Directional heterogeneity of environmental disamenities: the impact of crematory operations on adjacent residential values

Abstract

A hedonic study of residential house sales in Rawlins, Wyoming, was conducted to estimate the impact of an environmental shock from a new point source upon adjacent residential property values. We use a unique data base of house sale prices and associated house attributes, including structural and neighbourhood characteristics and geographic distances and directions from the source of the shock, atmospheric emissions from a new crematory. Our data spans 27 months of house sales: 7 months before, and 20 months after the startup of crematory operations. Results indicate that proximity, measured both in terms of direction and distance from the crematory, imparts a statistically significant negative impact on average house sale prices–an increase of 0.3 to 3.6% of average sale price for every one-tenth mile increase up to one-half mile in distance away from the crematory, but depending on direction from the crematory. This distance benefit increases somewhat with calendar time only for houses located west of the crematory.

Acknowledgement

Helpful comments and suggestions by an anonymous referee are gratefully acknowledged.

Notes

¹Thus the housing market could not anticipate the likely effects of crematory operations.

² Even if spatial correlation were present, an assumption that any spillovers among neighbouring sites are strictly pecuniary would permit the coefficient on the pollution variable in an OLS hedonic price regression to be interpreted as the complete marginal effect of pollution on house value (Small and Steimetz, 2006). Strictly pecuniary effects imply that the value of neighbouring sites affects the sale price of a particular site but does not affect the amenities of that site.

³This may seem to contradict our data, which indicates (in Table 1) a good deal of housing sales activity in the region northeast of the crematory. However, these homes are located further (about 1 mile on average) northeast of the crematory; open fields, a cemetery and school athletic fields occupy much of the nonresidential area directly northeast of the crematory.

⁴Our data indicates a higher correlation between multiple vehicle storage structures and distance away from the downtown area, implying larger lot sizes are most prevalent among residences located at the outer edge of the Rawlins city limits, well beyond the areas plausibly affected by crematory emissions.

⁵Adding covariates to a hedonic price function to avoid omitted variable bias has a cost. If the added covariate is imperfectly measured in the sense that it does not correspond exactly to that feature which the market actually values, measurement error will increase. As more covariates are added, the measurement error bias will increase, thus increasing the noise-to-signal ratio. Atkinson and Crocker (1987) and Graves et al. (1988) use the Bayesian diagnostics of Leamer (1978) to demonstrate that measurement error bias appears to be a more serious problem in hedonic price studies than does omitted variable bias.

⁶Interpretation of dummy variable coefficients in Table 2 requires a slight correction. For example, the correct marginal impact on SALEPRICE of the coefficient for FINBSMT is exp(β_FB )− 1, where β_FB is the coefficient estimate for FINBSMT reported in Table 2 (Halvorsen and Palmquist, 1981).

⁷Our Table 2 coefficient estimates of 0.018–0.019 for T are not an estimate of the average monthly appreciation rate for Rawlins houses over the time span of our data. This estimate captures an ‘embodied’ figure, reflecting both Rawlins-specific appreciation and the discount rate; the two cannot be separated (Kiel and McClain, 1995b).