Catching up with history: A methodology to validate global CGE models

Abstract

As a key adjunct to the process of policy formulation, market models are often called upon to quantify possible opportunities and threats. Significant improvements in computational power, database and modelling capacity contributed to a widespread usage of computable general equilibrium (CGE) frameworks in an array of policy fields. Curiously, however, in contrast to modelling efforts in, for example, the biophysical sciences, CGE model findings are seldom subjected to any systematic validation procedure. A cursory review of the literature reveals isolated single country CGE model validation exercises, although with a dearth of available data, there is a paucity of equivalent studies which implement such a procedure in a global CGE context. This paper takes a first step in this direction by proposing a systematic methodological procedure for evaluating global CGE model performance, using a consistent macro and sectoral historical time series dataset and validation statistics taken from the biophysical literature. Focusing on sectoral output trends, the results show that model simulation performs better than extrapolation from past trends. Notwithstanding, simulation error remains high in some sectors, particularly in small economies which have undergone rapid growth. Further econometric tests reveal that simulation error is mainly caused by sector specific factors rather than country specific characteristics. The latter observation is consistent with previous research on productivity specifications in CGE models, which in concert with the validation techniques proposed in this paper, serves as a promising avenue of future research.