Hertel undertook a project with the Australian Industry Commission (now called the Productivity Commission), under the direction of John Zeitsch. In Canberra, Hertel met Robert McDougall, the architect of the SALTER model and data base. From these interactions, Hertel learned about its structure and also grew to appreciate McDougall's tremendous analytical capabilities - an appreciation that would translate later into a job offer at Purdue University.

One of the big problems facing the SALTER project was the inconsistency of bilateral trade data and the need for reconciliation, prior to incorporation into a global CGE model. Fortunately, this was an issue that some of the Purdue staff had spent some time on. In particular, then graduate student Marinos Tsigas, Professor James Binkley with Hertel had worked on an approach to trade data reconciliation. In exchange for the reconciled trade data, SALTER shared its core I-O tables with the Center for Global Trade Analysis (hereafter Center), and the first GTAP Data Base was on its way. Subsequently, Mark Gehlhar, then a graduate student with the Center, took over the task of reconciling the trade data and has since vastly improved on the early approaches. His work on this topic now defines the state of the art in this area.

In December 1991, Hertel organized a session at the International Agricultural Trade Research Consortium (IATRC) meetings on CGE modeling. Alan Powell came over to give a presentation based on the IMPACT approach to CGE modeling and policy analysis and to assist Hertel recruit consortium members. After one of many late nights, the acronym GTAP (Global Trade Analysis Project) was born.

1991 was also the year that one of the other key figures in GTAP lore, Judy Conner, appeared on the scene. Upon his return from Australia, Hertel met Judy Conner, one who loved to be busy; they hit it off and began working together. At the time, Judy was working for several faculty members, but as the project grew, so too did the demands on Judy's time. Eventually the Center bought out all of her time, and she subsequently led the project administration until her retirement in 2006.


The Growing Years
The following year, 1992, was an intense period of model and Data Base development. Hertel, his graduate students, and collaborators from other institutions were busy putting all the pieces together. This is when the Robert McDougall's FIT program became famous (or infamous) at Purdue! This program was a tool that McDougall had developed, at the Australian Industry Commission for the SALTER project, that fitted the national Data Bases to international targets. Much of the work was very crude. For example, the tariff data were extracted by hand from various issues of the WTO Trade Policy Reviews. (Afterwards, the individual responsible, Bradley McDonald, knew the material so well that he was subsequently hired by the WTO.)

In 1993, the Center began offering the first week-long GTAP Short Course, in DOS. The Data Base, model, and applications were all released at once, leading to severe burn-out for everyone except Judy, who was ready to march on. Participants in this first course included many who are still active in the project. Authors were invited to submit proposals for studies to be included in a book documenting the GTAP Data Base and model. In December, these papers were presented at the International Agricultural Trade Research Consortium meetings, and they formed the core of the so-called GTAP book, published by Cambridge University Press three years later.

Being economists interested in maximizing social welfare, the GTAP staff made the first Data Base available free of charge. The one hitch was that one could only get a 10x10 aggregation, which was roughly the size of most large models in those days. In the wake of this experiment, the Center found that there was infinite demand for their product offered at a zero price, and were overwhelmed with aggregation requests. There was also another troubling negative externality: researchers did not take the Data Base seriously. This was the first of many failed experiments from which the GTAP staff learned valuable lessons. The version 2 GTAP Data Base carried a healthy price!

However, the Data Base sales revenue did not prove to be sufficient to cover the Data Base production costs, and another funding mechanism needed to be found. After lengthy discussions, the idea of a Consortium of leading agencies surfaced. This proved to be a critical innovation in the history of GTAP. Consortium members are represented on the GTAP Advisory Board, which meets annually to provide input into the strategic direction of the Data Base and model should be updated. They also discuss training courses and conferences, as well as funding new Center research. In addition, they network with other like minded groups because many have common goals.

The first Consortium member was the World Bank (represented by Martin), quickly followed by the Australian Productivity Commission (Phillippa Dee), the Australian Bureau of Agricultural and Resource Economics (ABARE-Vivek Tulpule), and the Economic Research Service (USDA/ERS-Matt Shane). In addition, Alan Powell, the grandfather of GTAP, was included as an At-Large board member with the goal of representing the public interest.

Getting this proposed institutional arrangement approved by Purdue University was extremely challenging, as their first inclination was to tax it with 52 percent overhead, the same way all government funds are treated. Dr. Wally Tyner, then-Agricultural Economics Department chair, provided leadership through the institutional maze, which was invaluable in making the Consortium idea a workable reality. At the 2nd Consortium meeting, the four members had doubled to eight; at the 3rd, it was 12; and at the 4th the total number had risen to 16 Consortium members!

Another key innovation for GTAP grew out of Visiting Professor Randy Wigle's sabbatical at Purdue University in the early 1990's. He was working with HTML language, which would later form the backbone of the World-Wide Web. It took Wigle several months to get Hertel to look at this new software, but when Wigle finally cornered him and explained the potential, a light went on for Hertel, who agreed that it is the perfect technology to disseminate information about GTAP and to promote the global network. This led to his commissioning Wigle to produce the first GTAP web site, which soon became the life-blood of the project.


The Uruguay Round
The Uruguay Round (UR) negotiations were a catalyst in moving the GTAP Data Base and model forward. Using version 2 of the Data Base with new protection data, then-graduate student Betina Dimaranan and Hertel developed one of the first major Uruguay round policy applications. This modified Data Base was used by the majority of the authors contributing to the influential volume, edited by Martin and Winters, which assessed the impact of the final Uruguay Round agreement. While the estimated gains from the UR were initially very far apart ($50 billion vs. $500 billion), once authors began to use the same Data Base, they narrowed considerably, and the remaining differences were readily tracked to differences in model structures. This change represented a great leap forward in serious global policy analysis within a CGE framework.

The years 1994-1995 brought critical changes to the Center and its work. With the initiation of GTAP Data Base Version 3, McDougall joined the Center and took over Data Base production. Dr. Martina Brockmeier spent her sabbatical year at the Center, which led to a 1995 short course in Germany. That special course motivated new EU consortium alliances, plus a workshop in which authors, mostly from the EU, presented their general equilibrium analyses.

Not everything that the Center initially tried continues today. Given the success of the short courses, some in the network expressed interest in learning more about the model and pressed for an advanced short course. In the fall of 1996 and 1997, the Center sponsored an advanced short course. However, enrollment was modest, and participants pointed out that most managers would prefer to send individuals to present their own work, as opposed to take more training. In 1998, the 1st Annual Conference on Global Economic Analysis was conceived at Purdue University.

The Center also continued to offer courses overseas, and in January 1998, the Center conducted a special South African Short Course with support generated by Dr. Will Masters at Purdue University. Relative to other Purdue-based short courses, the course participants' preparation level varied more widely. Hertel remarked that it would be nice if they all had access to some of the preparatory training at Purdue before they entered the course. This experience gave Melanie Bacou, the Center's illustrious Webmaster, the idea of putting together a web-based course based on the Hertel graduate course in applied GE analysis. This web-based course was initially offered on an experimental basis, and it has subsequently become a core part of the curriculum. As a consequence, there has been a huge jump in the quality of learning and discussion at the annual short courses. That 7-week web course now enables course participants to arrive for the one-week, on-site intensive course at a much higher, more consistent level, thereby increasing the level of exchange among instructors and participants.

The year 1999 brought additional important changes: Dr. Terrie Walmsley began her Purdue post- doc, having graduated from Monash University and been supervised by Powell. She began interacting with then-graduate student Elena Ianchovichina and McDougall, who had developed the GTAP dynamic model. In 2000, Walmsley helped to organize the GTAP Dynamic Course, with Ianchovichina and McDougall. Meanwhile, then-graduate student Ken Itakura began working with his Japanese connections, METI (Ministry of Economy, Trade and Industry), that led to several important projects utilizing the Dynamic GTAP Model, plus building support for GTAP in Japan.

There has also been a sustained effort in global climate change mitigation analysis, supported by the U.S. Department of Energy as well as the US Environmental Protection Agency. This interest has funded a series of researchers, including Drs. Gerard Malcolm, Truong P. Truong, Jean-Marc Burniaux and Huey-Lin Lee. These individuals have contributed to the GTAP-E Model and Data Base, which emphasizes energy use and CO2 emissions. With current support from the US-EPA, Dr. Lee has extended this model to include the emissions of non-CO2 gases as well as net emissions associated with land use and land use change.

Dimaranan joined the Center staff in 2000, after completing her dissertation at Purdue University. She took over the process of Data Base construction and documentation, and she has taken this to a new level of professionalism. She and McDougall make a great Data Base team and have since released versions 5.0 and soon-to be-released 6.0 together, breaking new ground and extending the project in creative ways that serve the constituency exceptionally well.

After the Taiwan Advisory Board Meeting, Hertel was convinced that the Center management needed to grow because he could no longer do it all on his own. In fall 2003, after a three-year stint at the University of Sheffield in the UK, Walmsley returned to Purdue as Associate Director for the Center, leading short course and Center management. A few months later, visiting scholar Dr. Sandra Rivera began the Center's strategic planning effort during her year away from the U.S. International Trade Commission (Washington DC). Both have collaborated significantly in moving the Center Strategic Plan forward, making the plan a living document.

In May 2004, the third special short course in Buenos Aires, Argentina began a long-awaited GTAP entry into Latin America. Sponsored by the Inter-American Development Bank (IADB), it enabled the Center to connect with critical policy makers and economists needed to improve the regional data. Graduate students Ernesto Valenzuela and Carlos Ludena were critical in the execution of the course, offering assistance in language skills for participant selection, managing the web course, and helping participants get the first few lectures under their belt in their native language. After just three months, the course has inspired several input-output tables for the next Data Base release as well as a variety of proposals for collaborative projects.

When Hertel spent his sabbatical year in 2005 at the World Bank, Tyner and Walmsley operated as Co-Directors to manage the Center and GTAP activities. Tyner's experience as Agricultural Economics Department Chair plus his international trade negotiation experience helped move the Center forward during this year.


GTAP Today and Beyond
GTAP today has an air of religious fervor about it. The network began small; however, many of the early disciples have remained intimately involved and committed to the project. Many professionals attending a GTAP course and/or conference become aware of what the Center and its GTAP Data Base/Model have to offer, and return to their organizations to sell it. To date, over 500 individuals from numerous countries have completed the GTAP Short Courses. What was a limited lingo shared by a few people in a handful of countries has spread to over 7,000 network participants spanning over 150 countries.

Ultimately, it is the people in the network who make GTAP and the Center what they are today. It is our devoted staff who compile the GTAP Data Base, undertake research and make sure that our courses and conferences are the most memorable experiences. It is every person who has attended a GTAP Short Course or presents a paper at the Global Economic Analysis Conference. It is the dedicated board members who consistently support the mission of the Center with their membership. And especially, it is the vision of the Center, collaborating to bring better ways of framing and understanding economic issues, together.

Adopted from the Short Range Strategic Plan (Sandra Rivera), 2004



A Brief History of the Development of the GTAP Data Base
The GTAP Data Base consists of bilateral trade, transport, and protection matrices that link individual country/regional economic data bases. The regional data bases are derived from individual country input-output tables, from varying years. The table below summarizes the history of the development of the GTAP Data Base. A complete archived list of GTAP Data Base releases may be found here.

The GTAP 1 Data Base relied exclusively on I-O tables inherited from the Australian Industry Commission's SALTER project. For this reason, GTAP adopted the SALTER concordance that identified 37 sectors/commodities.

In GTAP 3 Data Base, 12 of the 24 national data bases still traced their roots back to the Industry Commission's SALTER project. Of course they were updated for each new release using the FIT program1. These I-O tables were heavily concentrated in the Pacific Rim, reflecting SALTER's focus on APEC issues.

Six of these twelve SALTER inherited regions were updated in version 4 (New Zealand, China, Indonesia, Thailand, Taiwan, and Canada). In addition to these updates, version 4 featured updates of four more existing regional data bases, as well as entirely new data bases for 14 countries (Vietnam, Sri Lanka, Venezuela, Colombia, Uruguay, UK, Germany, Denmark, Sweden, Finland, rest of EU, Turkey, Morocco and South Africa). In version 4 the number of sectors was also expanded to 50 to include more agricultural sectors. A special project on energy and climate change led to the creation of the first GTAP Data Base with improved treatment of energy and carbon emissions. The result was the GTAP-E Data Base.

Version 5 updated 16 national data bases (Australia, China, Japan, Korea, Taiwan, Vietnam, India, Colombia, United States, United Kingdom,) and added 23 more countries (including 15 EU countries, 7 Southern African countries and Bangladesh). A system of interim releases was also established to allow new I-O tables to be incorporated into the Data Base as they were contributed. During these interim releases no changes were to be made to the macro data, the only difference would be the disaggregation of a region to incorporate the new country I-O data. Interim releases of version 5 have added 13 more countries in Central and Eastern Europe, Albania, Russia, Malaysia, the Philippines, Singapore, Indonesia and Thailand. This leaves Hong Kong as the last remaining I-O table inherited from SALTER. Since there is no actual I-O table in existence for Hong Kong, this had to be fabricated by SALTER staff, we may wish to contemplate a change in the treatment of Hong Kong in the future - possibly re-estimating this I-O table, or eventually combining Hong Kong with China. In version 5 the number of sectors was further increased to 57, this time services were disaggregated. The treatment of energy was also further improved and incorporated into the primary GTAP Data Base.

In version 6, 2 new countries were added, Madagascar (pre-release 2) and Tunisia (pre-release 3), and 11 other I-O tables were updated (Argentina, Australia, New Zealand, Singapore, India, Colombia, Brazil, Korea, The Netherlands, Turkey and Taiwan). All of these updated I-O tables are significant improvements on previous contributions in terms of a more recent I-O table being used or in the case of Singapore additional external data being used to disaggregate. In version 6.1, 6 new I-O tables were included for Nigeria, Pakistan, Bolivia, Ecuador, Iran and Mauritius. With the exception of the IDE tables, the original I-O tables are all available to consortium members on the GTAP Website.

The remaining 18 regions in the 87 regions of GTAP 6 Data Base are made up of composite data bases representing groups of countries. In light of recent requests the number of composite regions was increased in version 6 to allow for more accurate analysis of regional free trade agreements. As a result the 10 old composite regions have been sub-divided into 18 new composite regions.

The tariff data for the GTAP 6 Data Base also changed significantly in terms of sourcing, coverage, nature and quality, and data processing. In version 5 of the GTAP Data Base, agricultural tariffs and merchandise tariffs were obtained from the two separate sources. Agricultural tariff data were obtained from the Agricultural Trade Policy Database which is based on the Agricultural Market Access Database (AMAD) while data on merchandise tariffs were obtained from the World Bank and UNCTAD through an early version of the World Integrated Trade Software (WITS). In version 6 the tariff data was obtained from the Market Access Maps (MAcMap) data base contributed by the Centre d'Etudes Prospectives et d'Information Internationales (CEPII). The MAcMap data base is compiled from UNCTAD TRAINS data, country notifications to the WTO, AMAD, and from national customs information.

The GTAP 7 Data Base cycle began in April 2007 and was completed in November 2008. The GTAP 7 Data Base has a 2004 reference year, and divides global economic activity into 113 regions and 57 sectors. Version 7 incorporates 2004 data on macroeconomic aggregates, merchandise trade, tariffs, agricultural export subsidies, and the export tax equivalent of quotas on textiles and apparel under the ATC.

25 new I-O tables have been incorporated since the 87 regions of the GTAP 6 Data Base and 9 old IO tables have been replaced. The new tables include Pakistan, Bolivia, Ecuador, Iran, Mauritius, Nigeria, Cambodia, Paraguay, Egypt, Senegal, Nicaragua, Kazakhstan, Kyrgyzstan, Armenia, Azerbaijan, Georgia, Belarus, Ukraine, Costa Rica, Guatemala, Laos, Norway, Ethiopia, Panama and Myanmar. The pre-release data base also includes updated I-O tables for Austria, Australia, Chile, China, Canada, Colombia, Czech Republic, Denmark, Estonia, France, Germany, Greece, Indonesia, Ireland, Korea, Mexico, Morocco, Netherlands, Peru, Philippines, Poland, Romania, Russia, Spain, Switzerland, South Africa, Sri Lanka, Sweden, Turkey, UK, United States, Uruguay and Vietnam.

Bilateral services trade data was incorporated into the GTAP 7 Data Base thanks to Nico van Leeuwen and Arjan lejour from the CPB, Netherlands. Significant improvements were also made to energy, the consumer demand elasticities and agricultural production targeting.

In May 2010, the Center released the GTAP 7.1 Data Base, which included improved IO tables contributed for EU-27, Viet Nam and China as well as revised domestic support data for EU-25 and USA. As a step towards ensuring data quality, the Myanmar table was removed.

In March 2012, the Center released the GTAP 8 Data Base, which included reference years for 2004 and 2007. Newly added data for Bahrain, Cote d'Ivoire, El Salvador, Ghana, Honduras, Israel, Kenya, Kuwait, Mongolia, Namibia, Nepal, Oman, Qatar, Saudi Arabia and United Arab Emirates were added as well as updated/improved data for Australia, Bolivia, Brazil, Cameroon, China, Ethiopia, India, Malaysia, Mexico, Norway, Philippines, Switzerland, Thailand, Uganda, Venezuela and Zimbabwe were included in this release.

In October 2012, the Center for Global Trade Analysis celebrated the 20th Anniversary of GTAP with a panel discussion on "Purdue University's Role in International Outreach". Following the panel, a reception was held featuring photos from the past 20 Years of GTAP Short Courses and 15 Years of GTAP Conferences.


Project Timeline


1 The FIT program lies at the heart of the GTAP Data Base construction process. FIT uses entropy theoretic methods to update and create a consistent data base, where all the data - Input-output tables, trade, protection, macro and energy - are consistent with each other. It is this consistency which is the core value-add of the GTAP Data Base.