European Commission logo

Modelling International Relationships in Applied General Equilibrium

AgricultureClimateEconomyEnergyClimateagricultureCGEGTAPtraderecursive-dynamicex-ante Impact AssessmentNon-tariff barriers


AgricultureClimateEconomyEnergyClimateagricultureCGEGTAPtraderecursive-dynamicex-ante Impact AssessmentNon-tariff barriers

main purpose

MIRAGE is a Computable General Equilibrium Model (CGE). In the European Commission it is used mainly for trade policy analysis, but it has also been used for analysing other policy areas such as agriculture and climate.


MIRAGE is a recursive-dynamic, multi-region, multi-sector CGE model used to analyse policy scenarios. It has been developed by the Centre d'Études Prospectives et d'Informations Internationales (CEPII) in Paris.

It is described in detail in Decreux, Y. and Valin, H. (2007) [1] and in Bchir, M.-H., Decreux, Y., Guérin, J.-L., and Jean, S. (2002) [2]. It is programmed in the General Algebraic Modelling System (GAMS) software.

The model has been used in-house in DG Trade for several years.

The model can be used for ex-ante analyses of policy changes to answer “what-if” type of research questions. It has also been used for ex-post analyses. The scenarios are formulated by changes to economic policy parameters, such as tariffs, non-tariff barriers, subsidies and taxes. These policy changes are usually sector specific or (less commonly) factor-specific as well as country or country-pair specific.

The standard version in MIRAGE uses the GTAP 9.2 database (Global Trade Analysis Project: with the base year of 2011. In 2020, the version of the model used in DG Trade was updated to the GTAP 10 database with the base year of 2014, which is the most recent publically available version of the database also used by other CGE models such as e.g. MAGNET. For ex-ante questions, DG Trade uses recent macroeconomic projections by the IMF and the World Bank.

MIRAGE has been used in DG Trade for Impact Assessments and ex-post analyses, but would lend itself also to Sustainability Impact Assessments (SIA) carried during negotiations in order to inform negotiators, stakeholders and the general public or Economic Analyses of Negotiated Outcome (EANO) carried out after negotiations to inform co-legislators in the approval process (in fact a version of the model has been used by an external contractor for a EANO once:

[1] Decreux, Y. and Valin, H. (2007) "MIRAGE, Updated Version of the Model for Trade Policy Analysis with a Focus on Agriculture and Dynamics", CEPII Working Paper 2007-15

[2] Bchir, M.-H., Decreux, Y., Guérin, J.-L., and Jean, S. (2002) "MIRAGE, A CGE Model for Trade Policy Analysis", CEPII Working Paper 2002-17

model type


Third-party ownership (commercial companies, Member States, other organisations, …)
MIRAGE Consortium


Licence type:
Non-Free Software licence


details on model structure and approach

The model structure is described in detail in Decreux and Valin (2007) and in Bchir et al. (2002).

MIRAGE is similar in structure to most other CGE models, such as the GTAP model.

The key features by which MIRAGE differentiates itself from the dynamic GTAP model (Gdyn) are:

  • Recursive-dynamic formulation (optionally, the user can use a comparative-static version)
  • Imperfect Competition (optional)
  • Additional layer in Armington nesting to reflect quality differences between high-income and developing countries (optional, can be switched on for specific sectors)
  • Reduced labour mobility between rural and urban sectors (optional)

model inputs

  • GTAP database
  • CEPII estimates of non-tariff barriers in services
  • World Bank estimates of non-tariff barriers in goods
  • Macroeconomic projections by: IMF; World Bank; CEPII – based on various primary sources

model outputs

Macroeconomic variables:

  • GDP
  • Wages (by skill level)
  • Consumer Prices

Sector-specific variables (unless specifically mentioned, these can be added up to macro-level results):

  • Bilateral imports and exports
  • Total imports and exports
  • Output (cannot be added up)
  • Value-added
  • Factor demand (Labour and capital, cannot be added up)
  • CO2 emissions

model spatial-temporal resolution and extent

Spatial Extent/Country Coverage
EU Member states 27ALL countries of the WORLD
Spatial Resolution
World-regions (supranational)National
Country level (some countries are grouped into regions)
Temporal Extent
Short-term (from 1 to 5 years)Medium-term (5 to 15 years)Long-term (more than 15 years)
Base year 2014. Projections into the future are possible
Temporal Resolution