Testing aggregation of protein food products in urban areas of Iran: A comparison of different Generalized Composite Commodity tests

Although prices of all commodities play a role in consumers' decision-making process, the large number of individual commodities and prices creates problems in analyzing consumers' choice. As a result, to the study consumer behavior, data aggregation is used quite often. To this end, different theories are proposed that justify consistent aggregation of commodities. The Generalized Composite Commodity Theorem (GCCT) and the testing procedure proposed by Lewbel is the original one that considers a necessary condition for consistent aggregation. Davis proposed aggregation testing approaches by resorting to the Bonferroni, Simes, Holm and Hochburg statistical methods for testing Generalized Composite Commodity Theorem in which the necessary and sufficient conditions are considered. This study is aiming to examine the possibility of aggregating meats, meat products, dairy products and eggs in a group named “Animal Proteins” and aggregating all grains in a “Vegetable Proteins” group, and to compare the aggregation results using Lewbel's procedure and the approaches proposed by Davis using data over 1990-2012 period. Results show that based on the Lewbel's method, all animal products except dough, curd and egg can consistently be aggregated in the Animal Protein group, and all grains including split peas, beans and lentils can be aggregated in the Vegetable Protein group. While based on Davis methods, all animal commodities without any exception can be placed in Animal Protein group and all grains are consistently aggregated in the Vegetable Protein group. These results indicate that the latter approach lets a wider range of commodities to be aggregated in a group. This implies that selecting an inappropriate aggregation method can lead to a bias in evaluating consumer behavior.