| Abstract: | Summary This paper gives an empirical reexamination of the  Linear-Expenditure -hypothesis for Austria. It starts with a brief theoretical discussion of the principal properties and restrictions of the Linear-Expenditure-System (LES). To obtain empirical estimates of the parameters of the LES two different estimation procedures are applied, i.e. the original method used byStone and a simplified version of the Systems-Least-Squares-approach (following theMarquardt-algorithm). There are no essential differences between these estimates. They all seem plausible and satisfy the theoretical restrictions.Usually the stability (i.e. time-invariance) of the parameters is accepted without proof. Using the Moving-Window-Regression-technique, however, most of the estimates vary significantly in time. To obtain a direct proof of the time-dependence of the parameters the LES is reestimated now including trend-factors. Especially results considering time-dependent marginal-budget-shares are considerably better than the static-model results.The conclusion of this paper is that the static version of the LES does not explain the consumer behaviour in Austria and that much more effort should be spent on the estimation of dynamic demand systems. |
