Formal methods are used to characterize the uncertainty in the computable general equilibrium (CGE) model outputs to assess the use of the CGE model of China (integrated energy-economy- environment dynamic CGE, TEDCGE) for carbon tax policy issues. Monte Carlo experiment was used for the parameter uncertainty propagation and unconditional sensitivity analysis, using the variance of the con- ditional expectation (VCE) as the importance index to identify critical uncertainties. The results illustrate the statistical characteristics of TEDCGE outputs and sensitivities of the TEDCGE outputs to 50 uncertain elas- ticities. The results show that the carbon tax level for a predefined emission reduction goal is quite sensitive to both capital-energy substitution elasticity and inter-fuel substitution elasticity in the production function, while the key parameter for the GDP reduction rate was only the inter-fuel substitution elasticity. Among the various sectors, heavy industry and electricity are most vitally affected by a carbon tax.
