Addressing aggregation bias in zonal recreation models

Abstract

Models of recreation demand are often based on zonal data. Results from such models are susceptible to aggregation bias. We propose a zonal model of recreation that captures some of the underlying heterogeneity of individual visitors by incorporating distributional information on per-capita income from census sources into the aggregate demand function. This adjustment eliminates the unrealistic constraint of constant income across zonal residents, and thus reduces the risk of aggregation bias in estimated parameters. In addition, the corrected aggregate specification reinstates the applicability of generalized maximum-likelihood methods, and increases model efficiency.