Testing for Weak Separability in Food Demand

Re􀀂cognizing that parametric tests of separability are not invariant to functional form, the study employed two different demand systems--the AIDS and Rotterdam model--to conduc't a test of weak separability for Canadian food demanD In general, the two models generated different results concerning separability. Test results conditional on the AIDS model suggest that the meats group and also the fats and oils group were not separable from the other food groups examined, while fruits and vegetables were considered separable from beverages, cheese, eggs, cereals, and sugar and syrups. Beverages, in turn, were considered separable only from fruits and vegetables, and cereals. Regarding group separability, test results derived from the Rotterdam model coincided with those of the AIDS model only in the case of fruits and vegetables being separable from beverages, cheese, eggs, cereals, and sugar and syrups. These contrasting results from the two models point out the value of considering alternative specifications when conducting parametric tests of weak separability. Also, these results suggest the need to employ non-parametric tests alongside parametric tests to guide applied work in demand analysis.

Issue Date:
Oct 01 1990
Publication Type:
Working or Discussion Paper
Record Identifier:
Total Pages:
Series Statement:
WP 90/18

 Record created 2017-07-06, last modified 2018-01-23

Download fulltext

Rate this document:

Rate this document:
(Not yet reviewed)