Environmental Justice and Coasian Bargaining: The role of race and income in lease negotiations for shale gas

Christopher Timmins Ashley Vissing

May 2014

Abstract With the dramatic growth in U.S. shale gas development over the last decade (particularly in populated areas), lease negotiations have become an important part of the energy landscape, and constitute one of the primary potential sources of benefit for homeowners. We undertake an analysis of the determinants of the characteristics of shale gas leases. The bargaining process associated with the transfer of mineral rights from lessor to lessee shares many features of a classic Coasian bargaining framework. If bargaining were Coasian, it would reduce the need for costly government oversight of the leasing process. Using a unique combination of data sets, we test for whether the bargaining process exhibits characteristics of Coasian efficiency in one of the most active shale gas counties in the U.S. – Tarrant Co., Texas. Our results show that a number of important determinants of willingness-to-pay for avoiding shale gas development do indeed affect bargaining outcomes, suggesting Coasian efficiency – i.e., those who we would expect to have a larger willingness-to-pay to avoid exposure to shale development indeed negotiate for stronger lease terms. However, the argument in favor of Coasian efficiency becomes harder to sustain when we find similar results for race. There is no reason to expect different race groups to have a different willingness-to-pay to avoid environmental harm; fact that we find significant differences in lease quality across race groups, and can explain those differences with a measure of linguistic isolation, suggests an environmental injustice interpretation.

Timmins (christopher.timmins@duke.edu) and Vissing (ashley.vissing@duke.edu) – Department of Economics, Duke University, PO Box 90097, Durham, NC 27708.

! 2

1. INTRODUCTION

Natural gas stored in tight shale formations has grown to become a major source of US

energy supply. Facilitated by investments in technological innovations related to large-scale

hydraulic fracturing and horizontal drilling, large quantities of the resource that had hitherto been

considered inaccessible have been opened up to development. In addition to expanding domestic

US energy supplies, this growth has also increased the revenue of households who own the rights

to the mineral reserves, both in rural and urban plays. However, these benefits are accompanied

by costs external to the production process, including higher levels of air pollution from wells

(Colborn et al. 2012, Caulton et al. 2014, Roy et al. 2014), noise, road damage, air pollution and

accidents associated with increased truck traffic (Gilman, J., B. Lerner, W. Kuster, and J. de

Gouw 2013, Muehlenbachs and Krupnick 2014), and the potential for soil or water

contamination caused by radioactive salts and metals or chemicals used to the treat the wells

(Olmstead et al. 2013, Warner et al. 2013, Fontenot et al. 2013) among other unknown risks.

Under the US legal structure, homeowners are protected from these external costs in one

of two ways. First, ordinances are passed (at the municipal, state, and federal level) that restrict

activities at various stages of the drilling and production processes. Second, homeowners can

negotiate for protections in the terms of the lease agreements they sign, which transfer mineral

rights to operators who then develop the resource on behalf of the homeowner. Typically, the

homeowner receives a royalty payment based on a percentage of the value of the resource sold

by the operator in addition to a one time fixed bonus payment at the time the mineral rights

owner signs the lease. In addition, the lease agreement can specify other terms that restrict the

operator’s activities – noise restrictions, restrictions on how surface disruptions must be restored

after drilling, and restrictions on how long the operator can let the minerals go undeveloped

! 3

before rights revert back to their owner.

The leasing phase in most states is largely unregulated, yet it plays an important role in

the de facto regulatory process. It is therefore reasonable to ask how the outcomes of the lease

negotiation process perform in terms of the criteria that we would typically ascribe to an

environmental rule or regulation – in particular, efficiency and equity. The former is the

traditional purview of environmental economics. In the context of lease negotiation, that

discussion is most closely related to the literature on Coasian bargaining. The idea of Coasian

bargaining (Coase 1960) is that, in the absence of transaction costs and in the presence of well-

defined property rights, producers and recipients of an externality will come to an efficient

agreement about the quantity of externality to be generated, where efficiency is usually described

in marginal terms (i.e., the marginal benefit of the last unit of the externality-generating activity

to the operator is equal to the social marginal cost of that activity experienced by all affected

parties). The result of the Coase “theorem”1 is attractive, in that it establishes an efficiency result

without imposing a large information burden on a social planner.

Analysis of the equity implications of pollution and the rules used to regulate it falls

under the purview of the literature on environmental justice. The empirical literature on the

Coase theorem has sought to explain the determinants of bargaining outcomes – i.e., what are the

determinants of bargaining power, and do the willingnesses to accept payment of “victims”

affect where polluters choose to locate? Often, this leads to an analysis of the role of education

and income, although the relevant factors are not limited to these (Hamilton 1995, Maguire et al.

2005). These studies may find evidence that, for example, polluters pay less in compensation to

poor communities. If those communities have a lower willingness-to-pay to avoid pollution, this

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!1 Coase did not himself label this result as a “theorem”, but it was later coined as such by George Stigler in The Theory of Price (Medema 2014).

! 4

may be an efficient outcome. However, this brings us back to the second set of criteria

mentioned above – equity. In particular, the result of an efficient Coasian bargaining process can

potentially lead to an inequitable outcome, where certain groups (i.e., poor and minorities) are

disproportionately exposed to an environmental harm. Moreover, there may be features of the

bargaining outcomes that suggest inefficiencies according to the Coase theorem; for example,

Coase’s result requires that all parties to the negotiation be equally informed about all costs and

benefits.

This paper analyzes the outcomes of private lease negotiations transferring the mineral

rights for the purpose of shale gas development in the context of the Coasian bargaining and

environmental justice literatures. In particular we empirically analyze the relationship between

household characteristics (paying particular attention to income, race, and linguistic isolation) in

negotiations over “private” regulation (i.e., lease clauses) while controlling for the “public”

regulatory structure (i.e., municipal ordinances) that provides a backstop on the level of potential

harm. Our application considers leases negotiated in Tarrant County, Texas, which overlays the

Barnett tight-shale formation.

Evidence that negotiated lease terms vary systematically with determinants of

willingness-to-pay on the part of homeowners (e.g., income, education, and water source) is

consistent with efficient Coasian bargaining. Controlling for these factors, evidence that race

matters for lease terms hints at the possibility of discrimination; while not efficient from a

Coasian perspective, this outcome is more relevant in the context of environmental (in)justice.

Demonstrating that this race outcome can be explained largely in terms of English-speaking

ability suggests that bargaining outcomes may reflect an important information asymmetry,

further suggesting that the observed outcome may be far from efficient. These results raise

! 5

questions about the need for government regulation of the bargaining process (e.g., uniform lease

terms), or greater regulation of environmental harms with “public” tools (i.e., stronger municipal

ordinances).

This paper proceeds as follows. Section 2 reviews the two relevant literatures that this

paper draws upon – Coasian bargaining and environmental justice. Section 3 describes the

relevant institutional and legal frameworks surrounding shale gas development in Texas. The

specifics of the legal structure surrounding lease clauses, lease negotiation, and municipal

regulation play a particularly important role in our conclusions. Section 4 describes our data,

which are drawn from a novel source of information about the outcomes of lease negotiations

and proprietary data on housing transactions combined with publicly available data on the race

and income of homeowners. Section 5 motivates our empirical approach with a brief discussion

of the Coase Theorem. Section 6 describes our empirical modeling strategy and summarizes

results. Section 7 concludes with a discussion of our results with a particular focus on policy

implications.

2. LITERATURE REVIEW

In this paper, we explore the determinants of the bargaining outcome associated with the

leasing of rights for development of shale gas. In particular, we explore whether the leasing

process exhibits characteristics consistent with efficient Coasian bargaining, or whether other

factors (in particular, factors that could be associated more directly with race and language) play

a role. Our work is therefore related to the literatures on the Coase Theorem, environmental

justice, and several sub-literatures on the development of shale gas resources.

! 6

2.1 Coasian Bargaining

The literature on the Coase Theorem is extensive; we do not attempt to summarize it here

but rather refer the reader to one of several reviews (Schweizer 1988, Medema 1994, Zelder

1998, Medema 2014). Coasian bargaining theory states that, in the absence of transaction costs

and under well-defined property rights, parties will bargain to an efficient outcome in the

presence of externalities. This will place the resource in question into its highest valued use,

regardless of the initial allocation of property rights.

The size of the compensation paid by a polluter to a victim with property rights should

depend upon both the victim's willingness to accept compensation in exchange for tolerating the

externality and the magnitude of the externality. Bargaining power is also likely to have an

effect – residents who are better able to organize themselves are better able to obtain outcomes

favorable to their community. Jenkins, Maguire and Morgan (2004) analyze compensation

offers made by landfill operators to community leaders in a Coasian bargaining process in

exchange for being allowed to operate a facility. Compensation packages of this sort became

popular in negotiations between landfill developers and communities in the late 1980's. Using

data on host fees paid by 104 largest privately owned solid waste landfills in 1996, the authors

examine a number of characteristics that could potentially make bargaining payments larger.

The primary factors that are found to affect the size of payments include citizen participation in

negotiations, experience hosting a landfill, state mandates for minimum host compensation, and

industry concentration (implying oligopoly rents and greater ability to pay). The authors find

some indication of efficiency in the bargaining process, in that some measures of the severity of

the externality (i.e., sludge and tires) do affect the size of the payment. However, they also find

! 7

contradictory evidence in that lower host fees are paid the closer is the nearest subdivision to the

facility.

Whereas Jenkins, Maguire and Morgan (2004) take the size of the externality as given

and model the determinants of the compensation payment, we model the results of the Coasian

bargain over the size of the externality itself, and the subsequent division of the Coasian surplus.

We argue in Section 5 that it is in the former where one should look for evidence of efficiency or

inefficiency in the bargaining process.

In a study of the expansion plans of commercial hazardous waste facilities, Hamilton

(1995) tests three theories for why exposure to environmental harms might vary with race: (i)

pure discrimination, (ii) differences in willingness to accept payment for loss of environmental

amenities linked to income or education, and (iii) propensity for collective action. The latter two

explanations have connections to the theory of Coasian bargaining – in particular, firms will

avoid locations where residents require a greater payment. They will also avoid locations where

a tendency towards collective action will make payments in response to a greater willingness-to-

pay on the part of residents more likely. It can be difficult to break the simultaneity between

neighborhood characteristics and the presence of an environmental harm. Do nuisances locate in

minority neighborhoods in an effort to avoid compensatory payments, or do neighborhoods

become increasingly minority following siting decisions? Hamilton (1995) overcomes this

problem and tests the hypotheses described above by using information on the planned capacity

decisions of commercial hazardous waste facilities (i.e., looking to see how facilities' plans to

expand vary with neighborhood attributes). Because those expansions have not yet taken place

in the data, it is impossible for observed neighborhood demographics to have been determined by

them. Hamlton (1995) finds that neighborhoods (zip codes) targeted for expansion in 1987-1992

! 8

had non-white population of 25% compared to 18% not targeted. Looking more specifically at

the mechanism underlying this disparity, differences in the likelihood of raising firms’ costs via

collective action (measured by voter turnout) offer the best explanation, rather than pure

discrimination or a simple Coasian bargaining story.

2.2 Environmental Justice

Analysts typically date the emergence of the environmental justice movement to a set of

protests that followed the selection of a landfill site in the predominantly African-American

community of Warren County, North Carolina in 1982 for the disposal of PCB's that had been

illegally dumped as part of the Ward Transformer case (Bullard 1994). These protests were

organized, in part, by the Southern Christian Leadership Conference and the United Church of

Christ Commission for Racial Justice; the latter went on to be responsible for the first national-

level analysis to document the correlation between race and pollution (UCC 1987). These

protests also prompted the U.S. General Accounting Office to carry out a study in 1983

documenting that landfills were disproportionately located in Black communities, specifically in

the U.S. South. US GAO (1983) and UCC (1987) were followed by a series of papers that

documented the correlation between race, poverty and exposure to environmental harm (Bullard

1990, Mohai & Bryant 1992, Brown 1995, Szasz & Mueser 1997, Boer et al. 1997, Sadd et al.

1999, Ringquist 2005). These studies found a significant disparity in proximity by race, even

after controlling for income, local land use patterns, the percentage of employees in

manufacturing, population density, and other reasonable variables. Other analyses were more

specifically focused on risk-based measures of pollution exposure, but found similar results

(Morello et al. 2001, Pastor et al. 2002, Bouwes et al. 2003, Ash and Fetter 2002 & 2004,

! 9

Morello-Frosch and Jesdale 2006). In 2007, the UCC updated its 1987 analysis using

information on hazardous waste facility locations, demographic data from the 2000 Census,

finding that poor and minority groups were even more heavily concentrated around hazardous

facilities than had been previously thought.

Analyses documenting the correlation between environmental harms and disadvantaged

status (i.e., race and income) form the first strand of the environmental justice literature. The

second strand seeks to explain the mechanism behind those correlations (Been 1994);

understanding that mechanism is crucial for the design of effective policy. One story seeks to

explain correlations as the result of the siting of nuisances, paying particular attention to the

demographics at the time of siting. A second story focuses instead on residential sorting – i.e.,

the tendency for disadvantaged groups to move into polluted areas where residences are less

expensive ("coming to the nuisance").

One line of research seeks to explain siting decisions with sociopolitical variables – the

argument being that industry and government seek the path of least resistance when siting new

hazardous waste or polluting industrial facilities. This argument is similar to the Coasian

bargaining story found in Hamilton (1995). In particular, industry is aware that many

communities will actively oppose the siting of facilities. Because industry and government do

not want to generate controversy or experience delays with siting plans, they seek to avoid

communities that are most capable of mounting an effective opposition. These communities are

those with abundant resources and political clout and also tend to be affluent, white, and well-

connected. Poor communities and communities of color become an easier target because they

have fewer resources and are not well represented in the decision making of industry and

! 10

government (Bullard 2000, Cole and Foster 2001, Saha & Mohai 2005).2

There are a number of papers that directly compare the siting versus the sorting

explanations. Pastor et al. (2001) show that, over a 30-year period, the correspondence between

polluting facilities and minority communities in Los Angeles was based primarily on a pattern of

disparate siting of facilities in existing communities of color, rather than on geographic shifts in

these populations. Morello-Frosch et al. (2002) find strong evidence of correlation between risk-

based measures of exposure and race (even after controlling for income) along with evidence of

disproportionate siting. However, they find no evidence of sorting behavior. Depro, Timmins

and O’Neil (2014) demonstrate that the model used in these papers to identify the sorting process

is fundamentally unidentified; imposing structure on the model to achieve identification, they

find strong evidence in favor of the sorting hypyothesis.

We are not aware of any existing analyses of shale gas development from the perspective

of environmental justice. Our paper seeks to document the existence of correlation between race,

income and lease terms that may be conducive to exposure to environmental harms. Our

analysis of linguistic isolation explains the mechanism behind observed correlations and has an

environmental justice interpretation.

2.3 Analysis of Shale Gas

2.3.1 Housing Values

Hedonic models describe how homebuyers choose houses based on property and

neighborhood characteristics. That choice process provides a theoretical construct with which to

connect observed market outcomes to individual preferences, facilitating measurement of

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!2!Other siting papers to discuss: Arora & Cason (1999), Wolverton (2010). Other sorting papers: Been and Gupta (1997).!

! 11

welfare effects. Measuring the impacts of shale gas activity on property values is therefore one

way to quantify its welfare effects. There has been limited prior research into how gas drilling

affects nearby property values. A few notable exceptions include Boxall et al. (2005), who

focused on sour gas wells in Alberta, and Klaiber and Gopalakrishnan (2012), who measured the

temporal impact of shale gas wells in Washington County, Pennsylvania. Muehlenbaches et al.

(2014) use data from all of Pennsylvania and the southern tier of New York to conduct a triple-

difference analysis of the effect of shale gas development on groundwater dependent homes,

along with a double-difference analysis of the effect on all nearby homes regardless of water

source. While that paper finds some evidence of small gains for houses dependent upon public

water sources (likely from lease payments) it finds evidence of significant negative net effects on

groundwater dependent houses. We find similar evidence here using data from Tarrant Co.,

Texas, in support of our creation of a lease “quality” index. Other research has found similar

evidence of concerns over risks to a household’s water source (Throupe, Simons and Mao

2013).3

2.3.2 Employment Effects/ Dutch Disease

Aside from lease payments, one of the main benefits to local communities of shale gas

development cited by its proponents is increased employment. A number of papers have

examined the role of shale gas (or fossil fuel resources more generally) in promoting local

employment. Black, McKinnish and Sanders (2005) look at local shocks associated with a series

of oil shocks driven by political turmoil in the Middle East in the 1970’s. These shocks drove up !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!3!Other shale gas hedonics papers to include: James and James (2014).

!

! 12

the price of coal in the US, and created a positive economic shock in areas with coal resources to

exploit. By 1983, prices had dropped and the boom to these communities turned into a

bust. Since these shocks came from outside the local community, it is reasonable to think that

they would be uncorrelated with other local unobservable shocks, and therefore provides an

exogenous source of variation with which to identify the ripple effects in local traded and non-

traded sectors.

Marchand (2012) examines the differential growth in employment and earnings between

local labor markets with and without energy resources in Western Canada. Focusing on periods

of boom (1971-1981, 1996-2006) and bust (1981-1991) periods in energy markets, he finds

significant evidence of impacts on employment and earnings in energy sectors and modest

impacts in non-energy sectors (particularly construction, retail trade, and services) during boom

periods.

Fetzer (2014) uses variation in where exploitable shale deposits are to identify impacts of

shale gas development on employment in different sectors. Despite rising labor costs, he does

not find evidence of Dutch disease (i.e., contraction) in the tradable goods sector, while the non-

tradable goods sector does contract. He attributes this result to cheaper energy providing a

source of local comparative advantage in the tradable sector, and backs-up this explanation by

looking at pipeline-induced supply constraints that create regional variation in prices. Allcott

and Keniston (2014) also look for evidence of the Dutch disease as a result of resource booms

using an empirical strategy similar to that used by Bartik (1991). Their results indicate that a

resource boom that doubles national employment in oil and gas will increase total employment in

a county with one standard deviation larger oil and gas endowment by 3.5 percent. Wages also

rise, suggesting the possibility for Dutch disease. There is, however, no such evidence --

! 13

manufacturing employment, revenues, number of establishments and capital investments turn out

to be pro-cyclical with oil and gas.4

2.3.3 Health

Many of the most important ramifications of the way in which lease outcomes are

negotiated have to do with health impacts. To date, there have only been a handful of studies of

the health impacts of shale gas development. Colborn et al. (2011) details potential health

impacts to the brain & nervous system, kidneys, endocrine system, and the immune and

cardiovascular systems, as well as increased risk of cancer or mutation. McKenzie et al. (2012)

compares cancer risks from air emissions inside and outside of a ½ mile buffer around wells in

Colorado, with a focus on benzene. Bamberger and Oswald (2012) use case-study interviews to

analyze health impacts on humans and animals living near shale gas development, finding

common reports of respiratory, gastrointestinal, dermatological and neurological problems.

Hill (2013) analyzes birth outcomes to mothers of singleton infants located in close

proximity to wells in Pennsylvania between 2003 and 2010. Using a differences-in-differences

strategy for identification, she finds dramatic impacts for mothers who reside within 2.5km of a

well (relative to those in a control group consisting of mothers in the 2.5-5km distance range).

Specifically, she finds an increased risk of low birth weight (+1.36 percentage points) and

APGAR score less than 8 (+2.51 percentage points). An examination of household water source

suggests that these impacts are driven by air pollution or stress from increased local activity (e.g.,

noise and light pollution), and falsification tests rule out alternative explanations besides drilling.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!4!Other employment papers to include: Jacobsen and Parker (2013), Mastromonaco and Maniloff (2014).

! 14

3. LEGAL AND INSTITUTIONAL DETAIL

The following section describes the technological developments driving the natural gas

industry’s increased productivity and the institutional details behind the regulation of the

industry. Over the past 20 years, horizontal drilling technology has evolved to allow access to

natural gas contained in tight-shale formations spread over a large area while requiring fewer

well pads. Further, these technological developments have increased activity in urban plays,

literally bringing drilling into suburban households’ back-yards. Regulation guiding industry

practices, however, has been largely crafted for activity in less densely populated areas – the

more common setting for natural resource extraction. We describe the technological

innovations and regulatory structure relevant to our story of Coasian bargaining and

environmental justice.

3.1 Hydraulic Fracturing and Horizontal Drilling

The process of hydraulic fracturing enables firms to extract natural gas from tight shale

formations by artificially stimulating the strata. This increases the flow of natural gas within the

shale, resulting in its eventual release and collection at the wellhead. Horizontal drilling

techniques allow firms to drill wells accessing minerals located within a large radius surrounding

the wellhead requiring fewer drill sites to reach a larger subsurface area and better access broad

resource deposits. Horizontal drilling therefore allows firms to extract large quantities of natural

gas from a smaller surface footprint, facilitating extraction from areas of higher population

density. Unlike the situation under conventional drilling, individuals in (sub)urban areas have

subsequently found themselves to be parties to negotiations with operators over mineral rights

! 15

leases.

3.2 The Texas Railroad Commission and State-Level Regulation

The Texas Railroad Commission (TRC) oversees the majority of the oil and natural gas

industry in the state of Texas, which includes the approval of permits to drill wells.5 However,

prior to permit approval, firms must first amass a large and sufficient mineral estate acreage that

is spaced far enough away from existing well infrastructure to be approved and permitted by the

Texas Railroad Commission (TRC), the state-level regulatory body.6 Obtaining the minimum

amount of acreage is achieved by signing a series of leases with mineral rights owners.

Households signing leases with natural gas firms or ‘landmen’ are tasked with weighing the

trade-offs between future income derived from natural gas extraction from their mineral estate

with that of the potential known and unknown risks of living near an active well. Once a well is

permitted, the operator has two years to begin drilling the well before the permit expires.

The TRC’s jurisdiction regulating the industry extends to the drilling and production

phases; however, there they do not regulate noise, traffic, or well pad appearance, nor do they

require air pollution testing. By law, operators have access to surface water to be used to treat

the well, and chemical disclosure is restricted to only the non-proprietary chemicals used to

fracture a well. In general, the dis-amenities experienced by households from nearby shale gas

activities are unregulated by state and federal entities and are folded into the private leases

signed between them and firms.

More specifically, federal and state regulators generally do not have direct jurisdiction

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!5!The Texas Railroad Commission has jurisdiction over the “exploration, production, and transportation of oil and gas prior to refining or end use,” and the TRC executes its jurisdiction by enforcing rules written in the Texas Administrative Code, Chapter 3. !6 Texas Administrative Code, Chapter 3, Rules 37 & 38.

! 16

over the private contracts drawn between landowners and parties interested in leasing land for

exploration and production of oil and natural gas. Higher-level regulation is limited to royalty

payments stipulating when they are to be paid, the required information that must be provided

(and that which can be requested) by firms, notification upon re-assignment of leased rights, and

determining the consequences of delinquent payments. In addition, the TRC has jurisdiction

over enforcing and undertaking remediation from undue negligence on the part of firms and

broadly enforcing the protection of ground and surface water from contamination caused by the

industry. However, the TRC’s jurisdiction over the leases signed between households and firms,

and subsequently, the protection of households while a well is drilled and after production ends

is limited, and a well-informed household may negotiate more comprehensive contracts with

leasing firms to protect their interests beyond the minimal coverage of the law.

3.3 Municipal Regulation

In Texas, cities practicing `home rule' have written and passed sets of ordinances that

restrict how firms operate within the jurisdiction by forcing them to meet additional technical

standards; shielding residences from well pads by increasing the set-back requirements or

requiring specific landscaping; or increasing environmental or traffic testing required to be

approved for a permit to drill a well, to name a few. Local ordinances then supersede state-level

restrictions given that they meet the minimum state-issued baseline. While this paper is not

focused on the municipal ordinances, the model specifications often control for this layer of

regulation by estimating city-level fixed effects.

! 17

3.4 “Private” Regulation – The Lease Negotiation Process

The following sub-sections detail aspects of property law relevant to the setting studied in

this paper, including the purpose and content of the lease negotiations, the severed estate, and

overall rights of property owners, which is especially relevant when the estate is split.

Signing a comprehensive leasing document is important for households protecting their

rights while they are royalty interest holders.7 In particular, lease terms can compensate for the

absence of state or municipal regulations. These leasing agreements contain a set of basic terms

common to all leases drawn in the industry and sets of auxiliary terms that are negotiable

between lessors (mineral rights owners) and lessees (exploration and oil and natural gas

production firms). Basic terms include a careful description of the minerals leased to the lessor;

information about the royalty payments owed to the lessor, or mineral rights owner, once the

well begins producing in paying quantity; the duration of the lease, or primary term; and

opportunities for extension once the primary term has expired.

Auxiliary terms include clauses written into the agreement protecting one or both of the

parties, which may not be included in all leases. Negotiators may draft surface damage clauses

ensuring that the operator restores the surface to a condition agreed upon by both negotiators

once production is complete; environmental clauses requiring producers to perform regular

environmental quality tests on the surface and ground water or soil samples; and pooling

restrictions ensuring that the leased land value is not diluted in terms of royalty payments by

being grouped into an unnecessarily large acreage, to name a few.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!7!The risks are less when the property lies in a municipality having passed local ordinances written to protect lessors’ rights.

! 18

3.5 Severed Estate

Up to this point, we have assumed the signer of the lease is the household, or surface-

rights owner; however, the state of Texas allows the mineral estate to be severed from the

surface estate so the individuals signing a lease with a natural gas firm may not be the

individuals living in the house positioned on the surface estate. As early as 1953, Texas courts

declared that landowners may reserve mineral rights and the oil and gas contained as in the case

Benge v. Scharbauer [259 S.W.2d 166 (Tex. 1953)], thereby enabling the mineral estate to be

severed from the surface estate (Merrill, 19).8 In the event of severance, the mineral estate

dominates in terms of exploration and extraction, and the mineral lessee assumes the same rights

owed to the mineral estate owner since the leasing document is perceived as a temporary

transference of ownership.9 Colloquially, the mineral estate may lease the minerals to third

parties for exploration and law only requires the surface owners be notified of the ‘intent to

explore and drill;’ they have access to as much land as is necessary to explore and drill; they may

remove trees and fences to make way for well and equipment, and the well pad itself can take up

to one acre of land; and they may erect pipelines to transport the natural gas off the property

(Rahm, 2979).10,11

As an independent entity, the mineral estate may exercise its rights without consulting the

surface estate owners. Subsequently, a firm leasing the mineral rights for purposes of oil and gas

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!8 A grant or reservation of minerals by the fee owner affects a horizontal severance and the creation of two separate and distinct estates: an estate in the surface and an estate in the minerals [Acker v. Guinn, 464 S.W.2d 348, 352 (Tex. 1971)] (Fields,1). 9 If the minerals are not reserved at the sale date, the mineral estate automatically goes to the buyer along with the surface conveyance (Fambrough, 4). 10!Under the dominance of the mineral estate, there are five interests including the right to develop the mineral estate, or ingress and egress, which includes exploration; to lease; and to receive bonus payments, delay rentals and royalty payments (Vanham, 233).!11!There!are!three exceptions to the dominant mineral estate including excessive use of land in exploration and operation activities to access the minerals, unnecessarily injuring the surface, and not accommodating the existing surface use, the latter more formally entitled the Accommodation Doctrine (Letter of the Law, 1997).

! 19

exploration and extraction need only negotiate with the mineral estate owners whether they also

own the surface estate or not. The owners of the mineral estate may exercise their rights without

consulting the owners of the surface estate, and they are only required to inform the surface

estate when drilling is imminent on their property due to legislation passed in 2007 (Maxwell,

347).12 Additionally, the mineral estate may use water from the leased land to carry out

operations given use is not wasteful and inject wastewater into sub-surface formations.13 They

are able to use as much surface as is reasonably necessary to access the mineral estate (Warren

Petroleum Corp. v. Martin, 271 S.W.2d 410 (Tex. 1954)) while not accepting responsibility for

the full restoration of the property (Warren Petroleum Corp. v. Monzingo, 304 S.W.2d 362 (Tex.

1957)), nor are they required to pay surface damages as long as the damage is not unreasonable.

3.6 Rights of Surface Owners

Texas has not passed a law protecting the surface estate, or a surface damage act, as has

passed in other states with prominent oil and natural gas industries (including New Mexico,

Oklahoma, North and South Dakotas, and Montana). As mentioned above, surface owners are

not owed any remuneration for the opportunity cost of the lost piece of their property during the

drilling period nor must they be paid for reasonable damages to the land caused by drilling. If

there is any perceived misuse of the land by mineral rights owners, surface owners are

responsible for proving unreasonable conduct, which does not include surface damage or

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!12 Texas Natural Resource Code, 91.703(a): Not later than the 15th business day after the date the commission issues an oil or gas well operator a permit to drill a new oil or gas well or to reenter a plugged and abandoned oil or gas well, the operator shall give written notice of the issuance of the permit to the surface owner of the tract of land on which the well is located or is proposed to be located. 13 Unless specified in the deed, the water rights fall to the surface owner but they are accessible with reasonable use by the mineral estate (Vanham, 238).

! 20

inconvenience. Surface owners are marginally protected by the Accommodation Doctrine,14

which protects existing surface owner uses.

In lieu of state regulation, and in some cases, local ordinances, lessors can negotiate a

surface damage clause into the leasing agreement to protect the surface during production and

ensure remediation after production ends. Otherwise, the surface estate may only claim and

prove unnecessary damage through litigation. However, since the state of Texas does not require

operators to negotiate with the surface estate independent of the mineral estate, a severed estate

with a lease may not naturally include a surface damage clause. Property owners holding deeds

to both the surface and sub-surface can negotiate surface damage clauses which might include

complete restoration of the surface upon completion of the well and perhaps assign a property

damage fee; however, among mineral estates, there is little incentive for a mineral estate owner

to negotiate a surface damage clause with a potential operator. In the latter case, among severed

estates, surface owners have little legal protection.

4. DATA

The analysis follows from a combination of lease, well activity, housing, and

demographic data sets to capture our specific question of environmental justice of leasing terms.

This section details the sources of data and, when relevant, the construction of certain variables.

The data originates from several sources and is compiled into data sets for analysis by a series of

matching techniques, primarily based on text strings.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!14!Accommodation Doctrine: [W]here there is an existing use by the surface owner which would otherwise be precluded or impaired, and where under the established practices in the industry there are alternatives available to the [mineral owner] whereby the minerals can be recovered, the rules of reasonable usage of the surface may require the adoption of an alternative by the [mineral owner]. (Tarrant County Water Control & Improvement Dist. No. 1 v. Haupt, Inc., 854 S.W.2d 909, 911 (Tex. 1993)) (Merrill, 26).!

! 21

Of primary importance is the well and lease data with which we describe the precise

amount of well exposure and the types of leases signed at the transaction and appraisal dates

observed in the housing data. Secondarily, we have collected relevant housing transaction and

appraisal data from the Tarrant County Appraiser District office and Dataquick. Demographic

data are merged from information collected under the Home Mortgage Disclosure Act; and

inspection and violation data are collected from the Texas Railroad Commission. Further, we

have generated a variable identifying leases likely signed by a mineral rights owner from a split

estate.

4.1 Well exposure

Well permitting data are used to construct well exposure from the perspective of the

household at the appraisal date. In particular, well permitting information comes from two

sources: the Texas Railroad Commission and Drilling Info, a proprietary aggregator of drilling

activity accessible through Duke University’s Energy Institute. Exposure is tabulated at different

distance buffers surrounding the house, and if an operator has begun drilling a well by the

appraisal date, the well is included in the exposure variable. We tabulate exposure based on 500,

750, 1000, and 2000-meter buffers, and the calculation differentiates between wells that are

permitted, spudded, and producing. The primary exposure variable used in the data is the count

of wells that have either spudded or are producing in a 2000-meter buffer. Table 1.1 summarizes

the well exposure over time by tabulating the average well exposure for a given household

observed in the data throughout the sample period using the 2000 meter buffer as the exposure

distance.

! 22

4.2 Lease Data

The lease and lease contents are a primary and unique source of data used to determine

the observed Coasian bargaining outcome in our setting. A lease is drawn between two parties –

the lessee (i.e., who is typically an operator or third party ‘landman’) and the lessor (i.e., the

owner of the royalty interest, who is typically the property owner in the case of non-split estates);

signing the lease conveys the interests of the mineral estate from the lessor to the lessee. We

have collected data describing the terms of these privately negotiated leasing contracts. In

particular, we have data describing the primary clauses of all natural gas leases negotiated in

Tarrant County Texas between 2000 and 2013. Tables 1.2 and 1.3 summarize these primary

clauses including the average royalty, acreage, term length in months, and frequency of auxiliary

clauses.

In addition to the primary clauses contained in the leasing agreements, we have also

collected auxiliary terms for one third of the sample and matched the two sets of leasing data.

The data period for auxiliary clauses with a large enough sample size begins in 2006 and finishes

in 2011. Our specific sample was collected from the “Drilling Down Series”15 published by the

New York Times from 2011 into 2012. We were able to scrape these data and mine the files for

words and phrases indicating the existence of certain clauses using computer software. We

define 12 auxiliary lease clause categories. The comprehensive list of categories, variable names

used in the analysis, and category descriptions are included in the appendix.

For each of the auxiliary clause categories, we search a set of roughly 90,000 text

documents that were converted from the original PDF scans using OCR software for a set of

identifying words using software written in Python. We constructed a list of patterns for each

clause that were used to search and identify leases containing those clauses, and this resulted in !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!15 http://www.nytimes.com/interactive/us/DRILLING\_DOWN\_SERIES.html

! 23

sets of text files containing the parsed text from leases when the program successfully matched a

pattern. The use of regular expression functions allows us to search for fragments of words, and

account for miss-spellings and superfluous punctuation that might prevent a perfect match of

phrase. Perfect matches are particularly difficult in converted documents (PDF to text) because

the conversion is imperfect, resulting in miss-placed spaces, characters, numbers, and letters.

For example, below we describe surface damage and restricted surface access clauses; several

patterns were employed to extract words used to communicate these added restrictions including

the following:

1. No surface use a. No surface use: r'[rmn][ao0][\s+]*su[rmn]f[ao0][\w+]*[\s+]*use',\ b. No surface operations: r'[rmn][ao0][\s+]*[\w+]*[\s+]*[\\~|>/!\%_,.-

`"\'()$?@#\d:;]*[\s+]*su[rmn]f[ao0][\w+]*[\s+]*[ao0]pe[rmn][ao0]t\w+' c. Lessee shall not conduct any surface operations:

r'[l1]es[s]*[\w+]*[\s+]*sh[ao0][l1][l1]*[\s+]*[rmn][ao0]t[\s+]*c[ao0][rmn]d\w+[\s+]*[\w+]*[\s+]*[\\~|>/!\%_,.-`"\'()$?@#\d:;]*[\s+]*su[rmn]f[ao0][\w+]*[\s+]*[ao0]pe[rmn][ao0]t\w+'

d. Lessee shall not enter upon w surface: r'[l1]es[s]*[\w+]*[\s+]*sh[ao0][l1][l1]*[\s+]*[rmn][ao0]t[\s+]*e[rmn]te[rmn][\s+]*up[ao0][rmn][\s+]*[\w+]*[\s+]*[\\~|>/!\%_,.-`"\'()$?@#\d:;]*[\s+]*su[rmn]f[ao0][\w+]*'

e. Within (d) feet w w land (no surface use at all): r'withi[rmn][\s+]*\w+[\s+]*[({\[|)}\]]*[\\~|>/!\%_,.-`"\'$?@#]*[\d+]*[\\~|>/!\%_,.-`"\'()$?@#]*[\s+]*[\d+]*[\\~|>/!\%_,.-`"\'$?@#]*[\s+]*[\d+]*[({\[|)}\]]*[\s+]*fe[e]*t[\s+]*[\w+]*[\s+]*[\w+]*[\s+]*[l1][ao0][rmn]d',\

2. Damage and sound externalities

a. Lessee shall pay for damage: r'[l1]es[s]*[\w+]*[\s+]*sh[ao0][l1][l1]*[\s+]*p[ao0][v\\/y][\s+]*[\w+]*[\s+]*d[ao0][rmn][ao0]g[\w+]*'

The numerical description is the clause type and the alpha-description is the phrase or pattern

used in the text search to identify leases containing the identified clause. Following the pattern is

the regular expression used by Python to search for the phrase accounting for the listed

! 24

idiosyncrasies of the converted text files (as much as possible). After the initial extraction, the

data was cleaned further using regular expression functions and quantified into the binary form

used for analysis using STATA. Finally, the auxiliary clauses were matched to the observational

lease data using a record number assigned by the county clerk office.

In general, the auxiliary terms of leases fall into several categories including strict legal

requirements and cleaner definitions of liability across the two private parties; additional

environmental compliance; requirements for increased reporting by the lessee to the lessor

regarding well activity; surface damage protection; and restrictions on how a firm can access the

mineral estate. A clause of particular import in the state of Texas is the surface damage clause,

which we capture by searching for phrases describing cleanup efforts and damage remediation.

Surface owners are not owed any remuneration for the opportunity cost of the lost piece of their

property during the drilling period or for reasonable damages to the land caused by drilling. If

there is any perceived misuse of the land by mineral rights owners, surface owners are

responsible for proving unreasonable conduct that does not include surface damage or

inconvenience.16 Mineral rights owners can negotiate a surface damage clause into the leasing

agreement to protect the surface and use during production and ensure remediation after

production ends.

Mineral rights owners may also restrict how the firm accesses the natural gas, otherwise,

as the dominant estate, firms are legally allowed to access it through the surface estate as long as

it is not unreasonable. Restricting how the firm accesses the minerals is often written into the

lease as either no surface access or only pooling. No surface access explicitly prohibits boring a

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!16!Surface owners are marginally protected by the Accommodation Doctrine, which protects existing surface owner use. If there exist alternative extraction methods, then reasonable usage might require a change on the part of the lessee under the Accommodation Doctrine (Vanham, 240).!

! 25

hole in the surface estate to access the mineral estate while only pool states that to extract natural

gas from the mineral estate, the plot must be pooled with other properties and accessed from

another plot of land.

4.3 Housing Data

In addition to the leasing and regulatory data, we use appraiser data describing the

household attributes compiled from the Tarrant County Appraiser District (TAD) office and

Dataquick to form observations describing housing attributes, lease characteristics (if one has

been signed), and demographics of the inhabitants. TAD provided us historical appraisal data

from 2008 to the present in addition to a file delineating all transactions in Tarrant County. We

used this data to construct a cross section of appraised values for each property that is observed

being bought or sold. We then merged the TAD to Dataquick, which serves as our source of

loan information as it provides the lender name, loan amount, and other attributes of the loan.

By merging appraisal data with transaction information from Dataquick, we are also able to

incorporate demographic information of the house inhabitants at each appraisal date using the

Home Mortgage Disclosure Act data, a match that will be described in more detail below.

The housing data is the nexus for several connections between our data sources. In

particular, leases are merged to the housing data by address using various string matching

methods, and the housing data is merged to the HMDA data by lender name, loan amount, and a

geographic identifier. Table 1.4 summarizes the count of each type of transaction by year

including the distribution of the appraised, transaction, and lease observations across years.

! 26

4.4 String Matching

The address merge between leases and housing data (and between the two housing data

sets, TAD and Dataquick) is executed in a series of steps that separate addresses into parallel

pieces and controls for miss-spellings. The lease data is particularly dirty in its raw state in terms

of all variables generated using strings; these are the variables used to match leases to other data

sets for this analysis. The first step of the address merge is to parse both sets of data into the

address categories like a house number, suite or apartment number, street, street type, city, state,

and zip code.

The second step corrects miss-spelled words in the street city names. We constructed a

user-defined function that embeds a function, strgroup, designed by Julian Reif at the University

of Chicago. This function calculates the Levenshtein distance between all of the strings being

fed to the function, and normalizes by the length, or “edit distance”, of the smallest string in the

group. If the normalized distance is less than a specified threshold, the strings are grouped

together and output into a new group variable. The Levenshtein distances uses an algorithm to

determine a string distance based on the number of changes necessary to turn the first word into

the second. For example, “lessor” and “lessee” would have a Levenshtein distance of two as one

needs to change the “or” to “ee” using two steps such that words become identical. Our function

assumes that the “correct” spelling is the spelling used most frequently across both data sets,

applies strgroup iteratively, and assigns the correct spelling to all miss-spelled words. In the

end, we assemble a list of acceptable spellings of cities and streets that are then merged back to

the original data set and used to match the lease addresses to the housing data.

The address match is performed roughly fifty times using different combinations of

address variables that differ in the restrictiveness of the match. Using the described methods, we

! 27

merge roughly 73 percent of our leases to specific properties based on the provided grantor

addresses.

4.5 Home Mortgage Disclosure Act (HMDA)

To account for household demographic information in the regressions, including

household income, ethnicity, and race, we merge the housing data to data collected in accordance

with the Home Mortgage Disclosure Act (HMDA), legislation originally passed in 1975 and

amended in 1989 to ensure fair and adequate lending to nearby neighborhoods in which the

lenders are located and limit discriminatory lending practices. We merge HMDA to the housing

data by matching the tract, lender name, and lent amount across HMDA and Dataquick, our

source of property-level financing.

4.6 Violations Data

The final external source of data describes the well-level violations reported by the Texas

Railroad Commission and dockets, both past and present, reviewed by the state regulator. The

violation data is identified by the rule violated in Chapter 3 of the Texas Administrative Code, or

the state-level regulations enforced by the TRC. The current analysis only uses the counts of

violations and does not yet distinguish the violation by rule. None of these violations are a

response to complaints from citizens, but rather, they are the result of well inspections performed

by TRC staff.

The docket data comes from a mainframe database maintained by TRC tracking all cases

by permit, well, and lease, depending on the circumstances. The data includes all relevant dates,

which might include the date the case was initially filed, any hearing dates, when a settlement is

! 28

proposed, the date of the counter-offer, and the date when the case is closed, if relevant.

Additionally, the database describes the type of case, operators involved, and settlement values.

The data we have is for the state of Texas; however, we are only using the Tarrant County

portion and it is merged to our other well data by lease name and well number.

In the analysis, we tabulate the number of violations and relevant docket types for all

wells within the 500, 750, 1000, and 2000-meter distance buffers at the appraisal date. The

relevant docket types used in our analysis include: general enforcement issues, reimbursement to

the TRC, and proper well plugging.

4.7 Generation of Split Estate Identifier

We generate a variable describing whether each house is appraised after the mineral

estate is likely severed from the surface estate. This is an important variable to use as a control,

because mineral rights owners signing leases may attribute a different value to specific lease

clauses than those households signing a lease as a whole estate. For example, the lessor in the

case of a split estate may not value a surface damage clause as much as she would if she also

owned the surface rights and cared about the potential loss in property value attributed to well

activity.

Split estates are not directly identified in our data set. Rather, we determine what is a

likely split estate based on a series of matches that eventually compare the names of mineral

rights owners signing leases with those of the individuals buying and selling properties in our

housing data sets. We consider two cases as evidence of a split estate – (i) the lease is matched

to a property via address or another geographic identifier but the name of the signer does not

match that of the buyer or seller of the property, and (ii) the lease is matched to the property via

! 29

buyer or seller names, but the date the lease was signed is not consistent with when the house

was sold. The inconsistency arise when, for example, the transaction date of the house proceeds

the date the lease was signed, but the name on the lease matches the seller of the house.

Upon matching as many leases to properties as possible, we then merge all of the lease

records and property identification numbers back to a list of all buyers and sellers associated

with each house and at each date of transaction to determine if and when names match between

the housing and lease data. We first look for perfect matches between names of individuals

signing leases and buying or selling houses. We then proceed to identify close spellings using

the Levenshtein string distance measure described above in section 4.3. Using this function, we

can find those names that are nearly the same and differ likely from data entry errors across data

sets.

After identifying the name matches, we can then use the transaction and lease date

comparisons to finalize our split estate approximation. The intuition of this final step is that

sellers signing leases at a date after the transaction date are likely to have split the estate when

they sold it to the new buyers, otherwise the name on the lease would match that of the buyers.

4.8 Description of Samples

We begin the analysis with 134,611 housing observations matched to HMDA composed

of 77,359 unique households. Of the 134,611 observations, we only keep those houses that

signed a lease at any point in the data, or 70,887 observations.

For the purpose of our analysis using demographic information, it is important to know

when the lease was signed with respect to the transaction date. Since HMDA is matched for the

buyer using the loan information, any analysis using lease and demographic information together

! 30

requires us restrict the sample to those leases signed after the transaction date to ensure we are

measuring the correct information. Of the total number of leases signed in our data, 23,755 are

signed before the transaction dates observed in the data while 47,132 are signed after.

We further restrict our sample by those leases reporting positive royalty values, as those

not reporting are missing, and of the total, 37,452 observations have positive royalty values.

Hedonics analysis uses the 37,452 observations that have positive royalty values though we

compare the results to those from the full sample of houses with and without signed leases (and

positive royalty values) and observe very similar effects.

In the previous sub-section, we describe a method for identifying split estates. This is

particularly important in a hedonics exercise where the terms of a lease may be valued

differently depending upon whether an individual owns mineral and surface rights, or just

surface rights. One significant difference is the differential risks of pollution, air and water

quality, or the disamenities from well noise, traffic, or worse aesthetics. Of the sample with

positive royalty values, 35,240 are identified as non-split estate households and 2,212 as split

estates.

Parsing the split estate sample further between those leases signed before and after the

transaction dates, we observe the following:

1. If the lease is not active (signed after the transaction date)

a. 22,329 non-split estate households

b. 1,448 split estate households

2. If the lease is active (signed before the transaction date)

a. 12,911 non-split estate households

! 31

b. 764 split estate households

Lease clause analysis involving demographic characteristics use only those in the non-active

category of leases so that we are certain to have captured the correct demographics of the

household at the date the lease was signed.

Table 1.5 compares the summary statistics across the full sample of houses and the lease

sample, or those households having signed a lease with a known royalty value used for the

analysis. The biggest differences occur comparing the frequency of subdivision sales and active

leases, the house age at the appraisal date, and across the reading scores.

5. THEORY: COASIAN BARGAINING

We motivate our empirical specifications with a brief discussion of the Coase Theorem

and Coasian bargaining. The simplest version of the Coase Theorem states that, in the absence

of prohibitive transaction costs and with well-defined property rights, parties will negotiate to an

efficient equilibrium in the presence of an externality. Efficiency is defined either in marginal

terms (i.e., the cost of an additional unit of the externality borne by the victim is equal to the

benefit of that unit of externality to the polluter) or in discrete terms (i.e., property rights will be

allocated so that the resource in question is put into its highest value use).

In negotiations for the rights to shale resources, property rights are well-defined and lie

with the mineral rights holder. In the case of a non-split (i.e., “fee simple”) estate, the owner of

the mineral rights is the same as the owner of the surface rights associated with the property. In

the case of a split estate, the mineral rights and surface rights holders are different individuals.

In terms of transaction costs, mineral leases are legally binding agreements between a lessor and

! 32

a lessee that are enforced by U.S. courts. Many lease components are standardized. While

lawyers may be involved in the negotiation process (helping to mitigate problems of asymmetric

information), transaction costs are likely to be low. As such, this is an area where we might

expect the Coase Theorem to apply.

We motivate our empirical model with a simple theoretical construct. First, define all of

the disamenities associated with proximity to drilling activity by X; this could include (but is not

limited to) air, light, noise, and water pollution (or the risk thereof), surface damage and

alteration of the terrain. These are all the outcomes of activities that may be regulated by

municipal ordinance, but which can be privately “regulated” by clauses negotiated into leases.

Compliance with those lease clauses is costly for the driller. As such, the driller receives a

positive (but decreasing) marginal benefit from additional units of X.

We consider a simple case in which the landowner (who we assume for now is also the

mineral rights holder) has a constant marginal cost associated with each additional unit of X. In

the absence of any bargain being struck, property rights resting with the landowner would insure

that X = 0. However, for each unit of X up to X*, the marginal benefit to the driller exceeds the

cost to the landowner, so compensation is possible with positive remaining surplus. Beyond X*,

additional units of X provide the driller with benefits that are smaller than the compensation

required by the landowner. As such the driller voluntary restricts the externality to X*, paying

the rectangular area in compensation to the landowner. The triangle lying above that rectangle

represents Coasian surplus, which is divided between the landowner and the driller according to

their respective bargaining powers.

! 33

The graph becomes slightly more complicated when we introduce the revenues received

by the driller when the resource is sold, regardless of the level of X determined in the Coasian

bargaining process. This is not easily shown in the marginal graph, so we describe the

equilibrium in terms of total costs and benefits instead. The total benefit of X for the driller,

denoted by TB(X), begins at a height that reflects those revenues. Despite that complication, the

same bargaining principles described in the previous graph apply here. In particular, confronted

with a property-rights-holding landowner with total costs TC0 ( X ) , the driller would find it

optimal to set the externality to X0* . The Coasian bargaining surplus, which would then be

divided between the driller and the landowner, is given by the line segment (ab). If the driller

were instead confronted by a landowner with total costs given by TC1(X), it would instead be

optimal to set the externality to X1* , with the line segment (cd) representing Coasian bargaining

surplus.

X

MC(X)

MB(X)X*

! 34

Under efficient Coasian bargaining, X* is chosen to maximize the size of the bargaining surplus.

A separate negotiation takes place over the bargaining surplus, which typically involves the

payment of some percentage of the revenues collected from gas sales in the form of a royalty.

There are a number of variables that we would expect to be important determinants of the

slope of TC( X ) and which should, therefore, affect the equilibrium value of X. These are

variables that affect willingness-to-pay (WTP) to avoid additional pollution exposure. Hamilton

(1995) refers specifically to income and education. We also include the presence of children or

elderly in the household and a dummy variable indicating that the house relies upon

groundwater.

Variables like income, education, and family structure and water source might also

influence bargaining power, which determines how the Coasian surplus is divided between the

driller and the lessor in the setting of the royalty rate. We might also expect variables such as

TB(X)

TC (X)

TC (X)

0

1

XX X* *0 1

a

b

c

d

! 35

race and linguistic isolation to have an effect on bargaining power and royalty. It is, however,

difficult to think of a reason why, after controlling for income, education and family structure,

race and linguistic isolation should have an independent effect on TC(X); finding evidence that it

does suggests we are not in an efficient Coasian outcome (i.e., X is being determined by factors

aside from marginal costs and benefits). We test this hypothesis explicitly in our empirical

specification. Previewing our results, we indeed find evidence against efficient Coasian

bargaining, which we explain by appealing to asymmetric information, proxied by linguistic

isolation.

6. Model

6.1 Analysis of Lease Terms

We conduct our analysis of lease terms in two stages. The first treats each lease clause as

a separate bargaining outcome, but allows for the fact that they were negotiated in a single

bargaining process. As such, we treat the equations describing the outcomes of each lease clause

as a seemingly unrelated system of regressions. Results are described in Table 2. At this stage

of the analysis, we limit our analysis to the sample of only lease signers who own their full estate

(i.e., mineral and surface rights). Dependent variables are listed in the first column, and

regressors appear in subsequent columns. Further, we have sub-divided the parameters for the

regressors into race (top) and Coasian (bottom). Regression coefficients describe the likelihood

of each lease clause as a function of attributes of individuals who sign leases, tract-level control

variables, and city-level fixed effects. Lease clauses are generally defined to be “goods”, in that

having them in the lease should benefit the lessor, ceteris paribus. The two exceptions are lease

! 36

term, allowing the lessee a longer period of time before having to make the well productive

without losing mineral rights, and subsurface easement, enabling firms the ability to lay pipelines

and roads as necessary for transporting natural gas from the wellhead.

Increased household income is positive predictor of most lease clauses, including legal

clauses, reporting requirements, and clauses related to the cleanup of surface damage, noise

restrictions, environmental clauses, and limited surface access. Higher incomes tend to reduce

lease terms (i.e., allowing a shorter period of time before a well must be productive).

In addition to household income, we include several tract-level variables common to

Coasian analyses that are focused on describing potential trade-offs of negotiating households

increasing or decreasing their willingness-to-pay to avoid negative externalities. We include the

fraction of households with inhabitants older than 65 and younger than 18 as we might believe

those households are more cautious, having a greater willingness-to-pay to accept negative

externalities. Additionally, we control for whether the households depends on groundwater

thereby making them more susceptible to water contamination risks. The tract-level education,

household-level plot size, and whether the most recent transaction was a subdivision sale are

added as additional controls. Larger plot sizes are believed to be more valuable to firms, and it

has been anecdotally reported that many new developers in Texas keep the mineral rights to their

developments which might increase developers’ bargaining power through lease negotiation over

a larger plot size.

While the variables discussed above have a primary interpretation as determinants of

willingness-to-pay to avoid externalities from shale gas development, race variables do not. If

willingness-to-pay is indeed not a function of race after controlling for the variables described

above, then evidence that it matters for negotiations over lease clauses would suggest inefficient

! 37

Coasian bargaining. This is, in fact, what we find. In particular, we find that Hispanics are

likely to have far longer lease terms and third party negotiations, and they are less likely to have

legal protections like force majeure and the Pugh clause, and are less likely to have additional

reporting requirements. While they are more likely to have a noise restriction, they are less

likely to have surface damage or environmental clauses, or clauses limiting surface access.

There are similar effects observed for the other minority group that we might consider to

be disadvantaged in a lease bargaining process – blacks. They exhibit significantly lower

likelihoods of having a Pugh clause, surface damage or environmental clauses, and a higher

likelihood of a longer lease term

We examine one possible mechanism behind this race effect – asymmetric information.

In particular, of the two disadvantaged minority groups in Texas, Hispanic disadvantage may

derive from linguistic isolation. Formally, the Census Bureau defines a linguistically isolated

household to be one in which no person over the age of 14 speaks English at least very well. In

our data, we proxy for this situation with a measure of the percentage of English speaking

households in the census tract where the household lives, taken from the American Community

Survey 2008-2012 data. To test for this possibility, we interact the Hispanic race dummy with

the percentage of English speaking households from that household’s census tract. A value of

100 would indicate that the entire tract is comprised of English speaking households (ruling out

the possibility that the household in question is linguistically isolated), while a value of 0 would

indicate that none of it is (guaranteeing linguistic isolation). We would expect that linguistic

isolation would put a household at a severe disadvantage in lease negotiation. Interestingly, if

we can rule out linguistic isolation (i.e., a 100% English speaking census tract), Hispanic race is

no longer a disadvantage in terms of lease term, Pugh clause, reporting requirement, surface

! 38

damage, environmental clause, or restrictions on surface access. This result speaks to a

mechanism based on asymmetric information in lease bargaining and points most directly to a

breakdown in efficient Coasian bargaining.

Turning to the covariance matrix describing the residuals from the SUR procedure (Table

2.2), there are a number of notable features. Generally, all of the larger values are positive. This

indicates idiosyncratic bargaining ability (i.e., an individual who negotiates for one good clause

is likely to negotiate for other good clauses as well). We see this, for instance, in the case of

Pugh clauses and force majeure, noise requirements and legal fees, subsurface easement and

post-production cost, and environmental clauses and reporting requirements. There are very few

matrix elements with negative values bigger than -0.1, suggesting that individuals are not

systematically trading off one type of term for another.

6.2 Hedonic Estimation

Leases are multidimensional assets – in particular, they comprise a variety of monetary

payoffs (both today and in the future) accompanied by risks (both financial and environmental)

and external costs to be born by the owner of the surface rights (who may also be the lessor). In

the previous sub-section, we examined a number of these lease terms individually. In this sub-

section, we use implicit prices derived from a hedonic property value regression to generate a

“lease quality index”. In the following sub-sections, we explore how that index varies with the

attributes of the homeowners signing the leases.

Hedonic results are reported in Table 3, and they include hedonic regressions using a

subset of the data that is observed signing a lease and reports a positive royalty value (columns

1-3) followed by a second set of results using the full set of observations (columns 4-6). To our

! 39

knowledge, these are the first set of property value hedonic results combining exposure to shale

gas development with details of the lease attributes that determine how exposure affects houses.

In addition to housing characteristics, well exposure and lease attributes, we also include

variables describing loan attributes (variable rate, FHA loan), school quality (measured by 2012

8th grade reading scores in the school district), and a vectors of city and year dummy variables.

With the exception of school quality, all housing attributes and most lease attributes have the

expected signs and reasonable magnitudes.

6.3 Principle Component Factors

We consider a large number of lease clauses, many of which are highly correlated with

one another (see Tables 1.6, 1.7, and 1.8). This can make it difficult to identify separate effects

of each clause. To reduce the size of the variable list, we utilize a principal component analysis

(PCA). PCA is a statistical technique for data reduction that converts a set of correlated

variables into a smaller number of linearly uncorrelated variables called principal components.

The transformation determines the first principal component sot that it accounts for as much of

the variability in the data as possible. Subsequent components are also designed to explain as

much of the variability in the data as possible under the constraint that each is orthogonal with its

preceding component. Table 4.1 describes the results of our principle component factor analysis

of lease clauses. Table 4.2 describes the results of a hedonic specification using PCA factors to

summarize lease terms.

6.4 Lease Quality Regressions

Using the results of the PCA, we can summarize the value attributed to a bundle of lease

attributes using the implicit prices ascribed to the factors found in the hedonic regression

! 40

described in Table 4.1.

Table 5.1 describes the results of the decomposition of the lease quality index over

attributes of lease-signing households and their neighborhoods. It is easy to see that the two

minority groups who we might consider “disadvantaged” (i.e., blacks and Hispanics) have

systematically lower quality leases, even after controlling for variables that we would expect to

determine willingness-to-pay. This effect is particularly strong for Hispanic residents.

We examine the asymmetric information mechanism in the context of the lease quality

index and find results similar to those found using the SUR. In particular, the Hispanic

disadvantage all but disappears if we consider Hispanics in 100% English speaking census tracts.

This further confirms our interpretation of the Hispanic race effect as a breakdown of efficient

Coasian bargaining.

6.5 Bonus Payments

Bonus payments are not observed for all leases, so we separate out the analysis to another

section. The summary statistics of bonuses observed in the data are reported in Table 1.2.

Because of the large number of missing observations, we estimate our model using a tobit. The

goal of the bonus payment analysis is to demonstrate that these payments are not being used as a

source of compensation that systematically offsets the worse lease terms received by poor and

minority households. Estimating a simple model with the limited sample of bonus payments that

we observe, we find that this is indeed the case.

More specifically, we estimate a tobit regression of the log of bonus payments on land

area (payments are usually defined on a per acre basis), our income and demographic

characteristics from HMDA, whether the lease was signed by a well operator (as opposed to a

! 41

landman), and city-level fixed effects for two of the specifications. The results are reported in

Table 6. Similar to lease terms, blacks and Hispanics are at a disadvantage in the negotiations of

bonus payments, and we observe an increase in bonus payments when the household is likely

English speaking as explained by the effect of the Hispanic and English speaking interaction

term. A major determinant of bonus payments is income, which has a large large, positive, and

significant effect. Finally, those leases signed by operators result in larger bonus payments.

6.6 Split Estate Analysis

This split estate analysis follows a similar pattern to the prior analyses, and in particular,

we estimate a seemingly unrelated regressions system measuring the effect of a split estate on the

likelihood of a particular clause being included in the lease. Based on the results in Table 7.1,

we see that split estates decrease the likelihood of good clauses being written into the leases with

two exceptions: third party negotiations and offset wells. Split estates are more likely to include

a subsurface easement clause, a lease “bad,” enabling firms to lay pipelines through the mineral

estate even if only to access wells from which the property owner is not receiving royalties.

Further, we observe that split estates reduce the likelihood of a surface clause, which is an

intuitive result since the lease is negotiated between two parties that do not have an interest in the

status of the surface estate.

Table 7.2 compares the lease quality index regressions across households identified as

split estates and not. We find that in both samples being Hispanic has a negative effect on the

lease quality, and Hispanic interacted with English speaking improves the quality.

Overall, non-split estates have poorer lease quality than split estates as evidenced by the

negative and significant constants suggesting that non-split owners negotiate worse terms than

! 42

split estate owners. The results comparing the effects of split and non-split estate households on

the lease quality index are counter-intuitive to the SUR estimates which suggested that a split

estate does not increase the likelihood of most clauses being written into the lease. However, the

latter analysis relies on sample of households where leases are signed after the transaction date.

Sellers keeping the mineral estates while selling the surface estate may be more savvy

negotiators or have more experience with the oil and gas industry and thus know the value of

severed estates. However, this is not a result we cannot test.

Finally, Table 7.3 reports the results of a model describing the likelihood of a split estate

as a function of household attributes whereby the variable with the largest, significant effect is

income.

6.7 Lease Terms and Violations

The final set of specifications reports preliminary results describing the relationship

between lease clauses and violations as reported by data from Texas Railroad Commission.

To maintain consistency across the different analyses, we performed another set of seemingly

unrelated regressions of the lease clauses on the counts of violation types (separately) along with

the other tract-level control variables and city-level fixed effects. The partial results of the SUR

are reported in table 8. Each column reports the effect of the violation on the lease clauses from

the separately identified models (one model for each violation or docket type). We did not report

the estimates of the control variables for each specification to simplify the interpretation of the

table. We find that an increase in the count of any of these docket types or violations within a

2000-meter buffer around a house at the appraisal date decreases the likelihood that many of the

clauses were written into the leases.

! 43

These results are largely suggestive of future work addressing the relationship between

strict leasing documents and the future behavior of firms once they begin drilling a well and

extracting natural gas. Our results suggest when leasing clauses are absent, there are more

violations addressed by the TRC.

7. CONCLUSIONS AND POLICY IMPLICATIONS

We undertake an analysis of the determinants of the characteristics of leases negotiated as

part of the bargaining process for the assignment of rights to develop shale gas resources. With

the dramatic growth in U.S. shale gas development over the last decade (particularly in populated

areas), lease negotiations have become an important part of the energy landscape, and constitute

one of the primary potential sources of benefit for homeowners. This is significant, as there are

also many potential drawbacks to living near a shale gas well. The bargaining process associated

with the transfer of mineral rights from lessor to lessee shares many features of a classic Coasian

bargaining framework. If bargaining is, in fact, Coasian, that would suggest an efficiency result

that would reduce the need for costly government oversight of the leasing process. Using a

unique combination of data sets, we test for whether the bargaining process does indeed exhibit

characteristics of Coasian efficiency in one of the most active shale gas counties in the U.S. –

Tarrant Co., Texas. Our results suggest that a number of important determinants of willingness-

to-pay for avoiding shale gas development (e.g., income, water source) do indeed affect

bargaining outcomes. This result is indicative of Coasian efficiency – those who we would

expect to have a larger willingness-to-pay to avoid exposure to shale development indeed

negotiate for stronger lease terms. However, the story does not end there. The argument in favor

! 44

of Coasian efficiency becomes harder to sustain when we find similar results for race. There is

no reason to expect different race groups to have a different willingness-to-pay to avoid

environmental harm, so the fact that we find significant differences in lease quality across race

groups suggests a realization of environmental injustice.

The question then arises as to what enables that injustice. One possibility would be the

leverage that operators can exert in lease negotiations due to forced pooling. This is a topic that

we are taking up in other research. A second possibility is that it is the result of information

asymmetry. There are lots of potential sources of information asymmetry that one could think of

in the lease bargaining process, but the simplest (which is also most relevant for the most

disadvantaged minority group) would likely be linguistic – i.e., a household will have less ability

to bargain effectively if it is not English speaking. Our results suggest that this is indeed a

source of the outcome experienced by the Hispanic race group in the bargaining process.

Asymmetric information presents a clear mechanism for the breakdown of efficient Coasian

bargaining.

Moving forward, there are several alternative paths for communities tasked with

negotiating leases because of new natural gas discoveries. Access to information about lease

negotiation is increasingly available that details the potential terms and implications. Usually,

these are documents drawn by attorneys that represent property owners, and reading this

information decreases the asymmetry in information between property owners and firms that are

likely seasoned negotiators.

More broadly, states could adopt uniform leasing whereby leases are required to have a

minimum set of terms deemed fair to households relinquishing the rights to their subsurface

minerals for natural gas extraction. Establishing a minimum requirement ensures that all

! 45

households are protected up to a minimum threshold established by the regulating body which

could be based on the experience of communities with currently active natural gas industries.

Uniform leasing is not yet observed in practice, however, substituting uniform leasing are

stricter regulations overseeing the natural gas industry and their treatment of the surface and

subsurface estates during and after production. Already in the state of Texas local municipalities

are allowed to exercise ‘home rule’ which means they can pass stricter rules to be observed by

firms operating within their jurisdiction. As it stands, these rules are more or less strict across

cities with some requiring additional environmental testing, remediation and protection of the

landscaping, and local permitting standards, and restricting noise and the hours of operation,

among others. These rules are often considerably more focused on how the surface and

subsurface are treated during and after operations and benefit people living nearby. More

uniform adoption of these rules by local municipalities protects urban communities, and stricter

rules at the state level protect rural households, as well.

Leasing is largely an unregulated part of the natural gas industry enabling firms to

potentially exercise superior knowledge and market power to extract rents from households by

negotiating lenient lease terms. Disproportionately targeting certain types of households

exacerbates inequality felt by individuals living in regions with a lot of natural gas activity.

Policy-makers can address these problems by specifying a minimal leasing requirement that

protects the quality of life among households living near wells and decreases their risk exposure.

Further, policies, in general, that restrict how the industry operates increases the overall welfare

in these regions of the country such that they are not required to sacrifice safety and the aesthetic

of a home for a payment from the minerals beneath it.

! 46

REFERENCES

Ash M and T.R. Fetter. 2004. “Who lives on the wrong side of the environmental tracks? Evidence from the EPA’s Risk-Screening Environmental Indicators model.” Soc. Sci. Quarterly. 85:441–62. Bamberger, M. and R. Oswald. 2012. “Impacts of Gas Drilling on Human and Animal Health.” NEW SOLUTIONS: A Journal of Environmental and Occupational Health Policy. 22(1):51-77.

Been, V. 1994. “Unpopular Neighbors: Are Dumps and Landfills Sited Equitably?” Resources. Vol.115. Resources for the Future. Washington, D.C. Been and Gupta. 1997. “Coming to the Nuisance or Going to the Barrios? A Longitudinal Analysis of Environmental Justice Claims.” Ecology Law Quarterly. 24 (1):1-56.

Boxall, P.C., W.H. Chan, and M.L. McMillan. 2005. “The Impact of Oil and Natural Gas Facilities on Rural Residential Property Values: A Spatial Hedonic Analysis.” Resource and Energy Economics. 27(3), 248-269.

Brown P. 1995. “Race, class and environmental health: a review and systemization of the literature.” Environ. Res. 69:15-30.

Bryant B. and P. Mohai (eds). 1992. Race and the Incidence of Environmental Hazards: A Time for Discourse. Boulder, CO: Westview.

Bullard R.D. 2000. Dumping in Dixie: Race, Class, and Environmental Quality. Boulder, Colorado: Westview. 3rd edition. Bullard RD, P. Mohai, R. Saha, and B. Wright. 2007. Toxic Wastes and Race at Twenty 1987–2007: Grassroots Struggles to Dismantle Environmental Racism in the United States. Cleveland, OH: United Church Christ Justice Witness Ministry.

Caulton, D.R. , P.B. Shepson, R.L. Santoro, J.P. Sparks, R.W. Howarth, A.R. Ingraffea, M.O.L. Cambaliza, C. Sweeney, A. Karion, K.J. Davis, B.H. Stirm, S.A. Montzka, and B.R. Miller. 2014. “Toward a Better Understanding and Quantification of Methane Emissions from Shale Gas Development.” Forthcoming in Proceedings of the National Academy of Sciences.

Coase, R. 1960. “The Problem of the Social Cost.” Journal of Law and Economics.

! 47

James, A. and J. James. 2014. “A Canary Near a Gas Well: Gas Booms and Housing Market Busts in Colorado.” Unpublished Manuscript.

Colborn, T., C. Kwiatkowski, K. Schultz, and M. Bachran. 2011. “Natural Gas Operations from a Public Health Perspective.” Human and Ecological Risk Assessment: An International Journal. Colborn, T., K. Schultz, L. Herrick, and C. Kwiatkowski. 2012. “An Exploratory Study of Air Quality Near Natural Gas Operations.” Human and Ecological Risk Assessment: An International Journal. Cole, L.W. and S.R. Foster. 2001. From the Ground Up: Environmental Racism and the Rise of the Environmental Justice Movement. New York/London: N. Y. Univ. Press Chavis BF and C. Lee C. 1987. Toxic Wastes and Race in the United States. New York: United Church Christ. Fambrough, Judon. ``Hints on negotiating and oil and gas lease.” Technical Report 229, (2002), 1 - 11.

Feagin, J.R. and C.B. Feagin. 1986. Discrimination American Style: Institutional Racism and Sexism. Malabar, FL: Krieger. 2nd ed. Fields, Jack. “Resolving conflicts between surface and mineral development.” (2006), 1 - 23.

Fontenot, B.E., L.R. Hunt, Z.L. Hildenbrand, et al. 2013. “An Evaluation of Water Quality in Private Drinking Water Wells Near Natural Gas in the Barnett Shale Formation.” Environmental Science and Technology. 47(17):10032-40.

Gilman, J., B. Lerner, W. Kuster, J. de Gouw. 2013. “Source Signature of Vaolatile Organic Compounds from Oil and Natural Gas Operations in Northeastern Colorado.” Environmental Science and Technology. 47(3):1297-1305.

Hamilton, J.T. 1995. "Testing for Environmental Racism: Prejudice, Profits, Political Power?" Journal of Policy Analysis and Management. Vol. 14(1):107-132. Higgins, R. 1994. “Race, pollution and the mastery of nature.” Environ. Ethics. 16:251–63.

Hill, E.L. 2013. “Shale Gas Development and Infant Health.” Dyson School Working Paper 2012-12. Cornell University.

! 48

Jenkins, R.R., K.B. Maguire and C.L. Morgan. 2004. "Host Community Compensation and Municipal Solid Waste Landfills." Land Economics. 80(4):513-528.

Klaiber, A. and S. Gopalakrishnan. 2014. “Is the Shale Boom a Bust for Nearby Residents? Evidence from Housing Values in Pennsylvania.” Forthcoming in American Journal of Agricultural Economics.

Kovats, S., M. Depledge, A. Haines, L.E. Fleming, P. Wilkinson, S. B. Shonkoff, and N. Scovronick. March 1, 2014. “The Health Impacts of Fracking.” www.thelancet.com. 383:757-758.

McKenzie, L., R. Witter, L. Newman, and J. Adgate. 2013. “Human Health Risk Assessment of Air Emissions from Development of Unconventional Natural Gas Resources.” Science of the Total Environment. 424(1):79-87.

Marchand, J. 2012. “Local Labor Market Impacts of Energy Boom-Bust-Boom in Western Canada.” Journal of Urban Economics. 71:165-174.

Maxwell, B. “As tug of war: a survey of urban drilling and the issues an operator will face,” Texas Journal Oil \& Gas Energy Law, (2008-2009) Vol. 4: 337 - 362.

McFarland, John. “Checklist for negotiating and oil and gas lease.” 1 - 21, http://www.gdhm.com/images/pdf/jbm-ogleasechecklist.pdf.

Medema, S. 1994. Ronald H. Coase. New York: St. Martin’s Press.

________. 2014. “The Curious Treatment of the Coase Theorem in the Environmental Economics Literature, 1960-1979.” Review of Environmental Economics and Policy. 8(1):39-57. Merrill, Richard and Fabio Merrill. “Ownership of mineral rights under Texas law.” 1 - 44.

Mills CW. 2001. Black trash. In Faces of Environmental Racism: Confronting Issues of Global Justice, ed. L Westra, BE Lawson, pp. 73–94. Lanham, MD: Rowman & Littlefield. 2nd ed.

Morello-Frosch et al. (2002). “Environmental Justice and Regional Inequality in Southern California: Implications for Future Research.”

! 49

Morello-Frosch R, Jesdale B. 2006. “Separate and unequal: residential segregation and air quality in the metropolitan US.” Environmental Health Perspectives. 114:386–94.

Muehlenbachs, L.A. and A.J. Krupnick. 2014. “Infographic: Shale Gas Development Linked to Traffic Accidents in Pennsylvania.” Resources. Vol 185. Resources for the Future. Washington, DC.

Muehlenbachs, L.A., E. Spiller, and C. Timmins. 2014. “The Housing Market Impacts of Shale Gas Development.” NBER Working Paper 19796.

Oakes, Anderton, and Anderson (1996) “A Longitudinal Analysis of Environmental Equity in Communities with Hazardous Waste Facilities.”

Olmstead, S.M., L.A. Muehlenbachs, J.S. Shih, Z. Chu, and A. Krupnick. 2013. “Shale Gas Development Impacts on Surface Water Quality in Pennsylvania.” Proceedings of the National Academy of Sciences. 110(13):4962-4967. Pastor M., Sadd J., Hipp J. 2001. “Which came first? Toxic facilities, minority move-in, and environmental justice.” Journal of Urban Affairs. 23:1–21. Pulido L. 2000. “Rethinking environmental racism.” Ann. Assoc. Am. Geogr. 90:12–40 Rahm, D. “Regulating hydraulic fracturing in shale gas plays: The case of Texas,” Energy Policy, (2011) Vol. 39: 2974-2981. Ringquist, E.J. 2003. “Environmental Justice: Normative Concerns and Empirical Evidence.” In Environmental Policy: New Directions for the Twenty-First Century, ed. N.J. Vig, M.E. Kraft, pp. 249–73. Washington, DC: CQ Press. 5th ed. Ringquist E.J. 2005. “Assessing evidence of environmental inequities: a meta analysis.” Journal of Policy Analysis and Management. 24:223–47. Roberts J.T., Toffolon-Weiss M. 2001. Chronicles from the Environmental Justice Frontline. Cambridge, MA/New York: Cambridge Univ. Press

Roy, A., P. Adams, and A. Robinson. 2014. “Air Pollutant Emissions from the Development, Production, and Processing of Marcellus Shale Natural Gas.” J Air Waste Manage Assoc. 64:19-37. Saha R, Mohai P. 2005. “Historical context and hazardous waste facility siting: understanding temporal patterns in Michigan.” Soc. Probl. 52:618–48.

! 50

Schweizer, U. 1988. “Externalities and the Coase Theorem: Hypothesis or Result?” Journal of Institutional and Theoretical Economics. 144(2):245-266. Szasz A., Meuser M. 1997. “Environmental inequalities: literature review and proposals for new directions in research and theory.” Current Sociology. 45:99–120. Texas Administrative Code: Oil and gas division, Chapter 3.

Throupe, R., R.A. Simons and X. Mao. 2013. “A Review of Hydro ‘Fracking’ and Its Potential Effects on Real Estate.” Journal of Real Estate Literature. 21(2):205-232. US GAO. 1983. Siting of Hazardous Waste Landfills and Their Correlation with Racial and Economic Status of Surrounding Communities. Washington, DC: US Gov. Print. Off.

Vanham, R. “A shift in power: why increased urban drilling necessitates a change in regulatory authority,” Saint Mary's Law Journal, (2011-2012) Vol. 43: 229 - 288.

Warner, N.R., C.A. Christie, R.B. Jackson, A. Vengosh. 2013. “Impacts of Shale Gas Wastewater Disposal on Water Quality in Western Pennsylvania.” Environ Sci Technol. 47:11849-57.

Zelder, M. 1998. “The Cost of Accosting Coase: A Reconciliatory Survey of Proofs and Disproofs of the Coase Theorem.” In Coasean Economics: Law and Economics and the New Institutional Economics. Seven G. Medema (ed.). Kluwer Academic Publishers. Dordrecht, The Netherlands.

! 51

Table 1.1: Summarized well exposure by year (2000 meters) Obs. Mean Std. dev.

1995 225 0.00 0.00 1996 227 0.00 0.07 1997 266 0.00 0.00 1998 280 0.01 0.12 1999 358 0.02 0.13 2000 886 0.18 1.42 2001 863 0.60 3.06 2002 937 0.18 0.98 2003 985 0.52 2.53 2004 1252 0.41 1.74 2005 1107 0.46 2.47 2006 1855 1.56 6.28 2007 2911 2.74 9.04 2008 2810 11.30 20.68 2009 2023 10.70 16.24 2010 6181 7.41 14.15 2011 7734 20.26 19.05 2012 2692 24.75 20.41 2013 1847 30.08 23.30 ! !

! 52

!Table 1.2: Basic lease terms (lease sample used in the analysis)

N Mean

Std. dev.

Lease term (months) 37,396 44 12 Royalty rate 35,243 0.23 0.02 Bonus 1,350 15,637 6,915 Ext. Bonus 4,963 8,550 8,423

Table 1.3: Lease clause Mean Std. dev.

Royalty rate

0.23 0.02 Area

2.89 44.84

Bonus

15,634.00 6,917.00 Ext. Bonus

8,547.00 8,423.00

Ext. Term Months 0.38 0.49 Third party negotiation 0.01 0.08 Legal fees

0.03 0.16

Force Majuere 0.14 0.35 Sub-surface easement 0.04 0.19 Same bonus to extension 0.12 0.33 Pugh

0.12 0.33

Surface damage 0.12 0.32 Min. residential street access 0.02 0.13 Assignment restriction exception (often Chesapeake) 0.01 0.08 Indemnity

0.03 0.18

Insurance

0.00 0.04 No surface access 0.19 0.43 Environmental 0.14 0.51 Noise restriction 0.04 0.28 Assignment restriction 0.05 0.21 Post-prod costs 0.03 0.17 Offset well

0.01 0.10

! 53

Table 1.4: Frequency of each event type by year Appraised Transactions Signed leases

1993 753 1994 1,259 1995 225 1996 227 1997 266 1998 280 1999 358 2000 886 2001 863 2002 937

1 2003 985 4,013 108 2004 1,252 4,500 120 2005 1,107 5,086 417 2006 1,855 5,396 2,704 2007 2,911 4,298 15,350 2008 2,810 3,489 13,955 2009 2,023 3,296 1,530 2010 6,181 2,707 2,093 2011 7,734 2,450 650 2012 2,692 2,207 480 2013 1,847

44

Table 1.5: Compare attributes of two housing samples

Full Lease

Mean Std. dev. Mean Std. dev.

Log living area (sqft) 0.746 0.376 0.661 0.369 Log land (sqft) 9.135 0.622 9.127 0.596 Subdivision Sale 0.184 0.387 0.077 0.266 Bathroom 2.213 0.674 2.094 0.628 Bedroom 3.434 0.681 3.329 0.658 Variable rate mortgage 0.104 0.305 0.107 0.310 FHA loan 0.268 0.443 0.287 0.453 Active leases 0.176 0.381 0.365 0.481 Reading scores (2012 - 8th grade) 44.920 16.780 40.870 15.570 House age 0.202 0.204 0.270 0.224 Groundwater 0.655 0.475 0.636 0.481 Log well exposure 1.467 1.570 1.321 1.490 Groundwater * Log well exposure 0.917 1.444 0.818 1.360

Observation 134,618

37,455

! 54

Table 1.6: Correlation matrix of lease clauses

Royalty

rate

Post production

cost Lease term (months)

Force majuere Legal fees Pugh

Third party negotiation Offset well

Royalty rate 1 Post production cost 0.0525 1

Lease term (months) -0.0537 0.0188 1 Force majuere 0.0786 -0.0704 -0.0987 1

Legal fees 0.0347 -0.0284 -0.0344 0.2372 1 Pugh 0.0792 -0.064 -0.1212 0.8602 0.2795 1

Third party negotiation 0.0058 -0.0146 0.0079 0.0838 0.5145 0.222 1 Offset well 0.006 -0.017 0.0238 0.2401 -0.0158 -0.0364 -0.0081 1

Assignment restriction 0.0549 0.6108 0.0155 0.0629 0.3216 0.0612 0.1741 -0.0215 Reporting requirement 0.0664 -0.0253 -0.1116 0.6438 0.2711 0.6521 0.1242 -0.0291 Noise requirement 0.0345 -0.0253 -0.0413 0.2508 0.8986 0.2948 0.4315 -0.015 Surface damage 0.0933 0.4485 -0.0574 0.4776 0.1414 0.5233 0.0983 -0.035 Environmental 0.0682 -0.0479 -0.103 0.6759 0.2163 0.7229 0.1383 -0.0267 No surface access 0.0936 0.2039 -0.0902 0.7453 0.4984 0.7839 0.2524 -0.0425 Subsurface easement 0.0554 0.8765 0.0338 -0.0793 -0.0319 -0.0723 -0.0164 -0.0191

Assign.

restriction Reporting

req. Noise req. Surface damage Env.

No surface access

Subsurface easement

Assignment restriction 1

Reporting requirement 0.1642 1 Noise requirement 0.3792 0.3073 1

Surface damage 0.5489 0.6991 0.1787 1 Environmental 0.1291 0.9169 0.2533 0.7546 1

No surface access 0.4292 0.5749 0.5276 0.6416 0.6064 1 Subsurface easement 0.6273 -0.0331 -0.0299 0.467 -0.0539 0.2214 1

! 55

Table 1.7: Legal clause correlation matrix

Post prod.

cost Lease term

Force majuere

Legal fees Pugh

Third party neg.

Offset well

Assign. rest.

Report req.

Post production cost 1 Lease term (months) 0.0188 1

Force majuere -0.0704 -0.0987 1 Legal fees -0.0284 -0.0344 0.2372 1

Pugh -0.064 -0.1212 0.8602 0.2795 1 Third party negotiation -0.0146 0.0079 0.0838 0.5145 0.222 1

Offset well -0.017 0.0238 0.2401 -0.0158 -0.0364 -0.0081 1 Assignment restriction 0.6108 0.0155 0.0629 0.3216 0.0612 0.1741 -0.0215 1

Reporting requirement -0.0253 -0.1116 0.6438 0.2711 0.6521 0.1242 -0.0291 0.1642 1

Table 1.8: Surface clause correlation matrix

Noise

restriction Surface damage Env.

No surface access

Subsurf. easement

Noise restriction 1 Surface damage 0.1787 1

Envrionmental 0.2533 0.7546 1 No surface access 0.5276 0.6416 0.6064 1

Subsurface easement -0.0299 0.467 -0.0539 0.2214 1

! 56

Table 2.1: Seemingly unrelated regressions (SUR) of lease attributes Dependent Variables

Hispanic Asian Black

American Indian/ Alaska Native

HH Income

Hispanic * English speaking

HH Income squared

Royalty rate -0.005* 0.001 0.000 0.003 0.365*** 0.000** -0.000*** Post production cost -0.016 -0.010 0.001 0.036 -0.372 0.000 0.000 Lease term (months) 0.083*** 0.009** 0.011*** -0.003 -1.624*** -0.001*** 0.000** Force majuere -0.130*** 0.068*** -0.004 -0.042 2.470*** 0.002*** -0.000 Legal fees 0.033** 0.006 0.006 -0.008 0.393 -0.000* -0.000 Pugh -0.132*** 0.058*** -0.022*** -0.029 3.243*** 0.001*** -0.000 Third party negotiation 0.026*** -0.003 0.000 0.007 0.358** -0.000*** -0.000** Offset well 0.020 0.002 0.002 -0.006 0.005 -0.000 -0.000 Assignment restriction 0.011 0.010 0.007 0.060** -0.502 0.000 0.000 Reporting requirement -0.140*** 0.047*** 0.006 -0.028 1.080** 0.002*** -0.000** Noise restriction 0.068*** 0.009 0.011 -0.007 0.968* -0.001** -0.000* Surface damage -0.190*** 0.049*** -0.016** 0.033 1.128** 0.002*** -0.000 Environmental -0.234*** 0.120*** -0.030** -0.056 2.519*** 0.003*** -0.000** No surface access -0.090** 0.064*** -0.007 0.016 3.015*** 0.001** -0.000 Subsurface easement 0.003 -0.014* 0.001 0.067** -0.353 0.000 0.000 Coasian explanatory variables

Log land

(sqft)

Sub- division

Sale Ground-

water Over 65 (tract)

Under 18 (tract)

High school (tract)

Royalty rate 0.001*** -0.002*** -0.006*** -0.005** -0.020*** 0.060*** Post production cost -0.001 0.013*** 0.004 -0.067*** -0.047*** 0.014 Lease term (months) -0.010*** -0.009*** -0.013** 0.013 0.079*** 0.043*** Force majuere 0.005 0.004 -0.013 0.321*** 0.052* 0.203*** Legal fees 0.003 0.012*** -0.017** 0.001 -0.032** 0.048** Pugh 0.004 -0.006 -0.049*** 0.370*** 0.043 0.154*** Third party negotiation 0.001 0.000 -0.013** 0.036*** 0.040*** 0.015 Offset well 0.001 -0.005** 0.009* -0.033*** -0.003 0.045*** Assignment restriction -0.002 0.020*** -0.004 -0.042* -0.054*** 0.113*** Reporting requirement 0.007** 0.011 -0.026** 0.301*** 0.101*** 0.113*** Noise restriction 0.003 -0.008 -0.019* -0.018 -0.083*** 0.085** Surface damage 0.009** 0.017** -0.016 0.186*** 0.028 0.073 Environmental 0.018*** -0.013 -0.045** 0.591*** 0.094** 0.022 No surface access 0.007 0.029*** -0.076*** 0.213*** -0.000 0.284*** Subsurface easement 0.001 0.012** -0.008 -0.072*** -0.033* 0.025

Observations 22,061 City FE Yes Period FE No

*** p<0.01, ** p<0.05, * p<0.1

! 57

Table 2.2: Correlation matrix of SUR estimated error terms

Royalty

rate

Post production

cost

Lease term

(months) Force

majuere Legal fees Pugh

Third party negotiation Offset well

Royalty rate 1 Post production cost 0.1053 1

Lease term (months) -0.1778 0.0302 1 Force majuere 0.1922 -0.0848 -0.1043 1

Legal fees 0.097 -0.0374 -0.0273 0.2253 1 Pugh 0.1723 -0.0763 -0.1341 0.8389 0.2727 1

Third party negotiation 0.0422 -0.0157 -0.0017 0.0624 0.5004 0.2163 1 Offset well 0.0497 -0.0245 0.0298 0.257 -0.0186 -0.0403 -0.0091 1

Assignment restriction 0.1344 0.642 0.0306 0.0351 0.2881 0.0331 0.1277 -0.0298 Reporting requirement 0.1347 -0.022 -0.1286 0.6199 0.2716 0.6299 0.0997 -0.0316 Noise requirement 0.0806 -0.0336 -0.0397 0.2425 0.8871 0.2868 0.4067 -0.0182 Surface damage 0.191 0.4908 -0.051 0.4161 0.1234 0.4675 0.0719 -0.0423 Environmental 0.1386 -0.0539 -0.1154 0.6539 0.2161 0.7076 0.1135 -0.0282 No surface access 0.215 0.22 -0.0866 0.7137 0.4809 0.7574 0.2285 -0.0506 Subsurface easement 0.1216 0.8838 0.044 -0.0944 -0.0408 -0.085 -0.0181 -0.0268

Assign.

restriction Reporting

req. Noise req. Surface damage Env.

No surface access

Subsurface easement

Assignment restriction 1 Reporting requirement 0.1482 1

Noise requirement 0.3424 0.3097 1 Surface damage 0.5734 0.6522 0.1586 1

Environmental 0.1078 0.9072 0.2525 0.7056 1 No surface access 0.4263 0.5523 0.5141 0.6202 0.585 1

Subsurface easement 0.6563 -0.0304 -0.0383 0.515 -0.0602 0.2416 1

! !

! 58

Table 3: Hedonics regressions with lease attributes

(1) (2) (3) (4) (5) (6)

Households with leases Full sample

Log living area (sqft) 1.116*** 1.103*** 1.100*** 1.072*** 1.069*** 1.068*** Log land (sqft) 0.009** 0.007* 0.007* 0.020*** 0.020*** 0.020*** House age 0.119*** 0.129*** 0.128*** 0.034*** 0.044*** 0.043*** bedroom -0.113*** -0.111*** -0.110*** -0.102*** -0.101*** -0.101*** bathroom 0.111*** 0.109*** 0.108*** 0.106*** 0.106*** 0.106*** Subdivision Sale 0.006 0.011** 0.013** 0.002 -0.001 -0.001 Variable rate mortgage -0.023*** -0.021*** -0.021*** -0.025*** -0.022*** -0.022*** FHA loan -0.093*** -0.092*** -0.092*** -0.070*** -0.073*** -0.072*** Log well exposure 0.001 0.005** 0.005** -0.008*** -0.008*** -0.008*** Groundwater 0.033*** 0.036*** 0.037*** 0.011*** 0.008** 0.008** Groundwater * Log well exposure -0.018*** -0.019*** -0.019*** -0.003*** -0.003*** -0.002** Reading scores (2012 - 8th grade) -0.001*** -0.001*** -0.001*** -0.001*** -0.001*** -0.001*** Active Lease 0.063*** 0.065*** 0.064*** 0.027*** 0.041*** 0.040*** Royalty rate

0.627*** 0.615***

0.046*** 0.033**

Lease term (months)

-0.385*** -0.369***

-0.117*** -0.112*** Term (months) * Split

0.404*** 0.388***

0.112*** 0.102***

Noise restriction

0.034***

0.040*** Min. residential street access

-0.037

-0.042*

Environmental

0.033***

0.028*** Sub-surface easement

0.018

0.019

No surface access

-0.005

-0.023*** Surface damage

0.000

-0.006

Pugh

0.089***

0.111*** Force Majuere

-0.044***

-0.041***

Third party negotiation

-0.097***

-0.104*** Offset well

0.013

-0.010

Assignment restriction

-0.044***

-0.037** Reporting requirement

-0.065***

-0.050***

Insurance

-0.022

-0.019 Post-prod costs * Split

-0.022

-0.020

Pugh * Split

0.014

0.035 Force Majuere * Split

0.043

0.024

Negotiate * Split

0.141*

0.125* Offset well * Split

0.091

0.064

Assignment * Split

0.045

0.045 Reporting * Split

-0.097**

-0.109***

Insurance * Split

0.080

0.101

Observations 37,083 37,083 37,083 133,227 133,227 133,227 R-squared 0.775 0.780 0.781 0.818 0.818 0.819 City FE yes yes yes yes yes yes Period FE yes yes yes yes yes yes *** p<0.01, ** p<0.05, * p<0.1

! 59

!Table 4.1: Factor results all clauses

Factor1 Factor2 Factor3 Factor4 Uniqueness

Noise restriction 0.5677 -0.0462 0.7317 0.008 0.1402 Surface damage 0.7902 0.3443 -0.339 -0.0772 0.1361 Environmental 0.8271 -0.2503 -0.3251 -0.1147 0.1344 No surface access 0.8954 0.0434 0.0441 0.0257 0.1937 Subsurface easement 0.1933 0.9134 -0.1241 0.0471 0.1107 Post production cost 0.1921 0.9015 -0.124 0.0466 0.1329 Lease term (months) -0.1231 0.0974 0.0724 0.2752 0.8944 Force majuere 0.7739 -0.3285 -0.2615 0.2875 0.1421 Legal fees 0.5397 -0.0606 0.7738 0.0147 0.1061 Pugh 0.8175 -0.32 -0.1911 -0.0017 0.1927 Third party negotiation 0.3351 -0.0417 0.6037 0.0022 0.5215 Offset well 0.0013 -0.0686 -0.057 0.9424 0.1039 Assignment restriction 0.4536 0.7227 0.2041 0.0249 0.2297 Reporting requirement 0.8122 -0.2223 -0.2649 -0.1208 0.2061

Table 4.2: Factor regressions

Full index factor 1 0.013*** Full index factor 2 -0.016*** Full index factor 3 -0.007*** Full index factor 4 -0.013***

Royalty rate -0.011 Log well exposure 0.001 Groundwater 0.036*** Groundwater * Log well exposure -0.017*** Log living area (sqft) 1.125*** Log land (sqft) 0.011*** Bedroom -0.114*** Bathroom 0.111*** House age 0.129*** Subdivision Sale -0.008 Variable rate mortgage -0.037*** FHA loan -0.073*** Lease Signed (dummy) -0.018*** Reading scores (2012 - 8th grade) -0.001***

Observations 37,083 R-squared 0.774 City FE yes Period FE yes *** p<0.01, ** p<0.05, * p<0.1

! 60

Table 5.1: Index regression factoring all lease clauses together

(1) (2) (3) (4) (5)

Hispanic -0.116*** -0.093*** -0.554*** -0.499*** -0.519*** Asian 0.120*** 0.123*** 0.123*** 0.121** 0.116** Black -0.046* -0.037 -0.040 -0.057** -0.051** American Indian/Alaska Native -0.171* -0.168* -0.170* -0.160* -0.168** HH Income

9.769*** 9.251*** 7.289*** 7.417***

HH Income squared

-0.001*** -0.001*** -0.001*** -0.001*** Hispanic * English speaking

0.006*** 0.005*** 0.005***

Log land (sqft) 0.028** 0.011 0.011 0.037*** 0.040*** Subdivision Sale 0.008 -0.006 -0.013 0.012 0.016 Groundwater -0.089*** -0.092*** -0.099*** -0.053 -0.061 Over 65 (tract) 0.627*** 0.649*** 0.721*** 0.902*** 0.944*** Under 18 (tract) 0.184** 0.240*** 0.266*** 0.153* 0.182** High school (tract) -0.147 -0.081 -0.277* -0.068 -0.156

Observations 22,061 22,061 22,061 22,061 22,061 R-squared 0.006 0.007 0.008 0.026 0.030 City FE no no no yes yes Period FE no no no no yes *** p<0.01, ** p<0.05, * p<0.1

!!Table 6: Bonus tobits (log bonus) (1) (2) (3) (4)

Log land (sqft) -0.244 -0.437 1.306 1.106 Hispanic -42.30*** -39.70*** -50.25*** -47.01*** Black -2.651* -2.376* -2.393* -2.056 White 1.410* 1.037 0.966 0.662 Asian 0.248 0.0562 2.196 1.632 HH Income 533.3*** 547.2*** 596.7*** 599.4*** HH Income squared -0.0377*** -0.0408*** -0.0508*** -0.0521*** Hispanic * English speaking 0.479*** 0.450*** 0.567*** 0.530*** Lease signed by operator

12.40***

10.70***

Observations 23,465 23,465 23,465 23,465 City FE no no yes yes Period FE no no no no *** p<0.01, ** p<0.05, * p<0.1

! 61

Table 7.1 Seemingly unrelated regressions (SUR) of lease attributes using split estates Dependent variables

Split estate

Log land (sqft)

Subdivision Sale Groundwater

Over 65 (tract)

Under 18 (tract)

High school (tract)

Royalty rate -0.231*** 0.001*** 0.000 -0.006*** -0.000 -0.028*** 0.047*** Post production cost -0.010*** -0.002 0.013*** -0.001 -0.050*** -0.012 0.047*** Lease term (months) 0.001 -0.013*** -0.008*** -0.024*** -0.010 0.105*** 0.026** Force majuere 0.009 0.010*** 0.021*** -0.026** 0.341*** 0.017 0.184*** Legal fees 0.002 0.003** 0.013*** -0.007 0.020 -0.033*** 0.047*** Pugh 0.002 0.011*** 0.009 -0.042*** 0.367*** -0.019 0.139*** Third party negotiation 0.004* 0.002** 0.002 -0.009*** 0.051*** 0.032*** 0.006 Offset well 0.004* 0.002* -0.001 0.002 -0.011 -0.000 0.011 Assignment restriction -0.003 -0.003 0.019*** -0.001 0.002 -0.002 0.128*** Reporting requirement -0.004 0.008*** 0.015*** -0.026*** 0.308*** 0.079*** 0.131*** Noise restriction 0.003 0.004* -0.001 0.001 0.012 -0.091*** 0.074*** Surface damage -0.017*** 0.008*** 0.024*** -0.027*** 0.233*** 0.026 0.118*** Environmental -0.010 0.020*** 0.001 -0.047*** 0.606*** 0.025 0.045 No surface access -0.006 0.011*** 0.042*** -0.061*** 0.266*** -0.006 0.278*** Subsurface easement -0.009** -0.002 0.013*** -0.010 -0.043*** 0.001 0.047**

Observations 37,356 City FE Yes Period FE No

*** p<0.01, ** p<0.05, * p<0.1

! 62

Table 7.2: Comparing the effects on lease quality by split and non-split samples Not split estate Split estate

(1) (2) (3) (4)

Hispanic -0.499*** -0.519*** -0.682** -0.783** Asian 0.121** 0.116** 0.062 0.068 Black -0.057** -0.051** -0.061 -0.042 American Indian/Alaska Native -0.160* -0.168** -0.097 -0.098 HH Income 7.289*** 7.417*** 2.468 3.290 HH Income squared -0.001*** -0.001*** -0.000 -0.000 Hispanic * English speaking 0.005*** 0.005*** 0.007* 0.008** Log land (sqft) 0.037*** 0.040*** -0.026 -0.020 Subdivision Sale 0.012 0.016 -0.120 -0.132 Groundwater -0.053 -0.061 -0.124 -0.143 Over 65 (tract) 0.902*** 0.944*** 0.947** 1.016** Under 18 (tract) 0.153* 0.182** 0.140 0.063 High school (tract) -0.068 -0.156 0.071 -0.085 Constant -0.547*** -0.680*** 0.130 0.232

Observations 22,061 22,061 1,404 1,404 R-squared 0.026 0.030 0.051 0.065 City FE yes yes yes yes Period FE no yes no yes !Table 7.3: Split estate regression (probit) Hispanic -0.288 Asian -0.027 Black -0.015 American Indian/Alaska Native 0.164 HH Income 7.035** HH Income squared -0.000 Hispanic * English speaking 0.002 Log land (sqft) -0.072*** Hispanic (tract) 0.036 Subdivision Sale -0.140*** Groundwater 0.318*** Over 65 (tract) 0.084 Under 18 (tract) -0.470*** High school (tract) -0.900***

Observations 23,391 City FE Yes Period Yes Yes *** p<0.01, ** p<0.05, * p<0.1

! 63

Table 8: Violation seemingly unrelated regressions (SUR) of lease attributes Dependent variables

Violation (count) Enforcement Proper

plugging Reimbursement

Royalty rate

-0.001*** -0.000 -0.013*** 0.000 Noise restriction

-0.002*** -0.006 -0.045*** -0.013***

Surface damage

-0.001 -0.017** -0.056*** -0.023*** Environmental

-0.004*** -0.042*** -0.063*** -0.034***

No surface access

-0.001 0.004 -0.082*** -0.013*** Subsurface easement

0.001** -0.002 -0.021*** -0.009***

Post production cost

0.001 -0.004 -0.026*** -0.009*** Lease term (months)

-0.001*** 0.003 0.016*** 0.005***

Force majuere

-0.001* -0.015* -0.054*** 0.014*** Legal fees

-0.001*** 0.019*** -0.022*** -0.008***

Pugh

-0.001 -0.031*** -0.037*** 0.000 Third party negotiation

-0.000 0.005** -0.006*** -0.003***

Offset well

0.000 -0.001 -0.004 0.003** Assignment restriction

0.001** 0.005 -0.036*** -0.010***

Reporting requirement

-0.003*** -0.024*** -0.037*** -0.021*** *** p<0.01, ** p<0.05, * p<0.1

! 64

APPENDIX: OTHER LEASE TERMS

Primary Terms

1. Lease Term: The term of the lease often includes both primary and secondary terms in units of months or years. The primary term is the length of time allowed to drill a well and begin producing. Given that the well is producing in paying quantities, or is capable of producing in paying quantities, the primary term rolls-over into the secondary term of the lease, which remains in effect as long as the well is producing. A typical lease term ranges between three and five years. A longer lease term is generally considered to be bad from the point of view of the lessor, as it allows the lessee to hold mineral rights for a longer period without paying royalties.

2. Royalty: The fraction of earnings from the producing well paid to the lessors owning royalty interest in the well based on the acreage contribution of an individual lease to the producing well.

3. Bonus: A signing bonus is often negotiated at a per acre increment and is exchanged between the lessor and lessee at the time when the lease is signed. Bonus payments are frequently not reported in recorded lease agreements.

Auxiliary Terms

1. No surface access: May stipulate that all acreage must be pooled especially if the lessee owns a smaller tract of land. Other language may constrain where a well can be drilled in the context of a pooled agreement, for example, stating that the minerals may only be accessed through a well drilled on another pooled tract of land (might also be interpreted as a surface access clause).

2. Pugh Clause: This clause relinquishes ownership of the mineral estate back to the lessor at the end of the primary term in the event the producing well is not drawing from that portion of the lease.

3. Force Majeure: These clauses are often included to protect the lessee in the event of uncontrollable circumstances limiting or altogether halting operations on a well. To protect the lessor, additional clauses limiting the extent of delay or the definition of force majeure can be included.

4. Surface damage: In general, the lessee is not required to compensate the lessor for reasonable and necessary use of the surface to access the mineral estate. Lessors can negotiate on a variety of dimensions; however, it is important to note that in large urban areas many of these dimensions are regulated through municipal ordinance.

5. Noise Restrictions: Most commonly, leases will limit the amount of noise by restricting production to certain times of day or requiring mufflers be used with loud equipment.

6. Joint Negotiations: Some leases will be negotiated in bulk by a third party, perhaps a neighborhood association, and this can be noted in the leases including the name of the

! 65

third party negotiator. 7. Environmental clause: Lease clause limiting the types of substances allowed for

exploration and extraction activities. The clause encourages the use of safeguards to prevent contamination of soil, water, and surface and subsurface strata. In particular, the clause limits the use of hazardous substances as defined by the Comprehensive Environmental Response Compensation and Liability Act (CERCLA). There may be additional pollution restrictions and control mechanisms required by lessees.

8. Post production costs: Unless stipulated by the lease, post-production costs, like transportation and compression, might be deducted from the final royalty payment owed to lessors. Leases will often, to the benefit of lessees, contain language about royalty being calculated at the mouth of the well allowing courts leeway to rule in favor of deducting additional post-production costs. To protect property owners, leases should contain language constraining the post-production costs or stating that royalty be calculated at the point of sale. (McFarland)

9. Assignment Clause: Since the lease is an asset to lessees, they can exchange hands across parties over the primary term. Leases ordinarily do not control re-assignment of mineral ownership; however, there are several sets of language to constrain how re-assignment of ownership takes place. Leases may stipulate that lessors be informed of any assignment of the lease to a third party firm. Lessors may also hold the original lessee responsible for the negligence of the third party lessee and limit the type of interest that is transferable to a third party.

10. Indemnity: An indemnity clause shifts the liability from the lessor to the lessee in the event that a third party claims negligence on the part of the lessor for lessee activities. It is noted that the indemnity clause must be strong by satisfying the “express negligence” rule otherwise the court system is likely to not uphold the indemnity clause. This is achieved by including the phrase “including claims alleging that the lessor is guilty of negligence of other misconduct.” (McFarland)

11. Reporting requirement: Lessors may stipulate what information is to be provided to lessees including any reports on production or activity, geological or seismic surveys, assignments, description of the pooling unit, royalty calculations, and contracts for selling the oil and gas.

12. Offset wells: Similar to top-leasing, provisions allowing lessors to insist firms drill off-set wells shortens the time to a producing (and profitable) well. These clauses are included to protect lessors concerned with their minerals being drained by nearby wells.

13. Minimize residential street access: Lease clauses may restrict how the well is accessed in terms of roads built for well traffic and reducing traffic to the wellpad using residential streets.

14. Subsurface (perpetual) easement: Leases may state that the lessee gives the right to use the property to access wells (with pipelines) located on theirs and other property which may not be used to develop the lease signed, and that the easements can remain in place after the lease expires. This language seems particularly relevant for gathering lines.

15. Insurance: Leases may require specific general liability and worker compensation insurance coverage, for example.