Revised stochastic analysis of an input-output model

A main difficulty of regional analysis is the inaccuracy of regional input-output data. A natural framework for investigation is stochastic input-output analysis. In his study of Central Queensland, West (1986) assumes that input coefficients are normally distributed and derives formulas for the approximation of input-output multipliers means and variances. In his normality framework, these moments do not exits, however. Moreover, an inconsistency in the derivation will be exposed. We remedy these shortcomings by respecification of the stochastic structure and by direct evaluation of the moments through Monte Carlo calculations. West's formulas are quite accurate for an aggregated version of his data set. The leading terms of the formulas can be shown to be first order approximations to the means and the variances.(This abstract was borrowed from another version of this item.)