Skip to content
Back to search results

Computable General Equilibrium

Introduced by
Leif Johansen; developed by Herbert Scarf, John Shoven & John Whalley
Year
1960
Type
Multi-market numerical equilibrium simulation model
Family
Applied general equilibrium
Font size

What does it do?

A computable general equilibrium (CGE) model is a numerical simulation of an entire economy in which optimizing producers and consumers interact through markets that all clear simultaneously.

Building on Walras's general-equilibrium theory and a benchmark social accounting matrix, a CGE model is calibrated to reproduce a base-year economy and then solved for the new vector of prices and quantities that would prevail under a counterfactual policy — a tax reform, a tariff change, a carbon price — capturing how the shock reverberates and re-equilibrates across every market.

The intuition behind the method

Partial-equilibrium analysis looks at one market in isolation; input-output analysis traces quantity linkages but freezes prices. A CGE model does neither — it lets everything move at once. Raise a tariff and import prices rise, consumers substitute toward domestic goods, that shifts demand for labor and capital, which changes factor prices and household incomes, which feeds back into demand, until the whole system settles at a new set of prices where every market clears again. The model is essentially a giant simultaneous equation system encoding 'everyone optimizes, every market clears,' solved numerically to find the new equilibrium and the welfare change it implies.

History

Computable general equilibrium modeling traces to Leif Johansen's 1960 multi-sectoral growth model for Norway, but became practical only after Herbert Scarf developed a constructive algorithm for computing Walrasian equilibria in the late 1960s and early 1970s. John Shoven and John Whalley turned the approach into an applied policy tool in the 1970s and 1980s, surveying the field in their influential 1984 Journal of Economic Literature article and 1992 book. CGE modeling subsequently became a standard instrument at the World Bank, IMF, and national treasuries, with the GTAP project and GAMS-based platforms making large multi-region trade and climate models widely accessible.

When to use it

Use a CGE model when a policy or shock has economy-wide ramifications that work through prices and resource reallocation — tax and trade reforms, climate and energy policy, large structural changes — and when partial-equilibrium or fixed-coefficient input-output analysis would miss the substitution, income, and feedback effects. CGE models require a balanced benchmark SAM and credible elasticity estimates, and they assume optimizing behavior, well-defined market clearing, and (in the standard form) competitive markets. They are best for comparing equilibria under alternative policies rather than for short-run forecasting, and their results are conditional on the chosen functional forms, elasticities, and closure rules, which should always be subjected to sensitivity analysis.

Limitations and considerations

  • Calibration fixes most parameters from a single benchmark year and borrowed elasticities rather than estimating them, so results inherit those assumptions.
  • Outcomes are sensitive to functional-form choices, elasticity values, and the macroeconomic closure rule, sometimes more than to the policy itself.
  • Standard models assume competitive markets, optimizing agents, and an existing equilibrium, which may misrepresent imperfect competition, unemployment, or disequilibrium dynamics.
  • Complexity and limited transparency can make results hard to communicate and validate, and standard CGE models report point estimates without statistical confidence intervals.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Captures economy-wide price responses, input and consumption substitution, and feedback effects that fixed-coefficient input-output and SAM models cannot.
  • Grounded in explicit microeconomic optimization, giving internally consistent, theory-based welfare measures.
  • Calibrated to a benchmark SAM so it reproduces the observed economy exactly before any shock is applied.
  • Flexible and modular: can incorporate multiple regions, sectors, household types, taxes, trade, and environmental accounts for a wide range of policy questions.

Applications

  • Evaluating the economy-wide and welfare effects of tax reforms, tariff changes, and trade agreements.
  • Assessing climate and energy policy, including carbon pricing, emissions trading, and the GTAP-E family of models.
  • Analyzing structural adjustment, development strategy, and poverty and distributional impacts in low- and middle-income economies.
  • Simulating the regional and global effects of trade liberalization through multi-region models such as GTAP.
  • Quantifying the incidence of public-finance and social-policy changes across household groups.

How it works

How it works

1

Each production sector chooses input quantities to minimize cost subject to a production technology F_j (commonly nested CES functions combining capital, labor, energy, and intermediates). This yields factor and intermediate demand functions that depend on the prevailing prices, embedding flexible substitution that fixed-coefficient input-output models lack.

Formula
\min_{x} \; \sum_k p_k x_{kj} \;\; \text{s.t.} \;\; y_j = F_j(x_{1j}, \dots, x_{Kj})
2

Each household maximizes a utility function U_h subject to a budget constraint defined by its income I_h, which is itself earned from selling factors at market prices. The result is a set of demand functions describing how consumption responds to prices and income — the behavioral core that input-output and SAM multiplier models hold fixed.

Formula
\max_{q} \; U_h(q) \;\; \text{s.t.} \;\; \sum_k p_k q_{kh} \le I_h(p)
3

Equilibrium requires that, at the solution price vector, supply equals demand in every commodity and factor market simultaneously. These market-clearing conditions, together with agents' budget balance and zero-profit conditions, define the system of equations whose solution is the general equilibrium.

Formula
\sum_h q_{kh}(p) + \sum_j x_{kj}(p) = \bar{Q}_k + \sum_j y_{kj}(p) \quad \forall k
4

Rather than econometrically estimating every parameter, CGE models are calibrated: free parameters are set so that the model exactly reproduces the benchmark social accounting matrix as an equilibrium at base-year prices (usually normalized to one), with key substitution and supply elasticities taken from the empirical literature. Calibration ties the model to observed data and guarantees it replicates the starting economy.

Formula
\hat{\theta} : \; \text{model}(\hat{\theta}, p^0, q^0) = \text{SAM}^0
5

A policy change is introduced by altering an exogenous variable — a tax rate τ, a tariff, an endowment — and the model is re-solved for the new equilibrium price vector that drives excess demand Z(p) to zero in all markets. Comparing the counterfactual equilibrium to the benchmark yields the simulated effects on output, prices, incomes, trade, and welfare (often an equivalent or compensating variation).

Formula
p^{1} = \arg\, [\, Z(p; \tau^{1}) = 0 \,], \quad \Delta W = W(p^{1}) - W(p^{0})

Common pitfalls

  • Reporting a single headline number without sensitivity analysis over key elasticities and closure assumptions, which often drive the result.
  • Choosing a macroeconomic closure (e.g., fixed versus flexible savings, full employment versus fixed wage) that mechanically determines the sign of the welfare effect.
  • Using an unbalanced or poorly sourced benchmark SAM, so the model fails to replicate the base economy and propagates accounting errors.
  • Interpreting CGE comparative-statics results as time-path forecasts when the model only compares equilibria, not transitions.

Frequently asked questions

How does a CGE model differ from an input-output or SAM multiplier model?

Input-output and SAM multiplier models are linear, demand-driven, and price-free: output responds proportionally to final demand with fixed coefficients. A CGE model uses the same SAM as its benchmark database but adds optimizing producers and consumers, flexible substitution, and market-clearing prices, so a shock triggers price changes and resource reallocation. The trade-off is that CGE requires elasticity assumptions and is harder to validate, while input-output is transparent but ignores prices and substitution.

What is calibration and why is it used instead of estimation?

Calibration sets a model's free parameters so it exactly reproduces the benchmark social accounting matrix as an equilibrium at base-year prices, with elasticities drawn from the literature. It is used because a full CGE model has far more parameters than a single cross-section of data could identify econometrically. The cost is that the model's behavior rests on assumed elasticities and one benchmark year, which is why sensitivity analysis over those values is essential.

What is a 'closure rule' in a CGE model?

A closure rule specifies which macroeconomic variables are fixed (exogenous) and which adjust (endogenous) to balance the model — for example whether investment is savings-driven or savings adjust to investment, whether the labor market clears at full employment or wages are fixed, and how the government balance closes. Because these choices determine which adjustment margins absorb a shock, the chosen closure can substantially change the sign and size of the simulated welfare effect, so it must be stated and tested.

Related methods

Sources

  1. Shoven, J. B., & Whalley, J. (1992). Applying General Equilibrium. Cambridge Surveys of Economic Literature. Cambridge University Press.

    ISBN: 9780521319867
    Copied.
  2. Shoven, J. B., & Whalley, J. (1984). Applied general-equilibrium models of taxation and international trade: An introduction and survey. Journal of Economic Literature, 22(3), 1007–1051.

    Open publication reference
    Copied.