A non-parametric approach to the estimations of critical inputs to economic models based on consumption data

In the household consumption formulations of Computational General Equilibrium (CGE) modelling utility functions are usually selected a priori and do not take into account any information that can be directly data mined regarding preferences that the observed data might provide.

This research attempts to model consumer demand that is different to the standard parametric approach studied in economic modelling. The methods used here make no assumption about the functional relationship between commodities that are consumed by individuals. Instead models are fitted with reference only to actual observations.

Utility functions as used in economic modelling to represent consumer satisfaction level. The term “utility” was introduced as a numeric representation of consumer satisfaction level as a function of quantity and price. Dissatisfaction with the a priori use of special functional forms has lead to the introduction of the Axioms of Revealed Preferences and an associated program of research to side line the use of utilities. These Axioms define consumer preference based on data observations and are important in this work when trying to model consumer preference directly from these observations.

This work introduces a method via which a purpose built utility function is derived based on real consumer demand data consistent with the Axioms of Revealed Preferences. A different method is introduced in an attempt to simplify the problem and ease the computational load by removing the need to fit a utility function by defining an “ordering” of commodity bundles that are consistent with the Axioms.

Further to this, the construction of a utility allows a more accurate estimate of consumer demand and the change in demand due to price/income factors based on actual economic data.