What Do Property Values Really Tell Us? A Hedonic Study of Underground Storage Tanks

Groundwater Quality and Residential Property Values

To my knowledge the few studies investigating the effects of groundwater contamination on residential property prices generally find little or no effect. For example, Malone and Barrows (1990) found that nitrates in the groundwater did not impact home prices. Examining seven different towns in Wisconsin, Page and Rabinowitz (1993) reported that assessed values are not affected by well contamination from landfills, industrial sites, or pesticide run-off. Dotzour (1997) found that groundwater pollution did not lead to a significant difference in the average house price in Wichita, Kansas. These studies provided valuable contributions but are well over a decade old, and econometric techniques and data quality have improved greatly since that time.

More recently, Boyle et al. (2010) find that home prices decline by 0.5% to 1% for each 0.01 mg/l of arsenic above the 0.05 mg/l standard set by the U.S. Environmental Protection Agency. This depreciation appears to be temporary, since prices rebound within a few years. Boyle et al. speculate this rebound may be due to the availability of in-home water treatment systems or dissipation of perceived risks once media attention stops. Due to data constraints, the authors could not link arsenic test results to individual properties and instead relied on “neighborhood”-based measures. In contrast, in this paper I am able to link well contamination tests to private wells at individual homes.

Contaminated Sites and Residential Property Values

There is a significant literature on the effects of larger contaminated sites (mainly Superfund sites) on home values. Often the identification strategy in these studies is to account for proximity to the site, and allow the implicit price of proximity to differ before and after a contamination-related event, such as the discovery of contamination, listing on the National Priorities List, cleanup being undertaken, and cleanup completion.² Each event represents new information that may change public perceptions of environmental and health risks and, in turn, affect property values. The change in the premium for distance from a site is therefore believed to reflect a change in residents’ welfare.

Kohlhase (1991) and Michaels and Smith (1990) were among the earliest to study the effects of contaminated sites on property values. Farber (1998) reviews these and subsequent hedonic studies and finds that property values increase, on average, by $3,500 for each additional mile from a contaminated site. However, Boyle and Kiel (2001) find significant variation across studies, ranging from $190 to $11,450 per mile. Most studies find that home prices decrease when a site is placed on the National Priorities List (Kiel 1995; Farber 1998; Boyle and Kiel 2001; Jackson 2001), but Kiel and Williams (2007) find that this is not the case at all sites.

Chattopadhyay, Braden, and Patunru (2005) conducted a hedonic study on homes around a Superfund site on the Great Lakes. Similar to previous studies, they measured the disamenity in terms of distance of a home to the site. For residents within 5 miles of the site, they found that the aggregate willing-ness-to-pay estimates for full and partial cleanup were quite comparable to a parallel stated preference study. This lends support to the use of distance in measuring the magnitude of the disamenity, at least for these more local residents. This comparison, however, required certain assumptions as to how the price gradient for distance from a Superfund site changes after cleanup.

The evidence that home values rebound after cleanup is mixed (Kiel and Zabel 2001; Dale et al. 1999; McCluskey and Rausser 2003; Kiel and Williams 2007). Even though cleanup reduces objective risks, property values may not rebound, because of a lingering social stigma (Messer et al. 2006; Gregory and Scatterfield 2002). The site may still be perceived as a threat, and the surrounding community publicly shunned.

Overall, the mixed results from this literature suggest that there is significant heterogeneity in how different sites impact local property values, and more refined measures of environmental quality are needed to control for such heterogeneity.

Background on LUSTs

While there is a vast literature on how larger types of contaminated sites affect property values, comparability of these studies to LUSTs is unclear. LUSTs are more numerous, less publicized, relatively smaller in size, and pollution is more local.³ LUST sites are relatively homogeneous in that contamination mainly consists of petroleum products, and the sites are generally gas stations or similar types of commercial and industrial facilities. In contrast, Superfund and other contaminated sites are comprised of a wide assortment of prior land uses and pollutants.

There are about 595,000 industrial and commercial facilities that store petroleum or other hazardous substances in underground tanks (US EPA 2011d). Tanks could eventually leak as a result of corrosion, cracks, defective piping, or spills during refilling and maintenance. Leaking contaminants can seep into the soil and local groundwater and may migrate to surrounding water bodies and ecological systems via surface run-off or groundwater flows.

The majority of the regulated USTs contain petroleum substances, the by-products of which are carcinogenic and can also affect the kidneys, liver, and nervous system (US EPA 2011b). Exposure to these contaminants can occur through the consumption of contaminated groundwater, inhalation of vapors, and dermal contact. Those most at risk are among the 15% of Americans who rely on private groundwater wells. Private wells tap into the local groundwater, which could potentially be contaminated by a LUST in close proximity. Furthermore, private wells are not regulated by the Safe Drinking Water Act and are therefore not required to undergo routine testing, monitoring, and treatment (US EPA 2011c).

LUSTs and Residential Property Values

There are few studies on LUSTs and residential property values. Simons, Bowen, and Sementelli (1997) estimate a hedonic model using a cross-section of home sales in Cuyahoga County, Ohio, and find a 17% depreciation associated with homes within 300 feet of a leak at a registered UST facility. They find no effect associated with proximity to registered nonleaking tanks or nonregistered LUSTs. In a later analysis, Simons, Bowen, and Sementelli (1999) find that “contamination” from nearby gas stations reduces home values by 14% to 16%.⁴ Isakson and Ecker (2010) focus on 50 USTs in Cedar Falls, Iowa, that environmental regulators categorized as “no risk,” “low risk,” and “high risk.” They find that the prices of homes adjacent to a high-risk LUST are about 11% lower.

Due to the small sample size and cross-sectional nature of these studies, one should use caution when interpreting the results as causal. In contrast, I utilize a large panel of home sales over 11 years, which allows me to better identify the causal impact of LUSTs on property values. Using the same dataset, my study extends on an earlier analysis by Zabel and Guignet (2012), who emphasize the need to exploit both spatial and temporal variation in identifying the causal effects of LUSTs on home values.

Zabel and Guignet (2012) include neighborhood fixed effects and spatial econometric techniques to minimize potential biases from unobserved spatially correlated influences on house prices. Even more importantly, they observed home sales before and after the leak, allowing them to establish a pre-leak baseline, and analyze how prices change upon the discovery of a LUST, and completion of a leak investigation. They examined home prices within 100, 200, 500, 1,000, and 2,000 meters (m) of a LUST and checked whether the impact of a leak varied depending on the severity of contamination, the presence of an exposure pathway, and publicity surrounding the site.

In general, they found that the typical LUST had little effect on home values, but more publicized (and more contaminated) sites can cause more than a 10% depreciation at homes up to 1,000 m away. Focusing on these sites individually, however, revealed substantial heterogeneity in the price impact, ranging from a 12% depreciation to a 14% appreciation. This again suggests the need for a more refined measure of environmental quality.

In this paper, I devise a quasi-experiment focusing only on leaks at UST facilities registered with Maryland’s Oil Control Program. In contrast, Zabel and Guignet focus on all LUSTs, including historical sites where regulators were previously unaware that an old inactive UST was (or had been) present. When focusing only on registered USTs, a clear counterfactual exists: homes near nonleaking registered UST facilities. These can be compared to homes near leaking USTs, both before and after the leak. This framework allows me to estimate a difference-in-difference type of hedonic model.⁵

How to Measure Environmental Risk

Despite its widespread use, it remains unclear whether distance to the source of pollution is always an acceptable proxy for environmental and health risks, especially in the context of LUSTs. First of all, if the general public is unaware of the pollution problem, then no threat is perceived and distance is unrelated to perceived risk. In this case, there would be no premium for distance of a home from the disamenity.

While people can see gas stations, and other UST facilities, it is unclear whether they are always aware if a leak occurs at a UST near their home. USTs are underground, and there may be no obvious visual cues of contamination. When a leak does occur, there is little media attention, if any, and if there is, it is restricted to only the most severe cases.⁶

MDE requires a responsible party (usually the UST owner) to notify the public only in the most severe cases, and notification is required only for “members of the public directly affected by the release and planned corrective action” (COMAR 26.10.09.08).⁷ Under Maryland real estate disclosure laws, sellers are not required to disclose information about any nearby pollution unless the for-sale property is actually contaminated.

Additionally, simply looking at proximity to a LUST assumes that the spatial extent of the effect on property values is the same across all sites and homogeneous in all directions, but this may not be true. The spread of contamination plumes is complicated by unobserved groundwater flows (Page and Rabinowitz 1993). Cameron (2006) shows the importance of accounting for directional heterogeneity around a contaminated site and presents a method for doing so, but her approach is not applicable here because the effect of LUSTs on home values is too local and there are too few sales to statistically analyze individual sites.

Following the traditional approach in the hedonic literature, I examine the impact of proximity to a LUST on house prices, and how this varies across informational events, namely, the discovery of a leak, cleanup, and completion of cleanup. A key contribution of this paper is that, in contrast to previous hedonic analyses, I also utilize home-specific variation in information and environmental quality. I have compiled a unique dataset of private well contamination tests and correspondence from MDE, which allows me to identify households who are relatively well informed and face actual (or suspected) risks. Well contamination levels are observed at the end of the complicated hydrogeological processes and thus provide a measure of risk that already accounts for spatial heterogeneity of contamination around an individual LUST site and across sites.

A Difference-in-Difference Approach

Consider a single housing market, where the price of home i in neighborhood j at period t (p_ijt) is a function of structural characteristics of the home (e.g., interior square footage) and its location. The latter includes UST facilities and perceived environmental and health risks associated with that location, among other things. Formally, let p_ijt = f (x_ijt, UST_ijt, π_ijt), where x_ijt denotes home structure and neighborhood characteristics, UST_ijt is a vector summarizing the presence (or number of) UST facilities in close proximity, and π_ijt denotes perceived environmental and health risks.

Risk perceptions are formed from a given information set about a home’s location and nearby disamenities: π_ijt=π(UST_ijt, LUST_ijt, Test_it). The vector LUST_ijt denotes the presence of a leaking UST within a given distance of home i in each of the three stages of the contamination and cleanup process.

Based on MDE practice, if a leak is (1) discovered, then an investigation is undertaken by the environmental regulators to assess the situation and determine the appropriate actions. MDE may require that (2) cleanup be undertaken, which could include removal of the tank, excavation of contaminated soil, and the extraction and treatment of groundwater, among other things. Not all LUSTs undergo active cleanup efforts. Petroleum products naturally degrade over time, so if there is no public or environmental threat, then ongoing monitoring and natural attenuation are sometimes deemed the best course of action (US EPA 2004; Khan, Husain, and Hejazi 2004). If cleanup is undertaken, it is usually complete by the time the leak investigation enters the third and final stage, (3) closure of the case, which is reached when the regulatory agency no longer considers the LUST a threat.

If buyers and sellers in the housing market are aware of a LUST and perceive it as a disamenity or risk to human health, then one would expect home prices to decrease upon the discovery of a leak, and to rebound back to pre-leak levels after cleanup. If these conditions do not hold, then property values may be unaffected. It is also possible that prices may not rebound after cleanup because the site may still be perceived as a threat, or because of a residual social stigma (Messer et al. 2006; Gregory and Scatterfield 2002).

A unique aspect of this study is the inclusion of home-specific information regarding leaks and pollution in private groundwater wells, denoted by Test_it. If the MDE suspects that contamination has migrated into a private well, they will notify the residents, usually with a personal letter informing them about the LUST and requesting to test their well. After testing, MDE sends a follow-up letter with the test results and regulatory standards. If contamination is found, additional tests and notification letters may occur. This is not common to all homes near a LUST, thus the households whose wells are tested are relatively well informed about actual or potential risks.

I do not observe perceived risks directly and must therefore estimate a reduced-form hedonic model. Assuming a log-linear functional form, the model is

[1]

where v_j is a neighborhood-specific fixed effect to control for all unobserved time-invariant neighborhood influences, M_t denotes quarterly and annual fixed effects to capture overall market trends, and ε_ijt is a normally distributed error term. The coefficients to be estimated are β, γ, θ, and α.

The coefficient vector γ captures the baseline effect of desirable and undesirable characteristics associated with UST facilities and the surrounding area. For example, a UST site, such as a gas station, offers certain goods and services to surrounding residents but may also be associated with nuisances, such as displeasing aesthetics, traffic congestion, crime, and noise. The vector UST_ijt includes a variable denoting the number of UST facilities in close proximity to a home (whether leaking or not). UST_ijt also includes a dummy variable denoting a UST in close proximity where a leak is discovered at some point during the study period. This is included in order to account for any differences in the baseline price impacts across UST facilities where a leak does and does not occur.

This framework provides a clean quasi-experiment where the “treatment” is the discovery of a leak (denoted by LUST_ijt). Home sales around registered USTs that never leak serve as a control group, sales around LUSTs before the leak is discovered are the treated group before the treatment, and sales after a leak is discovered are the treated group after the treatment. This is similar to what Horsch and Lewis (2009) refer to as a spatial difference-in-difference approach.

Assuming that the unobserved characteristics captured by γ do not change over time in a manner correlated with LUST events, the elements of vector θ are the causal effects of (1) a leak being discovered, (2) contamination being cleaned up, and (3) a leak case being closed, on the value of homes in close proximity. These impacts are measured relative to the value of these homes prior to a leak being discovered nearby. The coefficient α is the additional impact of groundwater well testing and contamination on the value of a home.

A Spatial Autoregressive Combined Model

As a robustness check, spatial autoregressive combined (SAC) models are estimated (LeSage and Pace 2009, 32). A spatiotemporal lag of neighboring home prices is included in the right-hand side of the hedonic equation to soak up any local time-varying confounders. This lag is basically a weighted average of past home prices within some predefined neighborhood. The SAC model also allows the disturbance terms to be spatially correlated. The model is presented below in matrix notation,

[2]

where ε = λ W₂ ε+u. Let n denote the number of observed home sales, then P is a n×1 vector containing the natural log of the price for all sales, W₁ is a row-standardized n × n spatial weight matrix defining the neighbor relations, and ρ is the spatial lag coefficient, which is meant to absorb all unobserved and potentially confounding characteristics associated with the location of a home. W₂ is a row-standardized n ×n spatial weight matrix defining the relationship between neighboring disturbance terms, λ is the spatial autocorrelation coefficient, and u is a n × 1 vector of iid error terms.

Hedonic Results on Proximity to LUST Sites

The estimation results for several variants of equation [1] are shown in Table 3. The dependent variable is the natural log of the sale price (2007 dollars). All the attributes shown in Table 1, as well as annual and quarter fixed effects, are included as independent variables.¹³ A single hedonic price function is estimated for all three counties.

These counties could, however, be considered separate housing markets. For example, Baltimore City County is very urban, whereas Baltimore and Frederick Counties are primarily suburban and rural. Frederick County is also classified as a separate metropolitan statistical area. County interaction terms are included to allow the coefficients corresponding to attributes of the home and its location, as well as the year time effects, to vary across the counties.¹⁴

For ease of presentation, and because there are relatively few sales in close proximity to LUST sites, the estimated price effects of a UST and leak and cleanup events are constrained to be the same across the study area.

The main conclusions are the same, however, when separate regressions are estimated for each county (see Guignet 2012).

Perceived pollution risks are measured by three dummy variables denoting that a LUST is within 500 m and is in one of the three stages: (1) leak discovered, (2) cleanup, and (3) postclosure.¹⁵ The corresponding regression coefficients can be interpreted as a percent change in price relative to home prices before the leak was discovered. To absorb any unobserved confounding influences on prices associated with UST facilities and the surrounding area, I include the number of UST sites (whether leaking or not) within 500 m. To further capture the baseline price effects associated with being near a LUST, I also include a dummy variable denoting that a LUST, where a leak may or may not have yet been discovered, is within 500 m of the home. In a difference-in-difference framework, this variable denotes the “treated” group and is meant to capture any “pretreatment” price differences between the “control” and “treated” groups.

Model A in Table 3 includes all single-family home sales in the entire study area. Neighborhood fixed effects are included to control for all time invariant unobserved effects on house prices that are associated with a particular locale. Only the coefficient estimates of interest are shown, but the sign and significance of those not presented are as expected.¹⁶ The -0.0020 coefficient on Number of USTs within 0-500 m suggests that homes tend to sell for 0.2% less for each additional UST facility within 500 m, all else constant. The coefficient for LUST within 500 m suggests that homes near a UST that will eventually leak tend to sell for 1.5% less. The fact that this coefficient estimate is small and statistically insignificant lends support to the selection of the control group. In any case, neither result should necessarily be interpreted as causal.

The 0.0488 coefficient on Leak discovered suggests a 4.8% increase in property values when a leak is discovered (relative to the preleak value). As seen by Cleanup and Postclosure, Model A suggests a small and statistically insignificant price effect when cleanup is undertaken, and the investigation subsequently completed (again these effects are relative to the value of these homes before the discovery of a nearby leak).

The 4.8% appreciation upon the discovery of a leak is against initial expectations. It is possible that the public is unaware of the discovery of a LUST or does not perceive it as a threat, but this would imply no change in prices. It is also possible that surrounding residents always suspected a contamination problem, and when MDE finally opened an investigation they anticipated that cleanup would soon occur. If that were the case, however, we would not expect property values to revert back to the pre-leak values after cleanup, as suggested by the small and statistically insignificant coefficient estimates on Postclosure. In any case, this counterintuitive result is not necessarily robust in subsequent models that focus only on private well areas, which is where an exposure pathway is present and local residents are presumably more at risk.

Model B is estimated using only home transactions in areas where households rely on private wells for their potable water. In contrast to the previous model, it seems that among these homes the discovery of a leak has little impact on price (as seen by the statistically insignificant 0.0161 coefficient on Leak discovered). Again we see that cleanup activities have a small and statistically insignificant effect. The Postclosure coefficient suggests that upon completion of a leak investigation, when a LUST is presumably deemed to no longer be a threat, there is a 4.9% depreciation (relative to the pre-leak value). This suggests that a residual perception of risk or public stigma remains, even after the environmental threat is eliminated.

Model C focuses only on homes in private well areas and that are within 500 m of a registered UST. As such, this can be interpreted as a more refined difference-in-difference comparison that only compares homes near USTs where a leak did (the “treated” group) and did not (the “control” group) occur. The previous hedonic models contained a similar difference-in-difference comparison, but conditional on numerous observables. Focusing on a relatively more homogenous set of homes may reduce the influence of unobserved factors on the implicit price estimates.

Similar to the previous model, Model C suggests that the discovery and cleanup of a leak has a small and statistically insignificant impact on local home values. There is some evidence of a small depreciation after closure of a leak investigation, but this is only marginally significant.

Thus far, I draw no conclusive evidence that the discovery and subsequent cleanup and closure of a leak investigation impact local home values. The results are mixed, with estimated coefficients varying in statistical significance, magnitude, and sometimes sign. This holds true for numerous other regression results not presented here, including repeat sales models, county-specific regressions, spatial autoregressive combined (SAC) models, and “propensity score” models, which effectively compared homes around UST sites with a similar propensity for a leak (see Guignet 2012).

Perhaps there is no impact on property values because potential buyers and sellers of homes that are merely in close proximity to the LUST do not perceive it as a threat, or are completely unaware of it. In the next set of regressions I examine an alternative measure of the disamenity, where I know households are relatively well informed and face actual or suspected risks.

Hedonic Results with Well Tests

The regressions in Table 4 include a dummy denoting whether the well water at each individual home was tested by MDE prior to a transaction (Well tested). Model D considers all homes that rely on private wells. The significant coefficient on Well tested suggests that the price of tested homes decreases by 11.9%. The coefficient on Leak discovered is positive and marginally significant, and those corresponding to Cleanup and Postclosure are not statistically different from zero.

There are only 50 sales where MDE tested the well prior to the transaction. Despite this small number of observations, this result suggests a fairly large and statistically significant depreciation. To make sure these dummies are not just picking up unobserved heterogeneity associated with this subset of homes, Model E includes a dummy variable denoting observations where a transaction took place before the well was tested (Sold before well test). All else constant, if homes where well testing occurred prior to the sale are similar to those where testing occurred later, then this dummy controls for any unobserved heterogeneity associated with this subset of homes. The results are similar to those of the previous model, and in fact, if anything the inclusion of this control bolsters the result. I calculate the impact as and find a 12.5% depreciation among tested homes. A similar result is found in Model F, which focuses only on homes in private well areas that are within 500 m of a UST.

[3]

As a final robustness check, the SAC regression shown in equation [2] is estimated. As seen in Model G, the results again suggest an 11% depreciation among sales where the private well was tested. This is the subset of homes where households were relatively well informed about the LUST and groundwater pollution and face actual (or potential) risks.

Overall, I find no conclusive evidence that the discovery and subsequent cleanup and closure of a leak investigation impact the value of homes in close proximity. The only result that holds across some specifications is the counterintuitive appreciation upon the discovery of a leak. These leaks are well distributed both spatially and temporally, so the coefficient on Leak discovered is unlikely to be biased by unobserved influences associated with a particular location or time period. This coefficient could be biased though by unobserved influences that are systematically occurring at LUSTs in different locations and time periods in a manner correlated with the discovery of a leak.

Anecdotally, at least eight of the leak investigations were opened because contamination was found during redevelopment projects. This is one potential explanation as to why some specifications suggest that leak discovery is associated with an increase in home values. Unfortunately, I could not obtain a clear measure of local redevelopment to include in the hedonic regressions.¹⁷

On the other hand, across all specifications I consistently find about an 11% depreciation among homes where the private well was tested. Although the results are not reported here, in subsequent analysis I find that if a test reveals pollution levels above the regulatory standard, then prices decrease by about 13%. However, this is not statistically different from the 11% depreciation among homes where the tests revealed no contamination or that levels were below the standard. Again I emphasize caution in interpreting this result because only 50 transactions are observed where the private well was tested prior to the sale.