The Issue of Macroeconomic Closure Revisited and Extended

Léon Walras (1874) had already realised that his neo-classical general equilibrium model could not accommodate autonomous investments. In the early 1960s, Amartya Sen analysed the same issue in a simple, one-sector macroeconomic model of a closed economy. He showed that fixing investment in the model, built strictly on neo-classical assumptions, would make the system overdetermined, and thus one should loosen some neo-classical conditions of competitive equilibrium. He analysed three not neo-classical “closure options”, which could make the model well-determined in the case of fixed investment. His list was later extended by others and it was shown that the closure dilemma arises in the more complex computable general equilibrium (CGE) models as well, as does the choice of adjustment mechanism assumed to bring about equilibrium at the macro level. It was also illustrated through several numerical models that the adopted closure rule can significantly affect the results of policy simulations based on a CGE model. Despite these warnings, the issue of macro closure is often neglected in policy simulations. It is, therefore, worth revisiting the issue and demonstrating by further examples its importance, as well as pointing out that the closure problem in the CGE models extends well beyond the problem of how to incorporate autonomous investments into a CGE model. Several closure rules are discussed in this paper and their diverse outcomes are illustrated by numerical models calibrated on statistical data. First, the analyses are done in a one-sector model, similar to Sen’s, but extended into a model of an open economy. Next, the same analyses are repeated using a fully-fledged multi-sectoral CGE model, calibrated on the same statistical data. Comparing the results obtained by the two models it is shown that although they generate quite similar results in terms of the direction and - to a somewhat lesser extent - of the magnitude of change in the main macro variables using the same closure option, the predictions of the multi-sectoral CGE model are clearly more realistic and balanced.