The Washington Post ran a story recently titled “Shale gas boom makes some Pennsylvanians richer — and some poorer.” However, the study on which the article relies is one of those academic analyses that can be interpreted in multiple ways, ignores the obvious and, of course, recommends further research. There is, in fact, a dearth of hard evidence in it to indicate natural gas development has made anyone poorer.
The study is entitled “Shale Gas Development and Property Values: Differences Across Drinking Water Sources” and was issued as a working paper in September of this year by the National Bureau of Economic Research. The study examined over 41,000 property sales in Washington county, Pennsylvania, between 2004 and 2009, ultimately narrowing its database to 19,055 sales, split between areas with and without public water supplies. It used a number of relatively complicated methods to theoretically isolate the impacts of natural gas development from other factors impacting property values. The abstract reads as follows:
While shale gas development can result in rapid local economic development, negative externalities associated with the process may adversely affect the prices of nearby homes. We utilize a triple-difference estimator and exploit the public water service area boundary in Washington County, Pennsylvania to identify the housing capitalization of groundwater risk, differentiating it from other externalities, lease payments to homeowners, and local economic development. We find that proximity to wells increases housing values, though risks to groundwater fully offset those gains. By itself, groundwater risk reduces property values by up to 24 percent.
Is this, in fact, what the rest of the study says? Well, not exactly.
If one read between the formulas and convoluted explanations, there are a few contradictions and qualifications. Consider the following separate statements found in the full text of the report (emphasis added):
We find [with respect to the cross-sectional analysis] a positive and significant impact of proximity to a drilled well, though the interaction with groundwater is negative and insignificant. Inverse distance to a permitted well interacted with groundwater is positive but insignificant. The positive sign on the coefficient may be picking up the fact that proximity to a permitted well implies a likely lease payment. In fact, these lease payments increase with the amount of land leased, and lot sizes in groundwater areas are much larger than in the PWSA areas. Thus, the groundwater-dependent properties may positively capitalize on the permitting of the well before the negative amenities associated with drilling occur. However, given the insignificance of the coefficient on the interaction of groundwater with proximity, it is difficult to draw conclusions regarding the overall impact of proximity to wells for the groundwater area homes. (page 21)
Similar to the cross-sectional and [fixed effect or] FE results, we find that property values go up after a well pad has been drilled within 2000 meters, while properties that rely on groundwater are negatively affected by exposure. We find that permitted (but not drilled) wells do not have a significant effect on property values in our final specification, though controlling for these wells reduces the impacts (both positive and negative) of the treatment on property values relative to column 2 (Table 4, column 4). Though insignificant, the parameter estimate on the interaction term of permitted wells with the groundwater indicator is large and negative, providing some evidence that permitting may be negatively capitalized into the property value by groundwater homes. This could be due to the fact that the new home buyer is aware of the forthcoming drilling activity due to incoming lease payments or that construction has already begun to occur nearby. (page 29)
The estimates in the final specification (column 4) demonstrate that properties in the [public water service area or] PWSA positively capitalize proximity to a well pad by 10.7 percent, and this result is statistically significant. This is most likely due to lease payments, which allow properties in the PWSA to increase their values while avoiding the risks (or perceived risks) of contaminated groundwater. For properties that depend on groundwater, however, the estimate of the effect of drilling a well pad within 2000 meters implies a decrease in property values of 23.6 percent. The net impact of these two effects is made up of a statistically significant reduction in value of 23.6 percent attributable to groundwater contamination risk, partially offset by the 10.7 percent increase (likely) attributable to lease payments. Their difference (-12.9 percent) while not itself significant, suggests that, in contrast to PWSA homes, prices of groundwater dependent properties certainly do not rise as a result of nearby drilling, and may fall because of groundwater contamination risk. (pages 29-30)
Specifically, we focus on the potential for groundwater contamination, one of the most high-profile risks associated with drilling. We demonstrate that those risks lead to a large and significant reduction in property values. These reductions offset any gains to the owners of groundwater-dependent properties from lease payments or improved local economic conditions, and may even lead to a net drop in prices. (page 30)
To the extent that the net effect of drilling on groundwater-dependent properties might even be negative, we could see an increase in the likelihood of foreclosure in areas experiencing rapid growth of hydraulic fracturing. (page 31)
Suddenly, things aren’t quite so clear, are they? What was definitive in the abstract is but a “maybe” in the full text and what was seemingly significant in the former is statistically insignificant in the latter. Yet, the authors were able in the end, in the space of a single sentence, to move from what “might” be a negative impact to wholly unsupported conjecture regarding foreclosures.
This is hardly all that is wrong, however, for a study is only as good as its assumptions and data that go into it and the ones employed in this instance leave a great deal to be desired. Here are a few of the more obvious problems with the analysis:
It’s Not Just the Water
There are any number of factors that can affect property values and they can dramatically differ between rural and urbanized areas. This is why their is something called the Uniform Standards of Professional Appraisal Practice (USPAP) to guide analyses of property values. The Washington County study, of course, is not an appraisal per se. Nonetheless, one would hope it reflects the same considerations, born of experience, about how to fairly compare the values of different properties. This study does not.
One of the key USPAP standards, for example, is a requirement to “identify and analyze the effect on use and value of existing land use regulations, reasonably probable modifications of such land use regulations, economic supply and demand, the physical adaptability of the real estate and market area trends.“
It’s easy to see why this is important in the case of the Washington County study. The study theoretically compared values in two broad areas – those with public water and those without it. Practically speaking, however, it created a third category by excluding the cities of Canonsburg and Washington from much of the analysis on the basis they “face significant changes due to the economic boom associated with shale gas development” (page 19).
Leaving aside the obvious problem of artificially capping the positive side of the equation, this means the study effectively compared rural land with land on the urban fringe where residential development and redevelopment are most likely to take place and values are likely to be higher. Natural gas development is primarily occurring in the more rural areas, where, when an economy is in general economic recession as it was in 2009, values can quickly fall off as other development demands fade with the economy, whereas the urban fringe would retain more value, being on the advancing urban edge.
This is why an appraiser would not compare values or rates of change in values in the different areas without making adjustments for this factor. The Washington County study took no account of such locational factors as land use regulations (which typically restrict density in rural areas), demand, supply, adaptability of the real estate or market area trends. It also ignored the impacts of other mining activities, which are huge in Washington County. Here is the map from the study indicating the areas with and without public water service, along with gas wells:
Now, compare this with a map of coal reserves in the county from its Comprehensive Plan:
It quickly becomes apparent most of the area without public water service is also an area of coal mining activity. The coal industry hasn’t had particularly good economic times recently as the Federal government has greatly increased regulations on it during the time period of the study. Has this impacted value? Almost certainly. Yet, there is no consideration of this factor in the analysis. The word “coal” doesn’t appear in the text, nor do the words “mining” or “zoning.” How does one perform a credible analysis of property value differentials in Washington County without taking these factors into account? Short answer: It’s not possible.
When Did You Stop Beating Your Dog?
It’s hard to think of a better illustration, than this old saying, of perhaps the most serious problem with this study, which assumes hydraulic fracturing is a risky business and then sets out to conform data to that preconceived notion. The study ingeminates this false idea with statements such as this (emphasis added):
There are, however, many potential risks that accompany the drilling and hydraulic fracturing process. The processes required to develop and produce natural gas from shale rock use a great deal of water and require the injection of chemicals deep into the ground at high pressure. Compared with conventional natural gas development, this may result in greater risk to air, water, and health. Important for housing markets and local tax revenues, the environmental impact of shale gas development and the perception of the risks associated with these processes, as well as increased truck traffic or the visual burden of a well pad, could depress property values.
The risks associated with leasing one’s land to gas exploration and production companies are especially important for homes that depend on groundwater as a source of drinking water. One of the most often discussed risks associated with shale gas development is the potential for groundwater contamination. Faulty well casings or cement could provide a pathway for contaminants to reach a drinking water aquifer [SEAB, 2011, Osborn et al., 2011]. Another arises if hydraulic fracturing occurs too close to a drinking water aquifer [EPA, 2011] or if there are naturally occurring hydraulic pathways between the formation and the drinking water aquifer [Warner et al., 2012, Myers, 2012]. Even if shale gas operations do not contaminate groundwater in the short run, the possibility of future groundwater contamination may be capitalized negatively into the property value, resulting in important long-term consequences for the homeowner.
What the study argues, of course, is that perceptions can become reality and it then goes on to help to make this a self-fulfilling prophecy by repeating old canards that do not have a basis in fact. The EPA reference, for example, is to a news release regarding the agency’s shoddy work in Pavillion, Wyoming, about which we and others raised so many questions that it had to go back to the drawing board already and probably will have to do so again. Moreover, if the EPA is to be the guidepost as to what constitutes risk, perhaps these EPA statements should have been considered:
“In no case have we made a definitive determination that the fracking process has caused chemicals to enter groundwater.” Lisa Jackson, U.S. Environmental Protection Agency Administrator ( April 30, 2012)
“I’m not aware of any proven case where the fracking process itself has affected water.” Lisa Jackson, U.S. Environmental Protection Agency Administrator (May 24, 2011)
You’ll notice, also, that the first of those quotes was made after the EPA released its findings for Pavillion, Wyo.
There is also the research of the Ground Water Protection Council (GWPC) on more than 220,000 oil and natural gas wells that were in Ohio and Texas over a 25 year period. Their report, available here, analyzed more than 34,000 wells in Ohio, and found only 184 incidents where there were adverse impacts to groundwater, but here’s the rub; only 12 were related to failures of or gradual erosions to casing or cement. That’s a failure rate of 0.03 percent and near all failures took place prior to extensive regulations enacted in the past decade. Most importantly, there was not a single event where the hydraulic fracturing process was found to have affected groundwater, a fact confirmed by the state itself. Results from Texas were even better, with fewer failures and, again, no cases of fracturing contaminating a water supply.
Given this record, there is no basis for most of the risk assumptions made in the Washington County study. There are also unsubstantiated assertions that accompany these assumptions, as for example, when the authors suggested shale gas development would somehow involve greater risks than conventional natural gas development. There are no facts behind this, no citations and no arguments – just a stand-alone assertion. Such backup as is offered comes from later generic references to studies such as the Myers report, which, as hydrogeologist Don Siegel, noted, was fatally flawed and the Duke University report on methane migration, which has been repeatedly debunked here and elsewhere.
Mineral Rights Aren’t Chicken Feed
The Washington County study mentions mineral rights, but only to dismiss them as a significant factor in the analysis, stating that “If mineral rights are sometimes severed, this would simply reduce the size of the price premium we estimate on well proximity. This should not, however, affect our estimates of the capitalization of groundwater contamination risk unless the probability of mineral right severance is correlated with water source in the area around the groundwater-PWSA boundary.” Really?
Well, no. Severing of mineral rights typically becomes more common after natural gas development of an area has advancedbecause landowners who have sold leases desire to retain the potential future income from them upon sale of their property, that often being the greater proportion of the value associated with a rural property. The opposite is likely to be the case in the urban fringe area where the availability of infrastructure such as public water and sewer adds significantly to the total value of a property and renders mineral rights proportionally less valuable
There is also less likelihood an urban fringe property will get developed for natural gas because of density issues. Therefore, there is not the incentive to sever the rights. This is why severing of mineral rights is more common with larger properties, which the Washington County study acknowledges are more likely in the groundwater served areas. There is, in other words, a strong possibility there is more severing of rights in groundwater served areas than public water service areas – and the report fails to address that. The mere fact these same areas also have coal only increases the likelihood of severed rights in the one and not the other.
What all this means is that mineral rights are not trivial matters with respect to impacts on property values, and they are likely to be much more of a factor in the rural areas where homeowners are dependent on groundwater than the urban fringe areas where infrastructure is in place, development patterns are set, lots are smaller and public water is the rule rather than the exception.
The Washington County study also doesn’t match the reality of what’s happening on the ground in 2012. Here are two “heat maps” generated by Trulia, for example:
There are two takeaways from these maps. First, Washington County stands out as a higher value area, as do Bradford, Susquehanna and Tioga Counties, all of which are prime counties for natural gas development. Also of note: Bradford, Susquehanna and Tioga Counties have almost no public water service. Secondly, the Washington County heat map indicates higher valued homes in several areas of natural gas development that do not have public water service, including Cross Creek, Hopewell and South Franklin Townships, which together account for almost a third of all unconventional wells in the county. The following map illustrates:
The municipalities colored green on this map are communities with unconventional wells (see bolded numbers inside each community) that experienced a 50% or greater gain in median home value between 2000 and 2010, according to the U.S. Census. Washington County as a whole had a 52.6% gain, so these communities are near or above the average. Yellow communities also had unconventional wells but less than 50% gains in median home values. It there is any pattern at all it is that the communities adjacent to public water service areas, not necessarily within them (compare to map above), yield the greater value gains and the three communities without public water service and below average value gains (Amwell, Donegal and Independence) are all coal mining areas.
So, where’s the pattern? It doesn’t appear there is one. The focus of the NBER study was on a single factor under the theory other unobserved factors would be washed out in a generic comparison of neighborhoods largely on demographic factors. This is not the way property values are compared by any reputable appraiser and the results show it because, when all is said and done, there is very little to support the thesis advanced and a whole lot of equivocating and qualifying going on. In that spirit, let me observe this study makes us neither richer nor poorer when it comes to understanding the impact of natural gas development on home values.