1. Where We Work

Macro-Modeling At The World Bank

Computable General Equilibrium

Computable General Equilibrium (CGE) models are used for counterfactual policy and economic scenario analysis. Grounded in micro-founded economic theory, they capture how economic agents respond to changes in the economy given the resources available. The standard set of agents includes households, production activities, government, and the rest of the world, with multiple representative agents often modeled within each category — for instance, households grouped by socio-economic status and production activities grouped by sector.

CGE models are well-suited to estimating the long-term, economy-wide effects of policy reforms, since they capture the many direct and indirect channels through which a policy works. A subsidy reform, for example, improves the government budget balance and opens up fiscal space, while also shifting relative prices, consumption decisions, factor demand, and income distribution — and a CGE model can estimate the combined effect across all these channels. Combined with microsimulation modules, the models can also quantify distributional effects, helping policymakers weigh the costs and benefits of reform options and design appropriate compensation packages for affected industries or households.

The CGE modeling team provides modeling services to the World Bank Group and external clients, developing models either as standalone tools for transfer to national authorities or for analyzing specific client questions, and tailoring them along dimensions such as sectoral, regional, and household detail. Two flexible, climate- and energy-policy-aware frameworks underpin this work: the single-country model MANAGE-WB, used where detailed sectoral and household disaggregation is essential, and the global model CGEBox, used for multi-country analysis and regional spillovers. Both are open-source, co-developed by World Bank staff, coded in GAMS, and equipped with a Graphical User Interface (GGIG), supporting flexible production and demand nesting, vintage capital, multi-pollutant emissions accounting, climate damage functions, and the disaggregation of land markets to Agro-Ecological Zones.
 

«
»

  • MANAGE-WB model (single-country)

    The Mitigation, Adaptation, and New Technologies Applied General Equilibrium at the World Bank (MANAGE-WB) model is a recursive-dynamic single-country CGE framework developed at the World Bank (Beyene, Britz, Christensen, Dudu, and Galindev, 2025). It is developed either as a general standalone tool for transfer to national authorities or for analyzing specific questions determined by the client, whether a national authority or a World Bank Group country team. As a dynamic model, MANAGE largely follows a neoclassical growth specification. Labor force growth is mostly driven by demographics, while capital accumulation derives from savings and investment decisions. The model allows for a wide range of productivity assumptions, such as autonomous improvements in energy efficiency that can differ across agents and energy carriers.

    The model supports a wide range of extensions in a modular fashion, such as updating demand-function parameters based on income growth, capital vintaging, or endogenous technical progress. It accommodates flexible nesting structures in production functions, final demand, and segmented factor markets; multi-output, multi-input production; and a vintage structure for capital that allows for putty/semi-putty assumptions with sluggish mobility of installed capital. Its emission accounting framework tracks several pollutants relevant to global warming and air pollution, and supports the taxation of industrial process and product-use emissions as well as factor-based emissions, such as those from livestock production.

    The model can be calibrated to any Social Accounting Matrix (SAM) that follows a standard set of conventions in representing the economic structure. A single Excel workbook comprises the project-specific SAM, projections, and, for instance, nesting specifications. The SAMs are often built specifically for a project to base the analysis on up-to-date and detailed data for the country in focus. Alternatively, SAMs can be constructed in an automated way from the GTAP database. Emission coefficients can be calibrated to emission inventories from national authorities or the CAIT database.

    The open-source version of the model is available to the public. You will find it on the GitHub platform here: MANAGE-WB. There is a README section containing step-by-step instructions that will set up the model, generate a country data (a bridge file) and run simulations if you have GAMS 39 or later versions.

     

    CGEBox model (multi-country)

    CGEBox (Britz, 2016) provides a flexible, extendable, and modular code basis for multi-regional CGE modeling in GAMS. Like MANAGE-WB, it is open-source, coded in GAMS, and equipped with a GGIG-based graphical user interface, and it shares the same flexible nesting, vintage capital, and multi-pollutant emissions accounting features. Its core draws on the GTAP Standard model version 7 in GAMS by van der Mensbrugghe (2018). It can be used for comparative-static or recursive-dynamic analysis. Like most global CGE models, the default model set-up assumes perfect markets for products and factors, suppliers and demanders without market power, cost-minimizing firms, and utility-maximizing consumers. A representative household owns the primary factors, which are allocated to firms to maximize revenues. International trade is depicted by the Armington assumption, such that each region produces a specific differentiated product. Specific to the GTAP Standard model is the so-called global bank, which distributes global savings to maximize expected returns to changes in the capital stock. This mechanism allows for endogenous changes in the balance of trade in simulations. By default, the model embeds the regional household approach, which implies that income is collected by a virtual agent in each region and distributed to government and private consumption as well as savings to maximize a social welfare function.

    CGEBox comprises a wide range of modules that allow the model to be set up flexibly for a given project, such as support for multiple households and a separate government account with debt dynamics, depiction of some international markets based on a Melitz model implying monopolistic competition, disaggregation of bilateral trade flows to tariff lines, depiction of international migration, or disaggregation of production functions and factor markets to sub-national regions. The dedicated module G-RDEM (Britz and Roson, 2019) for long-term baseline generation comprises features such as an empirically estimated MAIDADS demand system with exponential Engel curves; income-dependent cost shares in production and expenditure shares for government and investment demand; savings rates driven by income level and demographics; debt accumulation from trade imbalances; differentiated productivity growth; and targeting of FAO cropland and yield projections.

    A database for CGEBox is built from different versions of the GTAP database and can be enriched with additional data to introduce further detail — for instance, for agro-food markets (Britz, 2022) — or to introduce sub-national data for a country. On demand, it can also include an SDG indicator framework (Wilts and Britz, 2024).

    More information on the model can be found on this webpage.

  • The core of a CGE model is the Social Accounting Matrix (SAM). A SAM1 records data on transactions among all economic agents in an economy over a given period. SAMs expand the explanatory capacity of Input-Output tables by explicitly introducing income and its primary and secondary distribution. A model calibrated to a SAM, therefore, does not need to rely on additional satellite accounts.

    A SAM is ultimately a square matrix in which activities, commodities, factors, and institutional sectors are represented by specific rows and columns. The income of each account is shown along its corresponding row, while its expenditures are recorded in the corresponding column. Each cell records the payment by the account in the column to the account in the row. Typically, a SAM contains six types of accounts: activities and/or commodities, factors, institutions (households and corporations/enterprises), government, capital accounts, and the rest of the world. The disaggregation of these six basic groups determines the size of the matrix. The basic structure of a standard SAM is shown in Figure 1.

    SAM


    Several primary databases are used to populate the cells of a matrix. The main ones are the set of National Accounts systems, household budget, and/or labor market surveys (and others of a socioeconomic nature), as well as statistics related to the foreign sector and international trade.

    SAMs for the MANAGE-WB model include two additional features: they split government and private investment, and introduce government borrowing and debt service from and to institutions (e.g., households, enterprises, and the rest of the world). In most cases, they include household types by income deciles or quintiles, labor by a combination of skill, gender, and formality, and detailed tax accounts.

    To allow for the automatic updating of SAMs when new information (e.g., national accounts data, household surveys, etc.) becomes available or is needed for a project, the MANAGE-WB framework uses a cross-entropy approach as presented in Robinson and El-Said (2000) and further improved with a higher posterior density estimator following Britz (2020).

    SAM estimation is kept separate from the MANAGE-WB model code and is run independently. Users can introduce new information (e.g., split any accounts in SAM, update macroeconomic totals or the whole of the macro SAM, etc.) and the code balances the SAM automatically. The code also allows for incorporating GTAP data to split activities when needed.

    The modeling team has developed Social Accounting Matrices (SAMs) for several countries for specific projects. Since these SAMs usually include proprietary data, they cannot be shared publicly. However, in some cases they can be used by other World Bank teams and researchers working on these projects.
     

    1. The definition of SAM is based on Mainar-Causape et al. (2020).

  • The MANAGE-WB framework has been implemented in over 100 countries in the past 5 years. Working with different country teams on varying analyses places the modeling team at the heart of knowledge spillovers throughout the World Bank. The team carries lessons learned from one project to another and builds on the developments achieved in previous projects. This allows World Bank country teams and country clients to benefit from the knowledge created elsewhere.

    Knowledge and model innovations that the team created in key projects are now at the core of the macroeconomic analysis featured in the Country Climate and Development Reports (CCDRs), and the development of the CGE models at the World Bank is continuously evolving.

    To increase the knowledge spillovers, the team collaborates with modelers around the world and organizes training for the World Bank teams and country clients. These training workshops aim to increase the client teams’ understanding of the CGE models and to encourage the use of them for evidence-based policymaking. The workshops are an important channel to spread the knowledge and experience accumulated via several projects.

FEATURED

MANAGE sticker
Working paper

MANAGE-WB Model Documentation

Model framework, specifications and technical description.
Read More Arrow

Contact Us

Paul Clare
Senior communications officer, World Bank
1818 H Street, NW Washington, DC 20433 USA
Email