Applied General Equilibrium Modelling ANNARES, STAGE and R23 Models

© Scott McDonald

GLOBE Model Genealogy

The GLOBE model project was started by Scott McDonald and Karen Thierfelder in June 2002 at a conference on Poverty, Trade & Tools for Development, which was hosted by the International Food Policy Research Institute in honour of Sherman Robinson's 60th birthday. During that conference Karen  and Scott discussed, and agreed on, the creation of a SAM based global CGE model calibrated using the GTAP database. Karen and Scott had both 'learnt' CGE models starting with versions of a single country CGE model developed for the US Department of Agriculture's (USDA) Economic Research Service (ERS) under the leadership of Sherman Robinson (Robinson et al., 1990; Kilkenny, 1991). Karen had worked on the NAFTA project and had subsequently used a global model that was descended from the NAFTA model. Scott had been introduced to GTAP by Terrie Walmsley and had used SAMs derived from the GTAP database for teaching single country CGE modelling. These teaching models and the PROVIDE project model (McDonald, 2003) provided the core single country code used to develop GLOBE.


The first full versions of the model were coded by Karen and Scott and were in use by late 2004 (the womod7*.gms series) while the first formal documentation (McDonald, Robinson and Thierfelder, 2007) appeared in May 2007 (based on the womod9*.gms series) as a working paper from the US Naval Academy (working paper 14). Subsequent developments continued until late 2009 when the technical documentation and user guide (McDonald and Thierfelder, 2009 a and b) were revised and an open source version of the model, GLOBE v1, was made available (glb1.gms). In 2012 the model developments since late 2009 were consolidated to produce GLOBE v2, which also formed the base for GLOBE_DYN. In due course GLOBE v2 will be made open source.

There are currently a number of variants of the GLOBE model: GLOBE_EN (energy model), GLOBE_MIG (migration model), GLOBE_IMP (imperfect competition), GLOBE_LAB (labour markets) and other less formal variants developed as parts of various research activities. Current developments of the GLOBE model family, as GLOBE v3, are on-going; an objective of this exercise is to consolidate several of the variants into a single version.

The GLOBE model is part of a suite of models that include a single country model (STAGE - a development of the PROVIDE model that was developed from the USDA ERS model), the R23 model and a range of teaching models - the SMOD suite. All these models use a (overwhelmingly) common set of notation and formats.

Open source version of GLOBE are template models that represent useful starting models for PhD students, and other persons new to global CGE models, who want a model that provides a basis on which they can build by adding their own behavioural relationships.