Nuclear Waste Transport and Residential Property Values: Estimating the Effects of Perceived Risk
by Jeff Smith
This paper attempts to use a current, highly publicized case regarding spent nuclear fuel shipments in South Carolina to test the effects on residential property values. As the title implies, the premise is that perceived risk may be all that’s necessary to affect human behavior, whether the perception is true or not. The authors cite previous studies that have empirically proven that environmental disamenities (hazardous waste sites, garbage dumps, etc.) reduce residential property values. Several other studies cited show that housing markets respond to both the ...view middle of the document...
A survey conducted in the 3 counties showed perceived risk among households to be greater in Beasley and Charleston county, while the perception was statistically less significant in Aiken (which is the home of the SRS). Consequently, there should be a greater effect on home values in Beasley and Charleston than in Aiken. A telephone survey of realtors in the three counties did not support the argument that nuclear shipments affected property values.
Data for the study included residential property sales in the three S.C. counties between 1991 and 1996 (4 shipments transpired during this period). 9,432 observations are in the sample, 471 from rural Aiken County, 1,834 from rural Beasley County and 7,228 from the populous, urban Charleston County. Therefore, the sample is largely drawn from Charleston County, a potential for bias. Prices were deflated using 1990 taxable values to control for relative annual appreciation. Several variables, including square footage, age of structure, distance from fuel shipment route and several census block level variables were also collected.
The authors use a hedonic house price model with individual structural characteristics (S), neighborhood characteristics (N), and distance (D).
where Pi – transition price, β and γ are vectors that measure marginal effects of their respective characteristics, and δ is the marginal effect of distance from the transport route. The authors note that spatial correlation is a prevalent and underexamined issue when using hedonic house price models. They use Moran’s I to diagnose the spatial correlation and an instrumental variable method to correct for this correlation. Therefore, their spatial hedonic regression model is:
Empirical analysis- the authors use a Box-Cox model to estimate the relationship. They determined, using the Box-Cox, that a log-linear model, with logged dependent variable and linear explanatory variables (other than NPRICE3) is preferred. This was determined without using the data regarding the timing of shipments. That was subsequently estimated by the use of dummy variables. The authors use two working hypotheses, (1) homes distant from the shipment route commanded a greater premium once shipments started, (2) in Aiken and Charleston Counties, the premium did not decrease as...