Research Findings on Environmental Impacts of Sustainable Shale Gas Development in Pennsylvania The Center for Energy Economics and Policy Alan Krupnick, PhD Director, Center for Energy Economics and Policy Widener University School of Law September 27, 2013 2 Risk Matrix 1. Expert survey of shale gas development risks 3. State-by-state regulatory analysis 2. Statistical analysis: a)Effects of shale gas activity on surface water quality in Pennsylvania b)Analysis of chemical assays of flowback/produced water c)Property Value effects 3 4. Public Survey 5. Summary Sloan Project on Environmental Risks Surveying the Experts: Who & What? 215 experts: NGOs (35): Most national environmental groups, some local Academics (63): Universities/think tanks Government (42): Federal agencies; about half the relevant states; river basin commissions Industry (75): Operating and support companies, trade associations, consulting firms, law firms Chose high priorities among 264 possible risks Overlap of each groups high priority routine risk pathways Consensus routine risk pathways 7 RFF project focuses on environmental risks from shale gas development Surface Water Quality Risk Study (PNAS, 2013) We exploit spatial and temporal variation in the proximity of shale gas wells, waste treatment facilities, and surface water quality monitors in Pennsylvania to estimate: 1.the impact of shale gas wells on downstream chloride and TSS concentrations; and 2.the impact of shale gas waste treatment and release to surface water on downstream chloride and TSS concentrations. 8 9 RFF project focuses on environmental risks from shale gas development Conclusions 10 No statistically significant impact of shale gas wells on downstream chloride concentrations. A positive result here would have been consistent with contamination problems from spills, dumping, etc. Release of treated shale gas waste to surface water by permitted waste facilities appears to increase downstream chloride concentrations. Effect is significant only for POTWs, not CWTs. Shale gas wells appear to increase downstream TSS concentrations. Public Survey Public concerns for shale gas development well known, but no information currently available on Risk valuation Risk preference tradeoffs We survey public in Texas and Pennsylvania to elicit attitudes and (monetary) preferences for five key risk attributes: Groundwater wells with problems Surface water bodies with problems Violations of air quality standards Time lost to congestion Percent habitat fragmented Three different information treatments 11 12 Degree of concern about the environmental consequences of shale gas development (1=none, 7=extreme concern) 13 Degree of support of shale gas development (1=not at all, 7=extremely supportive) 14 Primary (conservative) WTP Estimates (mean, 90% CI), Controlling for Information Treatment and Baseline Top 5 states by number of gas wells 16 Truck Traffic Accidents in Pennsylvania by Well Activity 17 Thank you 18 Changes in attitudes post-information treatment State-by-state regulatory analysis 31 states 27 regulatory elements across shale development process Sources of data: statutes, regulations, independent reports, and interviews with regulators 19 20 Fracking Fluid Disclosure 21 22 Some Findings Heterogeneity is the rule. States average regulating about 70% of our elements (18-95%) Mostly command and control Of these, numerical standards range from 9% to 75% of maximum stringency Lack of transparency and data availability 23 24 Non-consensus top government concerns White area: All groundwater effects, including from pit or pond storage of fracturing fluids, DWI, saline intrusion from drilling. Other for states: The above effects on surface water, drilling fluids and cuttings to both surface and groundwater; frack fluids hydraulic fracture propagation to groundwater Noticeably lower state priorities: community and habitat effects and air pollutants Top 5 states by number of gas wells 27 Survey Facts Knowledge Networks (GfK) Panel Two waves FebruaryMay ,624 completed responses 3 information treatments (neutral, industry, NGO) 3 baselines (high, medium, low risks) 2 states (PA and TX) Drop off those who took l ess than 10 minutes to complete Final data 1490 respondents 28 Survey Organization I.Motivation (PA/TX considering different government programs, need your input) II.Brief intro to shale gas development in PA/TX III.Elicit attitudes towards risks IV.Information treatment (randomized) V.Choice questions (five) VI.Elicit attitudes towards risks (repeat) VII.Debriefing Information Treatments: Shale Gas: What Everyone Should Know Different treatments similar in length (about ten short paragraphs, one for each topic) topics covered (overall economic potential, environmental risks, efforts to manage them) Industry Information American Petroleum Institute Ten Facts about Shale Gas Technical focus, emphasizes economic potential Assures risks not large, industry goes beyond what is required to address risks NGO Information Extracted from NGO websites, press releases, and so forth Highlights risks, downplays economic potential Mixes scare tactics, anecdotes, facts, problem cases Neutral Rational Middle, balanced view Factual, evidence based, notes expert agreement and disagreement Recognizes risks, highlights potential methods to reduce them Kind of boring & Not telling me how to think (focus group participants) Focus groups correctly identified the different treatments 29 30 $60 per year ($5 per month) $120 per year ($10 per month) 10,000 IndustryNGONeutral Hydraulic fracturing will account for nearly 70 percent of natural gas development in the future development applies the latest technologies and makes it commercially viable Without it, we would lose 45 percent of domestic natural gas production and 17 percent of our oil production within 5 years. Shale gas supplies are overestimated, and it is not as cheap as it has been touted to be. The big reason is that shale gas wells, unlike oil wells, peter out quickly the viability of shale gas as a solution to Americas high energy consumption level is only on an interim basis. Shale gas is more likely to be a stopgap, a 25 year solution rather than a 100 year one. According to the US Energy Information Agency, discovered and technically recoverable shale gas resources in the United States contain enough natural gas to satisfy over 100 years of demand in the entire country. Some question these estimates and claim actually recoverable shale gas supplies to be less. Regardless, shale gas resources are massive. Stylized Examples of Information Treatments: Overall Prospects of Shale Gas Development 32 Risk concerns and mean votes for SQ (total five votes) 33 WTP for Risk Reductions (mean, 90% CI), by Information Treatment (unpooled estimation) 34 WTP for Risk Reductions (mean, 90% CI), by State and Baseline (unpooled estimation) 35 Discussion Public is supportive of shale gas but concerned about environmental risks Groundwater a major concern in both states Surface water a big concern in TX, less so in PA Wildlife habitat a big concern in PA, not at all in TX Air quality a relatively small yet statistically significant concern Community impacts not a major concern in either state, on average Considerable heterogeneity; key area for future work Preferences are polarized, more than 1/3 not WTP 36 Discussion WTP fairly robust to information & baseline, some evidence that neutral information moderates WTP, non-neutral increases it lower risk baseline raises WTP per unit neutral information and high baseline conservatively estimate WTP Information effects deserve closer attention, still relatively poorly understood in valuation Top 5 states by number of gas wells 40 Treatment Effects Using Pooled Data. Estimated benchmarking to high baseline and neutral information as beta= (1+beta_treatment)*beta_pooled PennsylvaniaTexas Meanp-valueMeanp-value Industry Information (n=477) Status Quo Environmental Attributes NGO Information (n=513) Status Quo Environmental Attributes Medium Baseline (n=426) Status Quo Environmental Attributes Low Baseline (n=445) Status Quo Environmental Attributes 41 WTP for Risk Reductions (mean, 90% CI), by State, Pooled Data, No Controls for Baseline or Information Alternative Baselines: Environmental Risks in Status Quo 42 HighMediumLow Ground water (drinking wells with problems) 10,0005,0002,500 Surface water (surface water bodies with problems) 10%5% 2.5% Air quality (days of air quality violations) Wildlife Habitat (percentage fragmented) 10%5% 2.5% Community Effects (time loss) 10%5% 2.5% 43 Hypotheses to be Tested First PA SQ WTP less than TX; hopefully insignificant from zero. PA community WTP>TX TX surface and groundwater WTP > PA Groundwater concerns most valuable WTP (API) less than WTP (RFF) less than WTP (NGO) Attribute WTP (High) greater than Attribute WTP (medium, low) Standard Deviation of the WTP Distribution, by Information Treatment, Unpooled Data 44 Public Survey Sample Texas and Pennsylvania Willingness to pay for reducing various types of shale gas risks Three alternative information treatments describing risks: APIs NGO composite RFFs 45 46 47 Discrete Choice Econometric Modeling Random parameter multinomial logit model Estimation in WTP space (Train and Sonnier 2005) First, pooled results without information treatment and baseline controls Second, split sample results by information treatment and baseline sub-samples Third, pooled estimation to test and control for information treatment and baseline effects IndustryNGONeutral Ground water Each well contains multiple layers of steel casing and cementing to effectively protect groundwater NAS study found average methane concentrations in shallow drinking water wells 17 times higher than those in non-active areas, possibly due to leaky gas- well casings Some evidence of drinking water well contamination linked to shale gas development. Scientists say that it is unlikely that the fracking fluids will migrate up to the surface or to groundwater Surface Water Numerous protective measures are in place at well sites to control and direct any potential runoff at the site. Even if the wastewater does not seep directly into the soil, a heavy rain can cause a pit to overflow and create contaminated runoff infrequent reports of localized ecosystem disruption from spills. To minimize risks, operators could focus on stopping leaks and spills states could do more monitoring Air Quality Industry has made considerable strides in reducing emissions through voluntary measures working with EPA to define performance standards that will give rise to additional reductions Sublette County, a rural Wyoming community with a high concentration of gas wells, has recorded higher ozone levels than those in Los Angeles To minimize risks, states could impose more controls on emissions and/or industry could increase efforts to reduce such emissions 49 Relationship between risk attitudes and number of status quo votes 50 Treatment Effects Using Pooled Data. Estimated benchmarking to high baseline and neutral information as beta= (1+beta_treatment)*beta_pooled PennsylvaniaTexas Meanp-valueMeanp-value Industry Information (n=477) Status Quo Environmental Attributes NGO Information (n=513) Status Quo Environmental Attributes Medium Baseline (n=426) Status Quo Environmental Attributes Low Baseline (n=445) Status Quo Environmental Attributes 51 Treatment Effects Using Pooled Data. Estimated benchmarking to high baseline and neutral information as beta= (1+beta_treatment)*beta_pooled PennsylvaniaTexas Meanp-valueMeanp-value Industry Information (n=477) Status Quo Environmental Attributes NGO Information (n=513) Status Quo Environmental Attributes Medium Baseline (n=426) Status Quo Environmental Attributes Low Baseline (n=445) Status Quo Environmental Attributes 52 WTP Estimates from Pooled Data, Controls for Baseline and Information, Reference Point High Baseline and Neutral Information PennsylvaniaTexas WTP Distribution Mean Estimate p-valueEstimatep-value Groundwater (1,000 wells w/ problems) Surface Water (% water bodies w/ problems) Air Quality (days of standard violation) Local Community (% time loss) Wildlife Habitat (% habitat fragmented) Status Quo WTP Distribution Standard Deviation Estimate p-valueEstimatep-value Groundwater (1,000 wells w/ problems) Surface Water (% water bodies w/ problems) Air Quality (days of standard violation) Local Community (% time loss) Wildlife Habitat (% habitat fragmented) Status Quo 53 Attitudes After the description of shale gas development Likert scale for attributes and environmental quality (here we average over responses) Likert scale for opposition/support of shale gas development Repeated after information and WTP questions 54 Survey Design Features Payment Vehicle: Increased utility (gas/electric) bills First survey wave groundwater, surface water, air quality, and community attributes medium and high baseline experimental design using orthogonal level selection Second survey wave substitute habitat fragmentation for community impacts in some versions low and high baseline Bayesian D-efficient experimental design using parameter estimates from the first wave (Monte Carlo Modified Fedorov design algorithm following Kessels 2009) Air pollutants Drilling fluids and cuttings Saline water intrusion Fracturing fluids Flowback constituents (other than fracturing fluids) Produced water constituents Condenser and dehydration additives Habitat/community disruptions Other Site development and drilling preparation Vertical drilling Horizontal drilling Fracturing and completion Well production and operation Flowback and produced water storage/disposal Shutting-in, plugging and abandonment Workovers Upstream and downstream activities Groundwater Surface water Soil quality Air quality Habitat disruption Community disruption Occupational hazard Human health impacts Market impacts Ecosystem impacts Climate change impacts Quality of life impacts ActivitiesBurdens Intermediate Impacts Final Impacts Creating Risk Pathways (Risk Matrices on the web) 55 Example of Impact Pathways On-road vehicle activity Air quality Community disruption Conventional air pollutants and CO2 Noise pollution Road congestion Morbidity Climate change impacts Aesthetics Time loss ActivitiesBurdens Intermediate Impacts Final Impacts Surveying the Experts: Findings Some surprises: Surface waters dominate; groundwater risks identified less frequently Only two pathways are unique to the shale gas development process Habitat fragmentation Some expected findings: On-site pit and pond storage of flowback Freshwater withdrawals Venting of methane Treatment and release of flowback liquids Surveying the Experts: Responsibility Who should be responsible for managing risks? All groups prefer shared responsibility For consensus pathways, majority of industry supports government responsibility NGOIndustryAcademicGovtAll experts Government93.8%49.4%74.9%74.8%69.4% Industry6.2%50.6%25.1%25.2%30.6%