A revealed preference approach to the measurement of congestion in travel cost models

Travel cost models are regularly used to determine the value of recreational sites or particular site characteristics, yet congestion, a key site attribute, is often excluded from such analyses. One reason for this omission is that congestion is determined in equilibrium by the process of individuals sorting across sites and thus presents significant endogeneity problems. This paper illustrates this source of endogeneity, describes how previous research has dealt with it using stated preference techniques, and describes an instrumental variables approach to address it in a revealed preference context. We demonstrate that failing to address the endogeneity of congestion leads one to dramatically understate its costs. We apply our technique to the valuation of a large recreational fishing site in Wisconsin (Lake Winnebago) which, if eliminated, would induce significant re-sorting of anglers amongst remaining sites. Ignoring congestion leads to an understatement of the lake's value by more than 50%.