LARGE ECONOMETRIC-MODELS OF AN EAST EUROPEAN ECONOMY - A CRITIQUE OF THE METHODOLOGY

The W series of econometric models of the Polish economy are widely regarded as the most advanced models for East European economies. The econometric methodology of the W models is analysed and the consistency of these models with the objectives of econometrics is evaluated. It is found that the models are generally not consistent with the Christ objectives of econometrics. The model W3 does not produce admissible forecasts and the comparative statics analysis of the model W5 gives uninterpretable results. The models, and especially the parts where so called disequilibrium indicators of the second kind are used, are not built in line with either the traditional Cowles Commission methodology or the contemporary methodologies of Hendry, Sims and Engle-Granger. In particular, the W models suffer from the Lovell bias (due to data mining), endo-exogenous divisibility of variables is ignored, the models are not structurally invariant in an economic policy analysis since they lack forward looking variables, and in the equations with disequilibrium indicators of the second kind the variables are not cointegrated. © 1991.