Valuing Bundled Attributes: A Latent Characteristics Approach

Abstract

Lancaster wrote that attributes of goods may combine to produce characteristics that are the primary units of consumption (1966). Often, the attributes of goods are bundled and cannot be separated. We show that if attributes are technically bundled then valuation analyses that treat these attributes as independent will be wrong. We show that a latent characteristics approach improves upon other methods by incorporating more information about attributes than standard methods and by directly modeling the linkages among attributes. For the case of wilderness hiking, the latent approach provides a statistically better model of site choice than other standard methods. (JEL Q20, Q26)