Variations in price elasticities

Multilevel analyses have been performed on weekly scanner data obtained from 74 stores of a large retailing chain for 30 competing items of a household product group. Discrete price change parameters for the brand being modelled reflect the fact that prices do not change from week to week. Their estimates show that the price elasticity is not constant. Variance components models for four of the products suggest that a brand's inherent attractiveness is affected by the social and demographic characteristics of the retail outlet's catchment area. However, random slopes models indicate that there are no systematically different price effects between stores.