Document View

This article examines how an environmental hazard affects home values. It uses a geographic information system to obtain the straight-line distance (in feet) from the nearest source of contamination to the homesite to measure how the pollution problem affects home values in terms of distance. In addition, this article examines how home values change before and after remediation efforts. The results confirm that homes closer to the problem area suffer a greater loss. Also, the revelation of a contamination problem decreases home values, while a cleanup of the contamination has the reverse effect. [PUBLICATION ABSTRACT]

This article examines how an environmental hazard affects home values based on both the revelation of a soil contamination problem and the distance from the home to the contaminated soil.

In this study, the source of the contamination is a former creosote plant site, Lincoln Creosote, located in the heart of Bossier City, Louisiana. ArcView Geographic Information System (GIS) software is used to measure distances in the analysis of property values, as well as to visually depict the contamination problems in a residential neighborhood. The results show that home values are affected by the environmental problem; specifically, the closer a home is to the polluted area, the lower its value. Also, home values are shown to decline after the revelation of the contamination problem, but recover after the remediation actions are completed.

Literature Review

As the United States expanded westward after the Civil War, the demand for a method to preserve wood was born, especially for wooden piles and ties used by railroads, shipyards, and ports. Treatment of wood by creosoting became the most popular wood-preserving method in the United States, with the first creosoting plant being erected in 1865 by the Dighton and Somerset Railroad.1 The Louisville and Nashville Railroad established a much larger plant in 1875 in West Pascagoula, Mississippi; the establishment of this plant is often considered the beginning of the modern wood preservation era. In 1927, 127,000,000 gallons of creosote were used to treat wood products in the United States, and in 1940 over 280,000,000 gallons of creosote were used.2 Stirling5 notes the environmental problems associated with the toxic chemicals used in the treatment process, especially those associated with the intentional and sometimes unintentional release of chemicals into the air and, most notably, into the soil.

Bible et al.4 examine how the pollution problems of Lincoln Creosote affect home values. They use the dummy-variable approach to show that the neighborhoods closest to the plant site suffer an 8% loss in value compared to neighborhoods farther away. To more precisely measure how the pollution problem affects home value in terms of distance, the current paper uses the straight-line distance (in feet) from the nearest source of contamination to the homesite rather than a neighborhood dummy variable. In addition, this study examines how home values change between sales that occur before and after the revelation of a problem and before and after remediation efforts. Previous studies by Kohlhase5 and Smolen6 use regression models to examine the impact on price of homes located near Superfund waste sites and hazardous waste landfills, respectively; their distance measures were in miles and tenths of a mile. A thorough analysis of the effects of gas station leaks on adjacent property by Simons, Bowen, and Sementelli7 finds a 14%-16% negative effect in sale prices for residential properties sold after the gasoline contamination became known.

Weber8 provides a detailed discussion of best practice with respect to the valuation of brownfield properties; Weber's article emphasizes the importance of choosing expert witness appraisers (in cases involving the effects of contamination on property values) who use theory that has been tested and valuation techniques that are generally accepted in the relevant scientific community. The results of the current study provide relevant information regarding the stigma effects upon property values using a previously validated theory and a generally accepted valuation model.

A recent article by Jackson and Bell9 describes the importance of properly selecting and analyzing case studies when examining environmentally impacted properties. They provide a framework for appraisers to follow when valuing impacted properties; this framework emphasizes the importance of carefully selecting comparable properties (case studies) in the appraisal process, including factors such as contamination discharge, remediation lifecycle, and costs of cleanup. The study presented here is somewhat different in that rather than focus on a particular property (case), it examines the overall effect of an industrial contamination site on the average value of property sales in the residential neighborhoods in and around the source of contamination. The Louisiana study described here deals with over 500 homes and one major source of contamination and could be viewed as a case itself. This study could be compared to other similar situations in which residential homes are affected by creosote contamination (there are no similar sites in northwest Louisiana).

Anderson10 uses a detrimental conditions matrix approach for analyzing environmental contamination. He discusses environmental contamination within the context of the detrimental conditions (DC) matrix approach, which considers the cost, use, and risk in assessing environmental property. One of the issues discussed under ongoing risk is the stigma effect. Anderson notes that within the risk context, other studies often use a multiple regression analysis (an extension of the sales comparison approach) to evaluate market resistance or stigma. The study presented here uses a similar approach (a multiple regression analysis) that considers property characteristics, environmental conditions (nearness to the source of pollution), and a time factor that considers the date of the announcement of a problem and the date of the cleanup.

Background

A wood-treatment facility, Lincoln Creosote, operated in Bossier City, Louisiana, from 1935 to 1969. Since 1969, no substantial activities have occurred on the site. Wood products such as railroad ties and utility poles were pressure treated at the plant using creosote, chromated copper arsenate (CCA), pentachlorophenol (PCP), and polynuclear aromatic hydrocarbons (PAH) as wood preservatives.11 The original wood-treating process area was located in the western portion of the plant, with untreated wood stored in the northeastern part of the area; treated wood was generally stored in the southern portion of the plant. In addition, a petroleum pumping station12 was located in the southeast part of the woodtreating process area. The pumping station operated from 1946 to 1966.

The periodic release of hazardous substances that occurred as a result of chemical leaks and spills during the treatment processes resulted in the need for an investigation and remediation of the site. Both during the operation of the facility and after it closed in 1969 (before the cleanup in 1992), surface impoundments that were used to collect facility waste overflowed and resulted in the release of hazardous substances. Overflow from the surface impoundments migrated along drainage pathways and contaminated them with hazardous substances such as PCPs, CCAs, and PAHs. Some of these hazardous substances contaminated the soil in adjacent residential neighborhoods. When the plant closed in 1969, wood-treating wastes that contaminated the soil remained on the plant site. Figure 1 shows the original plant site (black area), the location of these hazardous substances (lightly shaded areas), and the locations of home sites sold (blackened squares).

Site Actions

Because of concerns expressed by the local government, the Environmental Protection Agency (EPA) investigated the plant site in 1985 and concluded from soil samples that high concentrations of hazardous substances existed. In 1989, under the direction of the Louisiana Department of Environmental Quality (LDEQ), Joslyn (the owner and operator of the plant from the late 1950s until the time of closing in 1969) investigated and found significantly elevated concentrations of hazardous substances (including arsenic) in the soil on the plant site.

Beginning in February 1990, the public generally became aware of these conditions as the local media began reporting on the environmental investigations.15 Joslyn undertook a cleanup of the site in January 1992 at a cost of $10 million. This was done with the oversight of the LDEQ.

In March 1994, the EPA began planning additional sampling in surrounding neighborhoods. The results of the 1994 sampling revealed that 16 PAH compounds and PCP attributable to the wood-treating chemicals formerly used at Lincoln Creosote were present in soils and ditch sediments in a neighborhood near the plant. With EPA oversight, Joslyn undertook the cleanup of residential area between April 1996 and October 1996. Following this cleanup, the EPA concluded that no further remedial action was necessary to ensure protection of human health and the environment in the nearby neighborhood.

The offsite remedial action removed approximately 15,000 tons of contaminated soil from residential sites extending along Bardot Street, a street adjacent to and downgrade from plant site. From one to two feet of soil were removed from the yards of homes, while the soil under the homes (built on concrete slabs) and under driveways was left undisturbed.

The ArcView GIS software provides a clear visual demonstration of the contamination and the location of nearby homes and is also used to measure the exact distance from a home to the nearest contamination. Since this analysis focuses on the effects on value, only homes that have been sold are included. Figure 2 shows the location of all residential sales and the pollutant paths; Figure 3 shows the location of all buildings in the neighborhood including the residential sales and pollutant paths.

Measuring the Effects of Contamination

A hedonic regression model is used to investigate the possible influence of being located near a source of pollution on the value of the homes. The dependent variable is the sale price adjusted by the appreciation of the price of homes in the metropolitan area (ADJUSTED PRICE).14 The independent variables include the number of heated and cooled square feet (SQUARE FEET), the building age (AGE OF HOME), the number of bedrooms (BEDROOMS), number of baths (BATHS), whether or not a fireplace exists (FIREPLACE; value of 1 if it exists, 0 otherwise), the type of exterior (EXTERIOR; value of 1 if brick, 0 otherwise), the type of air conditioning and heating (AIR CONDITION; value of 1 if central, 0 otherwise), the type of foundation (FOUNDATION; value of 1 if slab, O if pier and beam), the type of garage (GARAGE TYPE; value of 1 if garage, O if carport), the size of the garage (GARAGE SIZE) in terms of number of cars, and two district dummy variables. Based on this model, a time dummy variable and a distance variable are added depending on the sample examined.

Enlarge 200% Enlarge 400%

Enlarge 200% Enlarge 400%

Enlarge 200% Enlarge 400%

Two samples are tested. Both samples use two district dummy variables to control for possible effects caused by the location of homes in a particular area. The home sales are identified by their location in District 1 (south of the railroad track that borders the plant site), District 2 (south of Douglas Street and north of the railroad tracks), and District 5 (north of Douglas Street). The houses in District 3 tend to be newer and more attractive than the houses in Districts 1 and 2. The houses in Districts 1 and 2 are similar in age and price (older and less expensive), but are clearly separated by the railroad tracks. District 1 and District 2 are used as dummy variables.

The first sample includes 507 homes sold between 1988 and 1998. In February 1990 the public became aware of the pollution problem. To estimate how this event affects home value, a dummy variable (TIME) is added to the basic model. The variable takes on the value of 1 for homes sold after February 1990 and the value of O otherwise. It is expected that holding other variables constant, homes sell at a lower price after the recognition of the contamination occurs.

The second sample includes 344 homes sold between February 1990 and October 1996. Again, February 1990 is when the pollution problem was revealed. October 1996 is the time when all the cleanup work, including the plant site and the surrounding residential area, is completed. Since the pollution problem is evident during this time, it is interesting to see how the distance to the contaminated area affects home value; therefore, a distance variable (DISTANCE) is added to the basic model. The variable is the linear distance in feet between the site location of the home sold and the location of the nearest contaminated soil as identified by the EPA. It is expected that the shorter the distance, the greater the adverse impact of contamination on house value. A dummy variable (TIME) is also added to the model. The variable takes on the value of 1 for homes sold after June 1992 and a value of 0 otherwise; June 1992 is the time when the cleanup for the plant site (not including the surrounding areas) was completed. Holding other variables constant, it is expected that this cleanup, albeit partial, should increase home value. An interaction variable, INTDT, is added to detect the possible joint effect of distance and time. A distance dummy variable is created that takes on the value of 1 if the distance from the home to the contaminated site is greater than the median of the variable DISTANCE and 0 otherwise. INTDT is equal to TIME times this distance dummy variable. The reason to use the distance dummy variable instead of the original distance variable (DISTANCE) is to reduce possible collinearity problems caused by the correlation between INTDT and DISTANCE. Also, an interaction variable, INTDS, is added to detect the possible joint effect of distance and home size in square feet. A home-size dummy variable is created which takes on the value of 1 if the home size is greater than the median of the variable SQUARE FEET and 0 otherwise. INTDS is equal to the distance dummy variable created above, times this home-size dummy variable. The reason to use the home-size dummy variable instead of the original home size variable (SQUARE FEET) is to reduce possible collinearity problems caused by the correlation between INTDS and SQUARE FEET.

Results

First Sample

The descriptive statistics and regression results for the first sample are shown in Tables 1 and 2. The time dummy variable has a coefficient of-4,823 with a i-value of-4.13, (significant at the 99% level). This indicates an average price reduction of about $4,800 (about 9.5% of the average home price) due to the revelation of the contamination problem. This may be seen as evidence of the stigma effect resulting from the revelation of the environmental problem. The number of square feet is positive and significant, the age of home is negative and significant, while the number of bedrooms, the fireplace, the air conditioning type, foundation type, garage size, and garage type are positive and significant; these results are as expected by theory and previous empirical evidence. The model does not appear to have a heteroscedasticity problem since the Chi-square value from the White's general heteroscedasticity test is 99.32 with a p-value of .3604. Also, the model does not appear to have a multicollinearity problem since the highest variance inflation factor (VIF) value is 3.49.

Enlarge 200% Enlarge 400%

Enlarge 200% Enlarge 400%

Second Sample

The second sample consists of house sales from February 1990 through October 1996. The number of observations is 344. The simple statistics and regression results are shown in Tables 3 and 4. The distance variable has a coefficient of 4.48 with a /-value of 5.53 (significant at the 99% level); in other words, the price of a house is on average about $4.50 higher for every foot away from the contaminated site. The time dummy variable'"' has a coefficient of 2,663 and a p-value of 2.06 (significant at the 95% level), indicating that the houses sold after June 1992 (after plant site cleanup and the beginning of the offsite cleanup) sold for about $2,600, or 4% more. The two interaction variables are not significant. The regression results for other variables in the model are similar to those in the first test except for garage type and garage size. These two variables become insignificant. The model does not appear to have a heteroscedasticity problem since the Chi-square value from the White's general heteroscedasticity test is 101.17 with a p-value of .98. Also, the model does not appear to have a severe multicollinearity problem since the highest variance inflation factor (VIF) value is 5.42.

Conclusion

This article uses a specific case, Lincoln Creosote, to show that a major event in a neighborhood may affect home values. Specifically, the revelation of a soil contamination problem decreases home values, while completing a cleanup of the contamination has the reverse effect. The implication of these results for appraisers is that the possible effects of contamination on home values must be carefully considered, especially with respect to the time the contamination becomes know and the time the cleanup, if any, is completed. Using the modeling process demonstrated here may provide the appraiser with important information. Also, this paper demonstrates that GIS techniques can provide informative and interesting maps that clearly depict the location of houses with respect to sources of contamination, and calculate the distance between the house and the contamination area.

This study indicates that revelation of the contamination problem reduces, on average, house values by $4,800 (approximately 9.5% of the average house price), an important source of information that may be used in the comparable sales process to adjust for differences in the time of the sale. This type of information may be valuable to appraisers involved in the valuation of houses affected by contamination problems, especially in cases where a class action lawsuit is being pursued. In addition, an appraiser may find it especially useful to know the change in value (due to increased distance from the source of contamination) when making adjustments to comparable sales.

The article also provides insight into how appraisers can examine the effect of other events (such as a nearby change in land use) on home values in terms of timing and distance.