Linear approximations to the quadratic almost ideal demand system

Abstract

This paper investigates linear approximations to the recently popular quadratic almost ideal demand system (QUAIDS) by proposing a new composite variable and conducting a simulation study. The linear approximations are especially useful when one uses nonstationary time series, to which nonlinear systems are difficult to apply properly. The new composite variable performs well in combination with the price indices appropriate for linearizing the almost ideal demand system. The QUAIDS can be linearly approximated on a practical basis if the appropriate combinations of composite variables and elasticity formulas are employed and the base point is set to the point of evaluation.