Estimating the Impact of the Atlantic Sunrise Project on Natural Gas Consumers

Transcription

1 Final Report Andrew Kleit Chiara Lo Prete Seth Blumsack Nongchao Guo Kyungjin Yoo Estimating the Impact of the Atlantic Sunrise Project on Natural Gas Consumers John and Willie Leone Family Department of Energy and Mineral Engineering The Pennsylvania State University January 13, 2015 Acknowledgement: This report is an account of work sponsored by Williams Partners L.P. ( Williams ). The study s authors acknowledge that Williams provided financial support for the analysis contained herein. Contact author: ank1@psu.edu. Andrew N. Kleit, Professor of Energy and Environmental Economics, EXECUTIVE SUMMARY New production of natural gas in the Marcellus region of Pennsylvania has created tremendous opportunities to benefit consumers of natural gas in the United States. Unfortunately, the ability of this new production to create benefits is limited by the scarcity of pipeline capacity out of the Marcellus region. To address this issue, Williams wholly owned subsidiary, Transcontinental Gas Pipe Line Company, LLC ( Transco ), has proposed to expand its pipeline through the Atlantic Sunrise project. We seek to understand the market impact of the Atlantic Sunrise expansion by estimating the changes in equilibrium prices, demand and supply along the Transco pipeline that would result after the completion of the project. In this report, we construct a model assessing the impact of Atlantic Sunrise on flows, injections, withdrawals and natural gas prices along the Transco mainline from Alabama to New Jersey, using data from 2012 to the middle of In turn, this will provide insights on how this project will impact natural gas consumers on the eastern seaboard of the United States. The Atlantic Sunrise project would affect operation of the Transco pipeline system in two ways. First, it would permit additional deliveries from the Marcellus to markets on the Transco system. Second, it would afford more flexibility to the Transco system, with the ability to accommodate flow reversals in response to market conditions. In particular, the Atlantic Sunrise project will allow gas to flow southward from the Pennsylvania-Maryland-New Jersey Region into Virginia and further]south. 1

2 We have modeled the market impact of the Atlantic Sunrise project on flows and prices across the Transco system, focusing on the impact in three geographic areas: Pennsylvania- Maryland-New Jersey; Virginia, North Carolina and South Carolina; and Alabama and Georgia. Over the 30-month period examined in our study, we estimate that consumers served by Transco from Alabama to New Jersey would have enjoyed about $2.6 billion in total benefits because of the Atlantic Sunrise expansion. These benefits accrue due to lower prices and the opportunity for additional natural gas consumption (which is itself partially a consequence of lower prices). While we estimate that consumers would greatly benefit overall from Atlantic Sunrise, we wish to emphasize some specific aspects of our findings. First, the benefits to consumers are not uniform over time and will vary greatly with system conditions. As Table 5.6 shows, more than 60% of the estimated benefits of Atlantic Sunrise in our period of study would have accrued in January 2014 alone, because of the high level of gas demand associated with the polar vortex in that month. This finding in particular needs to be projected forward with care. Consumer benefits during the wintertime will generally be higher than in other seasons because of increased heating demand, but we estimate that these benefits would be roughly six to twenty times larger during very cold winters than during normal winters. If wintertime natural gas demand rises (due to cold weather, increased demand from power plants or other factors), this will magnify the consumer benefits of Atlantic Sunrise. Similarly, if very cold winters become relatively uncommon, the consumer benefits of Atlantic Sunrise will be correspondingly smaller over time. Second, the benefits to consumers are not uniform over geographic areas. Consumers from Alabama to Virginia would be the recipients of additional Marcellus gas flowing south due to Atlantic Sunrise, and would nearly always benefit from the pipeline expansion project. Because of the location of the constraints on the Transco system, we estimate that Virginia-North Carolina-South Carolina customers would benefit nearly three times as much as Alabama- Georgia customers. Pennsylvania-New Jersey-Maryland customers exhibit the highest benefits overall (across our 30-month estimation period), but will also be harmed during certain periods when exports from this region to Virginia cause prices north of Virginia to increase. These price increases, however, are far smaller than the price decreases that would occur due to Atlantic Sunrise during severe winter periods. 2

3 I. Introduction New production of natural gas in the Marcellus region of Pennsylvania has created tremendous opportunities to benefit consumers of natural gas in the United States. Unfortunately, the ability of this new production to create benefits is limited by the scarcity of pipeline capacity out of the Marcellus region. To address this issue, Williams has proposed to expand its Transco pipeline through the Atlantic Sunrise project. We seek to understand the market impact of the Atlantic Sunrise expansion by estimating the changes in equilibrium prices, demand and supply along the Transco pipeline that would result after the completion of the project. In this report, we construct a model assessing the impact of Atlantic Sunrise on flows, injections, withdrawals and natural gas prices along the Transco mainline from Alabama to New Jersey, using data from 2012 to the middle of In turn, this will provide insights on how this project will impact natural gas consumers on the eastern seaboard of the United States. In Section II of this report we describe the challenges posed by new sources of natural gas and how the Atlantic Sunrise project helps address those challenges. Section III presents our basic model, which includes arbitrage conditions, demand and supply elasticities, and the different equilibrium prices and flows at various points on the Transco system that can result from the Atlantic Sunrise expansion. Section IV examines the relevant flows and prices on four different days, one from each season, had Atlantic Sunrise been operational on those days. Section V presents the benefits Atlantic Sunrise would have had on natural gas consumers on the Atlantic seaboard from 2012 to the middle of Section VI contains our conclusions. II. The Atlantic Sunrise Project A. The Challenge of New Sources of Natural Gas In 1990, 52 percent of electricity in the U.S. was produced from coal and 12 percent from natural gas. By 2013, the share of coal had decreased to 39 percent, while that of natural gas had increased to 28 percent. 1 There are two main reasons for this change. First, the use of natural gas as an input for electric power generation provides significant environmental benefits: in addition to about half of coal s greenhouse gas emissions per unit of energy, natural gas has significantly lower traditional air pollutant emissions (nitrogen oxides and sulfur dioxides) than coal. Natural gas demand in the electricity sector is expected to further increase in the future, partly due to tightening environmental restrictions on the use of coal for power generation. 2 The use of natural gas as a transportation fuel has also been on the rise over the past ten years, though it still remains a negligible fraction of U.S. total consumption. 3 1 U.S. Energy Information Administration, 2 In June 2014, the Environmental Protection Agency proposed Clean Air Act rules to reduce carbon emission from fossil-fueled U.S. power plants by 30% by 2030, relative to their 2005 level. See 3 U.S. Energy Information Administration, 3

4 Second, the new supply of natural gas has lowered natural gas prices in the U.S. Figure 2.1 shows the price of natural gas in the United States from 1998 to Starting in 1998, the price of gas rose from near $2 per million BTU (MMBtu) to $8 in After that time, however, the price of natural gas fell rapidly, with wellhead prices falling below $3 per MMBtu in One of the most important reasons for this is the new production of gas from unconventional sources. Figure 2.1 U.S. Natural Gas Wellhead Price ( ) Source: EIA (2014), There are three types of unconventional gas: tight gas trapped underground in impermeable rock formations, shale gas from shale source rock, and coal bed methane (CBM) from coal source rock. This unconventional gas has been developed vigorously since the mid- 2000s thanks to the widespread use of hydraulic fracturing and horizontal drilling. (See, for example, Hefner (2014).) Hydraulic fracturing involves the injection of water, sand and other chemicals at high pressures to fracture hydrocarbon-bearing formations, releasing oil and gas. Horizontal drilling enables the recovery of larger hydrocarbon volumes with a smaller surface footprint, enabling higher recovery in hard-to-reach areas or locations where obtaining surface rights is prohibitive. 4 4 See Energy Information Administration, The Geology of Natural Gas Resources, 2014, 4

5 In terms of unconventional gas basins, there are seven major shale gas and/or oil production areas: Bakken, Eagle Ford, Haynesville, Marcellus, Niobrara, Permian, and Utica (Figure 2.2). These regions made up the growth in U.S. natural gas production during Figure 2.2 U.S. Lower 48 States Shale Plays Source: EIA (2011), Figure 2.2 depicts the major basins for U.S. unconventional energy production. In the early days of shale gas development, the Barnett shale in Texas was the largest producer of shale gas with over 4 million MMBtu per day. Since 2012, the Marcellus formation has yielded major new gas supplies in the U.S. Mid-Atlantic region. In 2013, the Marcellus produced 9.2 million MMBtu per day, contributing 13 percent of total U.S. natural gas supply. 6 The Marcellus Shale formation underlies significant portions of Pennsylvania, West Virginia, New York, eastern Ohio and other parts of eastern North America. It can be found 5 Energy Information Administration, Drilling Productivity Report, August 2014, 6 See 5

6 beneath about 60 percent of Pennsylvania s total land mass, occurring as deep as 9,000 feet below ground surface. 7 (See Figure 2.3.) A few thousand feet below the Marcellus underlying Ohio, Pennsylvania, West Virginia, New York, and Quebec, there exists another organic-rich rock unit named Utica shale. 8 Currently, the Utica Shale is receiving attention with a fast growth rate in energy production. It produced an estimated 1.3 million MMBtu per day in September In addition, the Utica play has a greater percentage of production from more profitable petroleum resources than the Marcellus. 9 Figure 2.3 Depth of Marcellus Shale Base Source: Penn State Marcellus Center for Outreach and Research, More than 90 percent of the natural gas consumed in the U.S. is produced domestically. 10 For decades, the focal point of natural gas production has been the U.S. Gulf Coast region. Natural gas transmission pipelines were designed and built to accommodate one-directional gas flow from the Gulf Coast area to the high-demand energy markets in the U.S. Southeast and 7 Penn State Marcellus Center for Outreach and Research, MCOR, 2014, 8 Utica Shale, Geology News and Information, 2014, 9 Drilling Productivity Report, Energy Information Administration, Aug 2014, 10 See and International Energy Agency,

7 Northeast. One of the largest of these pipelines, Transco, ships gas from the Gulf Coast to New York City. Transco is designed to deliver up to 11.2 million MMBtu per day of natural gas on a firm basis, and is crucial to maintaining natural gas supplies to distribution companies, industry and power plants. Bottlenecks on the Transco system have traditionally occurred during the winter, due to excess demand relative to available firm capacity. As noted above, over the past ten years the widespread use of horizontal drilling and hydraulic fracturing has allowed access to major natural gas deposits captured in shale formations in the Northeastern U.S. In particular, the Marcellus Shale formation in the northern Appalachian basin, in large part below the state of Pennsylvania, is rich in natural gas resources. Despite the existence of significant resources in close proximity to strong areas of natural gas demand in Southeastern and Northeastern states, natural gas pipeline capacity from Pennsylvania is severely constrained. This lack of pipeline capacity, in Pennsylvania and several other regions in the country, limits the potential economic gains from the extraction of unconventional gas. (See Joskow, 2013.) The lack of sufficient pipeline infrastructure has also induced price separation between Pennsylvania and the surrounding states. 11 The impacts of gas transmission constraints were felt particularly hard in the winter of 2014, when prices in surrounding states surpassed $100/MMBtu, while gas prices in Pennsylvania remained at less than one-tenth these levels. (See Figure 2.4 below.) In turn, low prices in Pennsylvania have economically constrained the completion of new Marcellus wells. Delivery constraints from limited transmission capacity also contributed to electric reliability issues during the polar vortex period in the Mid-Atlantic during January There are reports that only about half of the Marcellus wells being drilled in Pennsylvania are being completed. These economic constraints are particularly acute in the Northern Tier region of Pennsylvania, which produces primarily dry gas See, for example, Energy Information Administration, Some Appalachian natural gas spot prices are well below the Henry Hub national benchmark, October 15, 2014, 12 PJM Interconnection, LLC. Analysis of Operational Events and Market Impacts During the January 2014 Cold Weather Events, May 8, Available online at 13 See Andrew Maykuth, Natural-gas prices force down number of Marcellus drilling rigs, Philadelphia Inquirer, July 2, 2012, 08/business/ _1_natural-gas-prices-drilling-natural-gas. Dry gas is in contrast to wet gas, which is extracted with a variety of valuable associated products besides natural gas. 7

8 Figure 2.4 Natural Gas Prices per MMBtu on January 22, 2014 Source: B. Addressing the Challenge: The Atlantic Sunrise Project On the Transco system, bottlenecks have historically occurred in three different areas. First, the Leidy Line from the Marcellus region eastward into New Jersey is typically constrained, limiting flows of gas out of Pennsylvania. Indeed, the operations data we have reviewed indicate that this line is almost always constrained. Second, Transco currently restricts southward flow south of Transco Station 195 (located near the borders of Pennsylvania, Maryland and Delaware). During the summer this can be an important constraint, due to increased power generation demand to support air-conditioning load in mid-atlantic states. We thus observe that during the summer, natural gas prices can often be greater south of station 195 than north of 195. Finally, there is often a bottleneck on the Transco system east of station 90 in Alabama, which precludes more gas from flowing northward, and causes prices to rise at points north and east of Alabama. 14 The first two of these constraints in particular have contributed to gas supplies (both conventional and unconventional) being stranded in Pennsylvania, because transportation capacity is insufficient for much of that gas to reach customers outside of Pennsylvania. The stranding of gas supplies reduces production incentives and in extreme cases (such as the polar vortex) can have cascading impacts on other infrastructures, such as the electricity grid. 14 Historic data on flow constraints on the Transco system can be found at: 8

9 To overcome pipeline capacity constraints and help relieve Pennsylvania s stranded gas problem, Williams has proposed the Atlantic Sunrise project. The project goal is to expand and reverse the flow on part of the Transco pipeline, thus relieving the southbound bottleneck at Station 195. Atlantic Sunrise includes three components, presented in Figure 2.5: 1) The installation of two greenfield pipelines, the Central Penn Line North (55.79 miles of 30-inch pipe from compressor station 517 to Susquehanna County) and the Central Penn Line South ( miles of 42-inch pipe from compressor station 517 to Lancaster county, above station 195). The new lines will connect the Marcellus shale region to the Transco mainline near Station 195 in southeastern Pennsylvania. The Central Penn Line will provide 1.7 million MMBtu/day of additional natural gas transportation capacity to markets in the Mid-Atlantic and Southeast of the U.S. The increased system capacity associated with this expansion has already been allocated to market participants via contracts that are typically 20 years in duration. 2) The expansion of existing capacity of the Transco Leidy Line in the Marcellus region, running across the northern tier of Pennsylvania, 3) Pipeline replacements, construction of new compressor facilities, facility modifications and upgrading in five states, at various points on the Transco mainline. These additions and modifications will allow gas to flow bi-directionally on the main line. In particular, it is expected that the pipeline flow will often be reversed south of Station 195, allowing Marcellus gas to be supplied to customers in the Mid-Atlantic and Southeastern states. In turn, this will reduce the price of natural gas in these regions. Field surveys for the Atlantic Sunrise project began in Spring Williams is also seeking federal and state approval for the expansion. Compressor and pipeline construction is expected to start in 2016, while the project is scheduled for completion by the third quarter of The analysis in this report focuses on how the Atlantic Sunrise expansion would impact natural gas prices along the Transco mainline this corresponds to estimating the market impact of the second and third elements of the Atlantic Sunrise project. 9

10 Figure 2.5. Atlantic Sunrise Project Source: Williams, C. Other natural gas pipeline expansion projects to the East Coast and Mid- Atlantic regions Some of Williams competitors have also proposed (or are currently developing) pipeline projects to alleviate capacity constraints in the Marcellus production area. Our focus in this section is on proposed expansions carrying natural gas from Pennsylvania to markets along the East Coast and in the Mid-Atlantic region of the U.S. Owned by Columbia Pipeline Group (CPG), Columbia Gas Transmission is a major interstate natural gas system in the Northeast, and extends to the Midwest and Southeast regions. It consists of nearly 12,000 miles of transmission pipeline and can deliver up to 9.4 million MMBtu per day of natural gas. Its East Side Expansion project 15 involves upgrading and expansion of existing pipeline and compression facilities on the east side of the Columbia Gas Transmission line in Pennsylvania, to transport Marcellus shale gas to the Northeast and Mid- 15 Columbia Pipeline Group, 10

11 Atlantic markets. Pending regulatory approval, construction should begin in early 2015; the expansion is expected to be in service by the end of Spectra Energy owns and operates the Texas Eastern Transmission line, which runs for about 9,000 miles from the Gulf Coast to the Northeast, and can transport up to 8.5 million MMBtu per day of natural gas. The Texas Eastern Appalachia to Market 2014 (TEAM 2014) Project plans to expand existing capacity on the Texas Eastern Transmission line by approximately 600,000 MMBtu per day, via the addition of new lines ( looping ) and the installation of new compressor units. The project was authorized by FERC in February 2014, is currently under construction and is expected at this writing 17 to be in service in November Texas Eastern has also conducted an open season to solicit interest in the Access South project, delivering up to 320,000 MMBtu/day from the Appalachian region to markets in the Southeast by the end of Moreover, Spectra has recently announced plans to build a new underground pipeline, the Spectra Energy Pipeline project, carrying up to 1 million MMBtu per day of incremental capacity from Pennsylvania to the Mid and South Atlantic regions by the end of Tennessee Gas Pipeline (TGP), a subsidiary of Kinder Morgan Energy Partners, is a 13,900-mile line shipping natural gas from the Gulf Coast area to the Northeast. Through its Rose Lake Project, 19 TGP plans to expand the capacity of its 300 Line from the Appalachian and Marcellus region to the northeastern U.S. by about 230,000 MMBtu per day. The project received FERC approval in September 2013 and should be brought into service by November TGP has also proposed the Niagara Expansion project 20 to transport additional natural gas 16 In addition to the East Side Expansion project, CPG has proposed a West Side Expansion project, consisting of two components. The Gulf Bi-Direction project, which was completed earlier this year, included modifications and upgrades on the Columbia Gulf Transmission (another CPG company) to transport up to 540,000 MMBtu per day from Leach, Kentucky to Rayne, Louisiana. Moreover, the Smithfield III project will build new compressor stations on the west side of the Columbia Gas Transmission line, enabling 440,000 MMBtu per day to flow from Pennsylvania and West Virginia to Leach, Kentucky. The project received FERC approval in December 2013 and is anticipated to be in service by the end of See accessed November 19, Spectra Energy, Process/New-Projects-in-US/Texas-Eastern-Appalachia-to-Market-2014-TEAM-2014/. Spectra Energy is considering expansions of its Algonquin Gas Transmission system (crossing New England, New York and New Jersey for about 1,000 miles and a capacity of up to 2.6 million MMBtu per day) through the Algonquin Incremental Market (AIM) project and the Atlantic Bridge project. Once additional natural gas supplies from the Appalachian basin reach the states of New York and New Jersey, the two proposed expansions would allow gas to flow northward, satisfying demand in New England and Maritime provinces. The AIM Project is under regulatory review, while the Atlantic Bridge Project is in the early evaluation stages. 19 Kinder Morgan, 20 Kinder Morgan, 11

12 volumes from Pennsylvania to the Niagara region in northern New York. If approved, full operation is scheduled at the time of this writing for November III. Economic Modeling of the Atlantic Sunrise Expansion A. Arbitrage in the Presence of Constrained Pipeline Flows This section describes the economic model that we use to estimate the impact of the Atlantic Sunrise project on supply-demand balance along various points of the Transco system. While the process of price determination in natural gas markets is complex, the market itself is sufficiently mature that we can exploit two economic principles in the development of our model. First, if portions of Transco are unconstrained, we take advantage of the theory of arbitrage. (See, for example, Kleit (2001)). The theory of arbitrage states that if a product moves from Market A to Market B, and there is no shortage of transportation capacity between the two markets, the price in Market A plus the costs of transportation from Market A to Market B will equal the price in Market B. Thus, for example, assume that the price of natural gas in Market A is $5.00 per unit (MMBtu), we observe that gas shipped is shipped from A to B, and that the transportation cost from A to B is $0.30 per MMBtu. This implies that the price of gas in Market B will be $5.30. The intuition behind the theory of arbitrage is that if the price of gas in Market B rose above $5.30 per MMBtu, there would be profit opportunities in increasing shipments of gas from Market A to Market B. The influx of new supplies to Market B would have a depressing effect on prices in that market. Second, if the pipeline is constrained (here, east of Station 90 in Alabama) we can assume that the flow through Transco at that point is fixed. We note that because pipeline operating conditions change from day to day, that fixed amount will vary. (See Ruff (2012)). In our data, for example, we assume that the Transco system is constrained in Alabama by observing price differentials between Station 90 and Zone 5 that are not consistent with the theory of arbitrage. During periods where we observe constraints in the Transco system, we will fix flows at the constrained point at the observed level for each day. The Transco system is divided into zones for rate purposes. For the purposes of this analysis, the Alabama and Georgia parts of Transco are in Zone 4; South Carolina, North Carolina and Virginia are in Zone 5; Maryland and states to the northeast are in Zone 6 (See Figure 3.1 below). The Federal Energy Regulatory Commission (FERC) regulates the transmission rates between and within those zones. Station 195, which represents the southern terminus of the Atlantic Sunrise expansion (i.e., where Atlantic Sunrise would connect with the main Transco line), is on the western side of Zone In December 2013, TGP successfully completed an open season for an incremental 500,000 MMBtu per day on its system. The Utica Backhaul project, which began service on April 1, transports natural gas from the Marcellus and Utica production areas to the U.S. Gulf Coast. 12

13 Figure 3.1. Transco Zones Map Source: Using the principle of arbitrage and our assumption that flows are fixed at observed levels at constrained portions of the Transco system, we can describe a formal economic model for price determination at various points along the Transco system. We will let P j be the price in Zone j after the beginning of operations on the Atlantic Sunrise expansion. Thus, P 4, P 5, P 6 are the prices in zones 4, 5, and 6 respectively. (P 4 refers to prices in Zone 4 east of the potential bottleneck at Station 90.) We observe pre-atlantic Sunrise prices in Zone 5 and Zone 6. We also observe prices at Station 90, which we denote P 90. FERC regulated transportation rates for firm transportation service typically consist of two parts. The first is a reservation, or fixed, fee. The second charge is a usage fee, based on the variable costs to provide the transportation service. In addition, shippers are subject to a charge for fuel and line loss make-up -- a small percentage of the gas used in connection with the compression necessary to move the gas. [Thus, in practice, the transportation costs one way between zones will be slightly different than the transport costs the other way, depending on gas costs in each zone. In our model, we will define T jk to be the transportation cost from Zone j to Zone k. 13

14 Let Q 90,before equal the quantity flowing east out of Station 90 prior to Atlantic Sunrise and Q 90,after be the amount flowing east of Station 90 after Atlantic Sunrise. Let T 45 be the transport cost between zones 4 and 5, and T 54 be the transport cost between Zone 5 and 4. Let T 56 be the transport cost between zones 5 and 6, and T 65 be the transport cost between Zone 6 and 5. Let T 195,5 be the transport cost between Station 195 and Zone 5. Let T 195,6 be the cost of shipping gas from Station 195 to any point in Zone 6. Let Q 195,N equal the amount of gas flowing north from Station 195, with Q 195,S equal the southward flow, with Q 195,S 0. (These terms refer to the equilibrium amounts after Atlantic Sunrise goes into operation.) We consider how the Atlantic Sunrise Expansion would affect equilibrium prices and flows at three points on the Transco system: 1. The potential bottleneck east of Station 90; 2. The Zone 4/Zone 5 border (at Station 135, on the east side of the Georgia/South Carolina border); 3. Station 195, the injection point for the Atlantic Sunrise project, on the western side of Zone 6 in Maryland. 1. Equilibrium Price and Flow East of Station 90 If the flow east of Station 90 is constrained after the Atlantic Sunrise project, our assumption of fixed flows in the presence of constraints suggests that: (1) Q 90,before = Q 90,after In this case, the flow on the Transco is fixed. We note that allowable flows change from day to day. We will set the post-atlantic Sunrise flow equal to the pre-atlantic Sunrise flow on the relevant days in these circumstances. In the case of constrained flow east of station 90, we must have the following relationship between prices at station 90 and in Zone 4: (2) P 90 < P 4 Because of the constraint east of Station 90, arbitrage is not possible between Station 90 and the rest of Zone 4, and prices in the rest of Zone 4 are higher than at Station 90. On the other hand, if the flow east of Station 90 is unconstrained, prices will be equal across Zone 4, or: (3) P 90 = P 4 In this circumstance, the new flow from Station 195 will displace some or even all of the Station 90 flow, implying: (4) Q 90,before > Q 90,after 14

15 Finally, there are circumstances where there is no null point on Transco east of Station 90. In that case, the flow into Station 90 moves from east to west. If we define west-east flows with positive numbers and east-west flows with negative numbers, we would thus have: (5) Q 90,after 0 For all other circumstance, the Station 90 flow goes from west to east. 2. Equilibrium Prices and Flows Across the Zone 4/Zone 5 Border Three possibilities for flows across the border between Zones 4 and 5 exist: Flows north and east from Zone 4 to Zone 5; Flows south and west from Zone 5 to Zone 4; No cross-border flows. In the first case, gas flows north and east from Zone 4 to Zone 5. In this case, owners of gas will be indifferent between selling their gas in Zone 4 and selling it in Zone 5. This implies: (6) P 4 + T 45 =P 5 For example, assume that the price of gas in Zone 4 is $5.70 and the cost of transportation of gas from Zone 4 to Zone 5 is $0.30, and that gas flows from Zone 4 to Zone 5. This implies that the price of gas in Zone 5 will be $5.70+$0.30=$6.00. In the second case, gas could flow south and west across the Zone 4/5 border. This implies gas owners in Zone 5 are indifferent between selling in Zone 5, or paying the transport cost and selling in Zone 4, or: (7) P 5 + T 54 =P 4 The third circumstance is when no gas is shipped across the Zone 4/5 border. This implies that owners of gas in Zone 4 would prefer to sell it in Zone 4, while owners of gas in Zone 5 would prefer to sell the gas in Zone 5. In this case, the transportation costs are greater than any price difference. Therefore, P 4 +T 45 >P 5 (no shipments from Zone 4 to Zone 5) and P 5 +T 54 >P 4 (no shipments from Zone 5 to Zone 4). This implies: (8) P 4 -T 54 <P 5 <P 4 +T 45 For example, assume that the price in Zone 4 is $4.00, the price in Zone 5 is $4.10, and the transport cost between zones is $0.30. In this case, gas owners prefer not to ship gas between the two zones, and autarky (a state of no trade) exists between the zones. 15

16 3. Equilibrium Prices and Flows at Station 195 The impact of Atlantic Sunrise on flows at Station 195 is complex. We identify five different possibilities in equilibrium: Gas injected at Station 195 flows both north and south; Gas injected at Station 195 flows northbound only, and additional flows from Zone 5 to Zone 6 are facilitated by the pipeline expansion; Gas injected at Station 195 flows northbound only, but no additional flows occur from Zone 5 to Zone 6; Gas injected at Station 195 flows southbound only, and additional flows from Zone 6 to Zone 5 are facilitated by the pipeline expansion; Gas injected at Station 195 flows southbound only, but no additional flows occur from Zone 6 to Zone 5; Each of these equilibria has specific conditions under which they will occur. These conditions are described below for each of the five equilibria. a. The Injected Gas Flows Both North and South If gas injected at Station 195 goes both north and south, under the theory of arbitrage, owners of natural gas injected at Station 195 should be indifferent between their gas going north or south. This implies that the net return from northbound flow must be equal to the net return from southbound flow. In addition, both northward and southward flows are possible from Station 195. (9) P 6 T 195,6 = P 5 T 195,5 (10) -1.7 million MMBtu Q 195,S 0 Q 195,N 1.7 million MMBtu For example, assume the price in Zone 5 is $6.00, the cost of transporting gas from Station 195 to Zone 5 is $0.20 per MMBtu, while the cost of transporting cost from Station 195 to Zone 6 is $0.15. Moreover, assume that injected gas at Station 195 goes both north and south. Under the arbitrage assumption, gas owners at Station 195 are indifferent between sending their gas north to Zone 6, or South to Zone 5. Thus, the prevailing prices in Zones 5 and Zone 6 would need to satisfy (9) above, or P 6 - $0.15 = $ $0.20, implying that the price of gas in Zone 6 was $5.95. b. All of the Injected Gas Flows North There are two scenarios here. Assume that all gas injected at Station 195 goes north, and additional gas flows on Transco from Zone 5 to Zone 6. In this scenario, the price of gas in Zone 6 minus the transportation cost from Station 195 to Zone 6 will be greater than the price of gas in Zone 5, net of the transportation cost from Station 195 to Zone 5. (11) P 6 T 195,6 P 5 T 195,5 16

17 For example, if the price in Zone 6 is $5 and the two transport costs both equal $0.25, (11) implies that the price in Zone 5 must be less than $5. Assume now that gas from Zone 5 also flows into Zone 6, implying: (12) Q 195,N >1.7 million MMBtu Gas flowing north from Zone 5 to Zone 6 implies that owners of gas in Zone 5 are indifferent between selling in Zone 5, or paying the transport cost and selling in Zone 6, or: (13) P 5 + T 56 =P 6 For example, assume that transport cost between Zone 5 and Zone 6 is $0.35 per unit. Given a price in Zone 5 of $6.00, this implies the price in Zone 6 will be $6.35. Alternatively, all of the gas from Station 195 flows north, but no gas from Zone 5 flows into Zone 6. In this case, condition (11) still applies. However, the northward flow at Station 195 equals the injections from Atlantic Sunrise at 195, or: (14) Q 195,N =1.7 million MMBtu In addition, Zone 5 gas owners prefer not to send their gas to Zone 6, or: (15) P 5 + T 56 >P 6 In this case, we say that Zone 6 is in autarky with respect to Zone 5. c. All of the Injected Gas at Station 195 Flows South There are again two scenarios here. Assume that all of the gas injected at Station 195 goes south. This implies: (16) P 6 T 195,6 P 5 T 195,5 and that the payoff to sending gas south is greater than the payoff to sending gas north. Given that all the injected gas flows south, it is also possible that Zone 6 gas would also flow south. In this circumstance, we have that: (17) P 6 + T 65 =P 5 (18) Q 195,S <-1.7 million MMBtu In this case, the Zone 6 price plus the cost of transport from Zone 6 to Zone 5 will equal the Zone 5 price. 17

18 Alternatively, no gas would flow south from Zone 6. This implies: (19) P 6 + T 65 >P 5 (20) Q 195,S =-1.7 million MMBtu Once again, Zone 6 would be in autarky. The possible equilibrium conditions that we found are summarized in Tables below. (Numbers in the boxes refer to the relevant equilibrium/arbitrage condition. NA indicates that the relevant circumstance did not arise in our simulations.) 18

19 Pipeline Previously Unconstrained East of Station 90 Scenarios In all these scenarios Conditions 3: P 90 = P 4 ; and 4: Q 90,before > Q 90,after apply Table 3.1 Station 195 gas flows either both North and South, or only North to Zone 6 Gas flows from Zone 4 to Zone 5 Gas flows from Zone 5 to Zone 4 and the null point is in Zone 4 Gas flows from Zone 5 to Zone 4 and the null point is to the southwest of Station 90 No gas flows between Zone 4 and Zone 5 Gas flows both South to Zone 5 and North to Zone 6 6: P 4 + T 45 =P 5 9: P 6 T 195,6 = P 5 T 195,5 10:-1.7 million MMBtu Q 195,S 0 Q 195,N <1.7 million MMBtu 7: P 5 + T 54 =P 4 9: P 6 T 195,6 = P 5 T 195,5 10:-1.7 million MMBtu Q 195,S 0 Q 195,N <1.7 million MMBtu 5: Q 90,after <0 7: P 5 + T 54 =P 4 9: P 6 T 195,6 = P 5 T 195,5 10:-1.7 million MMBtu Q 195,S 0 Q 195,N <1.7 million MMBtu 8: P 4 -T 54 <P 5 <P 4 +T 45 9: P 6 T 195,6 = P 5 T 195,5 10:-1.7 million MMBtu Q 195,S 0 Q 195,N <1.7 million MMBtu Gas flows only North to Zone 6 6: P 4 + T 45 =P 5 11: P 6 T 195,6 P 5 T 195,5. Gas flows from Zone 5 to Zone 6: 12: Q 195,N >1.7 million MMBtu 13: P 5 + T 56 =P 6 No gas flows from Zone 5 to Zone 6: 14: Q 195,N =1.7 million MMBtu 15: P 5 +T 56 >P 6 NA NA NA 19

20 Table 3.2 Station 195 gas flows only South to Zone 5 In all these scenarios Condition 16: P 6 T 195,6 P 5 T 195,5 applies Gas flows from Zone 4 to Zone 5 Gas flows from Zone 5 to Zone 4 and the null point is in Zone 4 Gas flows from Zone 5 to Zone 4 and the null point is to the southwest of Station 90 No gas flows between Zone 5 and Zone 4 Gas flows from Zone 6 to Zone 5 6: P 4 + T 45 =P 5 17: P 6 + T 65 =P 5 18: Q 195,S <-1.7 million MMBtu 7: P 5 + T 54 =P 4 17: P 6 + T 65 =P 5 18: Q 195,S <-1.7 million MMBtu 5: Q 90,after <0 7: P 5 + T 54 =P 4 17: P 6 + T 65 =P 5 18: Q 195,S <-1.7 million MMBtu 8: P 4 -T 54 <P 5 <P 4 +T 45 17: P 6 + T 65 =P 5 18: Q 195,S <-1.7 million MMBtu No gas flows from Zone 6 to Zone 5 6: P 4 + T 45 =P 5 19: P 6 + T 65 >P 5 20: Q 195,S =-1.7 million MMBtu 7: P 5 + T 54 =P 4 19: P 6 + T 65 >P 5 20: Q 195,S =-1.7 million MMBtu NA 8: P 4 -T 54 <P 5 <P 4 +T 45 19: P 6 + T 65 >P 5 20: Q 195,S =-1.7 million MMBtu 20

21 Pipeline Previously Constrained East of Station 190 Scenarios Table 3.3 Gas flows from Zone 4 to Zone 5 Gas flows from Zone 4 to Zone 5 Gas flows from Zone 4 to Zone 5 Gas flows from Zone 4 to Zone 5 Station 195 Scenario: gas flows North from Injection Point to Zone 6; Gas flows from Zone 5 to Zone 6 Constraint remains after the injection at Station 195 Constraint is eliminated after the injection at Station 195 1: Q 90,before = Q 90,after 3: P 90 = P 4 2: P 90 < P 4 4: Q 90,before > Q 90,after 6: P 4 + T 45 =P 5 6: P 4 + T 45 =P 5 11: P 6 T 195,6 P 5 T 195,5 11: P 6 T 195,6 P 5 T 195,5 12: Q 195,N >1.7 million 12: Q 195,N >1.7 million MMBtu MMBtu 13: P 5 +T 56 =P 6 13: P 5 +T 56 =P 6 Station 195 Scenario: Gas flows South, no flow from Zone 6 to Zone 5 Constraint remains after the Constraint is eliminated after injection at Station 195 the injection at Station 195 3: P 90 = P 4 6: P 4 +T 45 =P 5 NA 16: P 6 T 195,6 P 5 T 195,5 19: P 6 +T 65 >P 5 20: Q 195,S =-1.7 million MMBtu Station 195 Scenario: Gas flows north from injection point to Zone 6; No gas flows from Zone 5 to Zone 6 Constraint remains after Constraint is eliminated after the injection at Station 195 the injection at Station 195 1: Q 90,before = Q 90,after 3: P 90 = P 4 2: P 90 < P 4 4: Q 90,before > Q 90,after 6: P 4 + T 45 =P 5 6: P 4 + T 45 =P 5 11: P 6 -T 195,6 P 5 -T 195,5 11: P 6 -T 195,6 P 5 -T 195,5 14: Q 195,N =1.7 million 14: Q 195,N =1.7 million MMBtu MMBtu 15: P 5 +T 56 >P 6 15: P 5 +T 56 >P 6 Station 195 Scenario: Gas flows from the injection point both south to Zone 5 and north to Zone 6 Constraint remains after Constraint is eliminated after the injection at Station 195 1: Q 90,before = Q 90,after 2: P 90 < P 4 6: P 4 + T 45 =P 5 9: P 6 T 195,6 = P 5 T 195,5 10:-1.7 million MMBtu Q 195,S 0 Q 195,N <1.7 million MMBtu the injection at Station 195 3: P 90 = P 4 4: Q 90,before > Q 90,after 6: P 4 + T 45 =P 5 9: P 6 T 195,6 = P 5 T 195,5 10:-1.7 million MMBtu Q 195,S 0 Q 195,N <1.7 million MMBtu 21

22 In all, there are 19 equilibrium scenarios described above. 22 We note that it is not directly possible to solve for which model applies. Rather, it is necessary to assume which equilibrium applies, solve out the relevant parameters, and see if the necessary conditions are met. In our modeling, we have also been able to solve for the null point on Transco. The null point is where the southward flows (from either Station 195 or the end of Transco in northern New Jersey) meet the northward flow coming from Station 90. In some circumstances, the null point is actually at or to the west of Station 90, as the flow across Zone 4 is westward, rather than eastward. If all of the gas injected at Station 195 flows north and east, that implies that the null point is in Zone 6. B. The Supply and Demand Model In this section we outline the conditions used to set supply equal to demand in all zones after Atlantic Sunrise. Combined with the arbitrage conditions discussed above, these conditions will allow us to model the economic impact of injecting 1.7 million MMBtu of natural gas per day at Station 195. We first assume that a perfectly elastic supply of gas is available at Station 90 at a constant price. This infinite elasticity assumption is reasonable to the extent that gas sent through the Transco at that point is a relatively small amount of the total gas produced in Southwestern stateswe also assume that the price of gas in the Marcellus Region is always below that of the price at Station 195 minus the costs of transport to Station 195 from central Pennsylvania. This is because there is currently a great deal of stranded gas in the Marcellus, so much that even an additional 1.7 million MMBtu per day will not fully alleviate the excess production problem. Demand at any withdrawal point takes the constant elasticity form, Q id = A id P i ED i, where Q id is the withdrawal amount at point i, P i is the price of natural gas at this point, A id is a point specific constant and ED i is the elasticity of demand of point i. We categorize all the withdrawal points into three categories: Local Distribution Companies (LDCs), Power Plants, and Other Industry. LDCs consists of two types of customers: Residential and Commercial. The Energy Information Administration (EIA) 23 estimates short run U.S. natural gas price elasticities for residential, commercial, industrial and electric power to be , , and respectively. The EIA reports that natural gas consumption by residential and commercial 22 In theory, there are many more potential equilibria possible than listed in the table above. The equilibria listed here are the ones we found in our simulations. 23 EIA, Reduced Form Energy Model Elasticities from EIA s Regional Short-Term Energy Model (RSTEM), 22

23 customers in 2013 was trillion and trillion cubic feet respectively. 24 averaging, we calculate a price elasticity of demand for LDCs of Using weighted Each injection point has supply function from the sources listed in Table A-1 of Q is = A is P i ES i, where Q is is the injection amount at each point, P i is the price of natural gas at that point, A is is a point specific constant and ES i is the elasticity of supply, assumed to be equal to 0.76 (Ponce and Neumann, 2014). 25 Based on the withdrawals and injections data, we calculate the point specific demand and supply parameters A id and A is. A list of injection points is given in Table A-1. A summary of the other suppliers in given in Table 3.4 below. Table 3.4 Summary of Other Suppliers on Transco Zone Number of Suppliers Total Injections by Zone (MMBtu) on January 28, 2014 Zone ,568 Zone 5 4 1,048,981 Zone ,663,258 Arbitrage price differences between zones are governed by interruptible (IT) rates on Transco. Between Zone 4 and Zone 5 the IT rate is 36 cents plus 1.28% of gas costs that day. We will assume gas costs here imply the gas costs in the originating zone. To simplify our calculations, we will assume gas costs are based on pre-atlantic Sunrise prices. 26 From Zone 5 to Zone 6, the IT rate is 0.77% of gas costs plus 26 cents. From Zone 6 to Zone 5, the current IT rate is 26 cents. Currently, there is no charge based on the cost of gas from Zone 6 to Zone 5. However, we expect that one will be imposed by FERC, once gas starts physically flowing south on Transco: we assume it will also be 0.77% of gas costs. Gas flowing north from station 195 pays the Zone 6 intra-zonal rate, which is 14 cents. Gas flowing south from station 195 pays the Zone 6 to Zone 5 rates, 0.77% of gas costs plus 26 cents. Transportation rates are summarized in Table 3.5 below, assuming a market cost of $4.00/MMBtu We note that this is likely to be a very high number for elasticity of supply. This estimate is taken from well-head production of natural gas, not pipeline supply. In practice, the elasticity of supply from many pipelines may well be zero, as they may also be capacity constrained. Because the lower the elasticity of supply the greater the price effects of Atlantic Sunrise, this assumption serves to reduce our estimates of the actual consumer benefits of the expansion project

24 Table 3.5 Interruptible Transportation Rates assuming a gas price of $4.00/MMBtu To From Fixed Fee/MMBtu Percent of Gas Costs Total Rate/ MMBtu Zone 4 Zone 5 $ % $0.411 Zone 5 Zone 6 $ % $0.291 Station 195 Zone 6 $0.14 0% $0.14 Station 195 Zone 5 $ % $0.291 We have two sets of data. The first is composed of flows, injections and withdrawals along Transco from January 1, 2012 to June 27, The second includes price data supplied by Williams. We have prices at Station 90 ( Transco-85 ), Zone 5 ( non-wgl Transco Z5, near the Virginia/North Carolina border) and Zone 6 ( TETCO-M3, near Philadelphia). Our model allows us to calculate the change in consumer surplus as a result of the Atlantic Sunrise project. This represents both the saving consumers receive through lower gas prices and the benefits they gain because they purchase more gas, since prices are lower. Given a demand function of the constant elasticity form, the change in consumer surplus at a point i is expressed by the following formula: (21) ΔConsumer Surplus i = A id 1 ED (P ia 1 ED P i B 1 ED ) where P i A is the zonal price before Atlantic Sunrise project, and P i B is the price after Atlantic Sunrise project. We calculate the change in consumer surplus at each milepost. We use the withdrawals at each out of system milepost, injections at each into system milepost and the price before the Atlantic Sunrise project to calculate the constants A at each milepost. Using equation (21), we obtain the change in consumer surplus at each milepost. There are some important caveats here. First, it is a distinct possibility that there are some days that consumer surplus will decline in Zone 6. This is because there are days where gas is significantly less expensive in Zone 6 than Zone 5, and, without Atlantic Sunrise, this gas cannot flow south. Allowing a southward flow on Transco relieves the underlying bottleneck and allows gas to flow from Zone 6 to Zone 5, potentially resulting in higher prices in Zone 6. Second, in the data we often observe uneconomic flows of gas between zones. For example, on July 28, 2013, the price in Zone 4 is $3.54, the Zone 5 price is $3.61, the transportation cost from Zone 4 to Zone 5 is $0.41, and yet there are substantial flows of gas from Zone 4 to Zone 5. We assume that these uneconomic flows are the result of difficulties of renegotiating long-term contracts between parties. Our model assumes that these uneconomic flows will not continue following the construction of Atlantic Sunrise. In reality, these 24

25 uneconomic transactions may continue but there is no way to predict the extent to which they will do so. We take three modeling steps to address this issue. These steps are both conservative in nature (i.e., lower our estimates of consumer benefit) and consistent with the economic modeling assumptions. First, in circumstances where the resulting change in Zone 5 consumer surplus is negative, we set the change in consumer surplus to zero. We note that there is no economic reason to believe that the Atlantic Sunrise project would cause Zone 5 prices to rise (since Atlantic Sunrise would, in effect, increase supply deliverability into Zone 5). This assumption serves to reduce our estimate of consumer benefits on days where the Zone 5 price would have declined, because the decline is from an uneconomically low level. Second, on days where there are no shipments for gas from Zone 6 to Zone 5, and the estimated consumer surplus change in Zone 6 is negative, we set the change in Zone 6 consumer surplus to zero. Again, there is no reason to believe in these circumstances that Zone 6 consumer surplus would decline, because all gas sales in Zone 6 would effectively serve customers within Zone 6. Finally, on days where we estimate that Atlantic Sunrise would induce shipments of gas from Zone 6 to Zone 5, we allow estimated consumer surplus in Zone 6 to decline. On these days, there are economic reasons to believe Zone 6 prices would rise, since buyers in Zone 5 would be willing to pay higher prices for gas supplied from Zone 6 than gas supplied from Zone 5. Again, this makes our estimates of the total consumer gains from Atlantic Sunrise conservative, as we do not take into account any adjustments for uneconomic flows. In addition, there are days where our model estimates that gas would have flowed into Station 90. These supplies serve to displace shipments of gas eastward from Texas and Louisiana, and result in additional supplies going to other (non-transco) markets. We do not estimate the benefits of these shipments to customers in other markets, but it is worth noting that the benefits of additional gas transportation infrastructure would not be limited to markets served directly by Transco. IV. Illustration of the Impacts of Atlantic Sunrise on Five Different Days To illustrate how our economic model can be used to estimate the market impacts of the Atlantic Sunrise expansion, we choose five days in different seasons (February 11, 2013, July 28, 2013, October , January 28, 2014 and April 28, 2014) and solve for the market equilibrium (injections, withdrawals and prices in each zone), given the establishment of the Atlantic Sunrise project. These days represent different demand/supply balances along the Transco system, and thus allow us to describe the market impacts of Atlantic Sunrise under various system conditions. 25