Hedonic Pricing in Realistic Urban Structures

This paper explores the importance of continuity assumptions in hedonic price functions. Using a set of actual housing data to mimic a realistic city, and using a set of known preference orderings, it simulates a housing market to recover equilibrium prices. It then estimates price regressions on these data to assess the quality of the estimated marginal values for amenities. Despite good fit to the data, the mean marginal values are generally computed with large errors, typically in the range of 25 to 30 percent and sometimes much larger—even though there are no measurement errors or other data problems and no omitted variables. Evidence is presented that, for some amenities with discrete distributions and/or lower priority in preferences, households with different demands cluster on similar amenity levels. This evidence suggests that either not all marginal values can be priced into the equilibrium and/or that they are so discontinuous that even very flexible functional forms cannot identify them