Linking CGE and microsimulation models: a comparison of different approaches

In the literature that studies income inequality and poverty, a recent development has been the development of models that link together a macroeconomic model (usually a Computable General Equilibrium (CGE) model) and a microsimulation model. Linking the two types of model allows the modeller to take into account full agents? heterogeneity, whilst at the same time considering the general equilibrium effects of a proposed policy reform. In this paper, I first review in detail three approaches to building linked CGE-microsimulation models: one in accordance with the fully integrated approach, and two following the layered approach (the so-called Top-Down and Top-Down/Bottom-Up approaches). The principal goal of the paper is to present a considered evaluation of the merits and demerits of these alternative methods currently used to link CGE and microsimulation models. To do so I use all three approaches to model the macro- and micro-economic impacts of a policy shock to an archetypical economy (constructed using fictitious data), and compare the results. This analysis highlights the importance of (i) consistency between the underlying macro- and micro-data; and (ii) the precise mechanisms by which feedback effects are passed between the macro and micro models. I develop this latter point further by detailed analysis of the TD/BU approach outlined by Savard (2003), leading to a proposed refinement in the way that feedback effects from the micro level of analysis are incorporated back into the CGE model.