Comparing Market Areas Using Kernel Density Estimation

Abstract

From a managers' point of view it is useful to know whether or not a market area has changed over time, and whether two market areas are different. In this paper we propose, using the kernel density estimation method, to esti mate the density of a market area and show how the estimat ed market area densities may be compared for differences using Bayesian cross-validated likelihood methods. The proposed methodology is highlighted using two applica tions. One application involves comparison of lunch and dinner markets of a restaurant and the other application involves identifying the optimal area for a direct mail pro gram.