Specifying Parameters in Computable General Equilibrium Models using Optimal Fingerprint Detection Methods

Joint Program Report
Specifying Parameters in Computable General Equilibrium Models using Optimal Fingerprint Detection Methods
Koesler, S. (2015)
Joint Program Report Series, 27 p.

Report 276 [Download]

Abstract/Summary:

The specification of parameters is a crucial task in the development of economic models. The objective of this paper is to improve the standard parameter specification of computable general equilibrium (CGE) models. On that account, we illustrate how Optimal Fingerprint Detection Methods (OFDM) can be used to identify appropriate values for various parameters. These methods originate from climate science and combine a simple model validation exercise with a structured sensitivity analysis. The new approach has three main benefits: 1) It uses a structured optimisation procedure and does not revert to ad-hoc model improvements. 2) It accounts for uncertainty in parameter estimates by using information on the distribution of parameter estimates from the literature. 3) It can be applied for the specification of a range of parameters required in CGE models; for example, for the definition of elasticities or productivity growth rates.

Citation:

Koesler, S. (2015): Specifying Parameters in Computable General Equilibrium Models using Optimal Fingerprint Detection Methods. Joint Program Report Series Report 276, 27 p. (http://globalchange.mit.edu/publication/15992)
  • Joint Program Report
Specifying Parameters in Computable General Equilibrium Models using Optimal Fingerprint Detection Methods

Koesler, S.

Report 

276
27 p.
2016

Abstract/Summary: 

The specification of parameters is a crucial task in the development of economic models. The objective of this paper is to improve the standard parameter specification of computable general equilibrium (CGE) models. On that account, we illustrate how Optimal Fingerprint Detection Methods (OFDM) can be used to identify appropriate values for various parameters. These methods originate from climate science and combine a simple model validation exercise with a structured sensitivity analysis. The new approach has three main benefits: 1) It uses a structured optimisation procedure and does not revert to ad-hoc model improvements. 2) It accounts for uncertainty in parameter estimates by using information on the distribution of parameter estimates from the literature. 3) It can be applied for the specification of a range of parameters required in CGE models; for example, for the definition of elasticities or productivity growth rates.