“…As a result, movements along that curve will reflect an unknown combination of marginal values and changes in the mix of preferences at each location. A number of approaches have been developed to identify demand functions, including treating geographically isolated cities or different housing types as separate market segments (Day, Bateman, & Lake, ; Palmquist, ; Zhang, Boyle, & Kuminoff, ), using repeat sales (Palmquist, ; Ries & Somerville, ), or using owner characteristics, such as age and income to estimate equations in which the marginal values from the bid function are used as dependent variables and then imposing assumptions about the shape of the utility function (e.g., Klaiber & Phaneuf, ). In our case, these methods are unpromising, first, because the small geographic scale of Japan, coupled with its high population density and efficient transport network precludes treating separate cities as isolated markets and, second, because the transactions database cannot be linked to demographic information about the owners.…”