F530

Contents lists available at ScienceDirect

Energy Policy

journal homepage: www.elsevier.com/locate/enpol

Natural gas and spillover from the US Clean Power Plan into the Paris Agreement

Jeffrey C. Peters

James S. McDonnell Postdoctoral Fellow in Studying Complex Systems, Department of Management Science & Engineering and Energy Modeling Forum, Stanford University, United States

A R T I C L E I N F O

Keywords: Natural gas Clean Power Plan Bridge fuel Fugitive emissions Paris Agreement Electric power

A B S T R A C T

Climate change has been identified as one of the today's great challenges, and mitigation likely requires policy intervention. As such, in 2015 the United States introduced the Clean Power Plan (CPP) which aims to reduce CO2 emissions from electricity production 32% from 2005 levels by 2030 and the Paris Agreement, which seeks to reduce national greenhouse gas (GHG) emissions, measured by global warming potential (GWP), 28% from 2005 levels by 2025. However, it remains unknown how the more narrowly-scoped CPP might affect the ability to achieve wider-scoped national GHG targets like the Paris Agreement. In our current state-of-world, characterized by inexpensive natural gas, the CPP will be met through large shifts from high-emitting coal power to less-emitting natural gas power, which translates to a 9.6% reduction in total US 100-year GWP without accounting for the fugitive methane. Spillover from fugitive methane could cut this reduction modestly by 0.2–1.4% or as much as 4.4% if evaluated using 20-year GWP – elucidating how different assumptions leads to different perspectives of natural gas as a "bridge fuel". The results here demonstrate the need to coordinate policies – either through additional policy (e.g. regulation of fugitive methane) or a larger-scoped CPP that includes upstream activities.

1. Introduction

In 2015, the United States Environmental Protection Agency (EPA) announced the Clean Power Plan (CPP) with a projection to reduce CO2 emissions from the electric power generation 32% from 2005 levels by 2030. The EPA identified: i) improving the heat rate of existing coal plants, ii) substituting gas power in place of coal power at existing plants, and iii) substituting zero-emitting renewable power in place of coal power as the three best system building blocks for achieving the 32% reduction target (Clean Power Plan, 2015).1

The second building block is a key mechanism, because gas combustion emits approximately half the CO2 emissions of coal combustion in electricity generation. Following the sharp decline in gas prices as a result of the US shale boom, several studies indicate that the most economic way of meeting the nationwide emission target is, at least in part, via fuel-switching from high-emitting coal power to lower- emitting natural gas as well as further capacity expansion in gas power. In fact, this transition can be observed prior to the CPP in data

following the fall in gas prices in 2008–2009 (see Fig. 1). The fall in natural gas price raised the idea of gas as a "bridge fuel"

to a low-carbon electricity future, with much debate (Kerr, 2010; Howarth et al., 2011; Levi, 2013; Shearer et al., 2014; Howarth, 2014; Davis and Shearer, 2014). While this debate continues, Fig. 1 demon- strates that relatively inexpensive natural gas has already led to fuel- switching from coal power to gas power as well as a decline in total electricity sector CO2 emissions. When measured by CO2 emissions in the electricity sector, it is reasonable to state the natural gas is, at least, a short-term bridge fuel.

Because natural gas (i.e. methane) is 86 times more potent than CO2 in terms of GWP over a 20-year period and 34 times more potent over a 100-year horizon (Myhre et al., 2013), one of the more recent concerns about increased gas power production is the accompanying methane emissions from extraction and transmission that occur prior to combustion in the power plant.2 While it is difficult to generalize pipeline specifications serving gas power plants across the entire United States, it stands to reason that increased demand for gas power

http://dx.doi.org/10.1016/j.enpol.2017.03.039 Received 19 August 2016; Received in revised form 17 March 2017; Accepted 18 March 2017

E-mail address: [email protected]. 1 Demand-side management is another key mechanism for reducing electricity sector emissions. This study uses EIA Annual Energy Outlook projections of future electricity demand

(EIA, 2016b), which treats energy efficiency implicitly. 2 Aerial observations have shown that atmospheric methane has increased over the past decade; however, isotope signatures indicate that the increase may not be attributable to oil

and gas extraction (Schwietzke et al., 2016; Schaefer et al., 2016; and Nisbet et al., 2016). These studies assume that methane isotopes from shale and conventional sources are identical; however, Golding et al. (2013) suggest that isotopes from shale sources may be slightly different.

Energy Policy 106 (2017) 41–47

Available online 22 March 2017 0301-4215/ © 2017 Elsevier Ltd. All rights reserved.

MARK

will not contribute significant additional methane leakage across the distribution network (i.e. the part of the network that serves munici- palities and households), which accounts for approximately 20% of methane leakage (EPA, 2014).3 Estimates of fugitive emissions from production, gathering, boosting, processing, transmission, and storage range from 1.1% (EPA, 2016) to 5.6% (the mean of the high-end of conventional and shale gas development minus distribution from Howarth et al., 2011) of total production, with a majority of studies settling slightly above the low-end EPA value (Brandt et al., 2014; Lyon, 2016).4 In an effort to address fugitive emissions, the EPA established a federal rule seeking to reduce fugitive emissions by 40– 45% (New Sources Performance Standards, 2016). Whether this target will be met through the regulation remains to be seen.

Regardless of what the actual rate may be in 2030, these fugitive emissions would not fall within the current scope of the CPP, but are still relevant in climate change mitigation. Despite its limited scope, the CPP is cited as a major component of the US contribution to the United Nations' Paris Agreement, a more broadly-scoped target that seeks to reduce net US greenhouse gas (GHG) emissions, measured by 100-year global warming potential (GWP), by 28% from 2005 levels by 2025 (US Department of State, 2016).

It remains unknown how the more narrowly-scoped CPP (i.e. CO2 emissions in the US electricity sector) might spillover into wider- scoped national GHG targets like the Paris Agreement. A spillover effect is a situation where an activity has an unintended consequence on another seemingly unrelated activity. In terms of the CPP, the spillover is cross-sectoral in that emissions may increase in sectors outside the scope of the mitigation policy (i.e. outside the CO2 emissions in the electricity sector). This idea echoes arguments for life-cycle instead of production-based emission policies as well as for life-cycle accounting in the study of energy and economic systems (e.g. Burnham et al., 2011; Weber and Calvin, 2012).

The question here is: how will the CPP mechanisms, designed to reduce CO2 emissions in the electricity sector, affect broader climate change mitigation goals like the Paris Agreement considering the spillover from fugitive methane emissions? In the process of answering

this question this article provides: i) life-cycle emission analysis combined with energy-economic modeling of the electricity sector, ii) clarity in the debate about whether natural gas is or is not a bridge fuel, and iii) a consensual path forward that would help reduce policy spillover through measurement and policy adjustment to mitigate fugitive methane emissions.

Section 2 introduces the scenario planning framework and energy- economic model used to explore the interaction between the CPP and the Paris Agreement targets. The four scenarios explore technological contributions to total US electricity generation in 2030 under low and high natural gas prices, with or without CPP implementation. Section 3 describes the specific data and assumptions used to project technolo- gical contributions to 2030 US electricity production and the accom- panying emissions from both combustion and fugitive emissions. Section 4 discusses the scenario results and the fugitive emissions for the four scenarios. Section 5 draws conclusions from the analysis and suggests a path forward through measurement and policy that could be met with broad consensus to reduce spillover from the CPP into broader climate change mitigation objectives like the Paris Agreement.

2. Methodology

There is, of course, great uncertainty in answering the question of spillover. First, models that predict large shifts to gas power assume that current natural gas prices represent a new normal. This may not be the case due to price rebound effects, especially in the face of possible liquefied natural gas exports, or even moratoriums on horizontal drilling and hydraulic fracturing (i.e. fracking). Second, the CPP itself faces legal challenges in the Trump Administration, US Congress, as well as the US Supreme Court where it is, at the time of publication, put on hold. Third, the GWP of fugitive emissions depends on assumption regarding the emission rate, effectiveness of EPA regulation, and the time horizon of the analysis. The following sections describe the scenario-based analysis using an energy-economic model to project technological contributions to US electricity generation and the corre- sponding CO2 emissions from combustion as well as fugitive methane emissions from expanding gas infrastructure.

2.1. Scenario analysis

The uncertainty in the spillover question is well-suited for scenario planning where different “states-of-the-world” are simulated in the same modeling regime in order to tease out the important mechanisms and assumptions leading to different projections of the future. The four scenarios here assume pre- or post-shale boom gas prices (2007 and

Fig. 1. US electricity sector CO2 emissions drop because of fuel-switching from coal to gas power as a result of falling gas prices following the shale gas boom in 2008. Source: EIA, 2016a.

3 Continued low gas prices could lead to increased gas demand in households, businesses, and industries where total fugitive emission rates could be higher because these sectors us the distribution network. This impact is not explored in this particular work.

4 Schneising et al. (2014) suggests fugitive emissions from shale plays could be much higher than the high-end used in this article; however, this estimate seems to be an outlier from the wide body of literature that fits into the range used here. The implications of using the Schneising et al. (2014) estimate is straightforward in the results (that is, more fugitive emissions than the high-end presented here).

J.C. Peters Energy Policy 106 (2017) 41–47

42

2014 prices, respectively) in combination with or without CPP im- plementation using an effective CO2 tax on combustion.

Table 1 shows the underlying future perspective of these four scenarios. The high gas price (HG-) scenario uses the 2007 price of natural gas, which can be interpreted as either that the shale gas boom is a short-lived phenomena or that state or federal moratoriums ban the practice of fracking. This could be described as the "past as future" scenario, where the impact from the shale gas boom quickly dissipates. The low gas price (LG-) uses the 2014 gas price and can be interpreted as an optimistic outlook for the price of shale gas as the "new normal" due to sufficient reserves, a high elasticity of supply from the reduced cost and timeline of drilling unconventional wells, and continued technological advance in drilling. The business-as-usual (-BAU) sce- narios assume that coal regulation and investment tax credits for wind and solar are maintained to 2030, while the CPP (-CPP) scenarios use an additional tax on CO2 to meet the 32% reduction target of the CPP.

Section 2.2 describes the model used to project technological contributions to electricity production and emissions in 2030. Section 3 describes the additional data and assumptions that are used in the model to study the four scenarios.

2.2. Electricity sector and emission projections

These four scenarios are analyzed using a non-linear partial equilibrium energy-economic model that projects US electricity gen- eration by technology and their associated emissions (Peters and Hertel, 2017). The set of technologies included in the model are: nuclear, coal, gas base load (e.g. combined-cycle), gas peak load (e.g. combustion turbine), oil, hydroelectric, wind, solar, and other (pri- marily consisting of geothermal and waste) power. Short-run changes in utilization with existing capacity in response to prevailing economic conditions, such as the gas price drop, are tied to longer-run capacity additions and retirements via returns to capital making it an ideal model to study the technological contributions to total generation.

Utilization is defined here as the substitution of electricity genera- tion between existing capacity, and is a function of the ease that dispatchable technologies can substitute for one another in the existing system and dispatchability – the ability of a generating technology to schedule operations. Utilization of each technology can adjust by substituting operating and maintenance (O & M) costs for capacity – represented by a constant elasticity of substitution (CES) parameter. Dispatchable technologies (i.e. coal, gas, oil, and 'other' power) have high CES parameters, while non-dispatchable technologies have a parameter value of zero indicating these technologies cannot adjust utilization (i.e. nuclear, hydroelectric, wind, and solar power). Fuels are represented as a Leontief input – i.e. demanded in equal proportion to utilization demand.

The derived demand for utilization for each technology is deter-

mined separately for base and peak load markets where nuclear, coal, hydroelectric, gas base load, wind, and other (primarily consisting of geothermal and waste) power are base load technologies and gas peak load, oil, and solar are peak load technologies. The purpose of the separation into base and peak load substitution nests is to represent specific operational differences inherent in technologies. Utilization within each market is given by the following equations for base and peak load markets, respectively. The equations are shown in log-linear (i.e. percentage change) form – the same manner they are represented in the non-linear GEMPACK model (Harrison et al., 2014).

c C σ p P= − ( − )bl g bl

bl bl g bl

(1)

and

c C σ p P= − ( − )pk g pk

pk pk g pk

(2)

The variables, cbl g and cpk

g are the percentage changes in utilization and pbl

g and ppk g are the percentage changes in generation costs for base and

peak load markets, respectively. The variables C bl and C pk are the percentage change in aggregate utilization for base and peak load technologies, respectively, and Pbl and P pk are the percentage change in the aggregate, volume-weighted price indices for base and peak load technologies, respectively. The parameters σbl and σpk are additive constant elasticities of substitution (ACES) parameters that represent the ability to substitute technologies using existing capacity in the base and peak load markets, respectively, and is a system-level characteristic. Peters and Hertel (2017) argue that the ACES specification, which minimizes a disutility of costs, is well-suited for electric power utilization because simple cost minimization may ignore important considerations of the electricity sector, namely reliability of supply. Further, the ACES specification ensures that the generation from all the individual technol- ogies equals to the total generation; the traditional CES does not and deviations can be large (van der Mensbrugghe and Peters, 2016).

In the case of inexpensive gas, utilization of gas increases at the expense of coal and oil power. Because construction can take several years and there is enormous sunk capacity, capacity supply is highly inelastic in the short-term and increasingly elastic in the long-term. Therefore, increasing gas utilization drives returns to capacity for gas power upward until all the necessary capacity replacements occur. The case is similar for renewables. Existing investment tax credits for wind and solar power increase the returns to capacity for these technologies, and these technologies also expand in the long-run.

The returns to capacity can affect the economic lifetime of capacity; increasing returns extends the technical lifetime, while decreasing returns retires the existing fleet at a faster rate. Retirements are difficult to capture and are calibrated based on observed rates and price vectors, rather than captured endogenously in the model. Shares of capacity expansion is also a function of rates of return to capacity, given by the following multinomial logit model.

s e

e =

∑ t c

U α p a t

s T U α p a t

( , , , )

∀ ∈ ( , , , )

t t c

t c

t c

s s c

s c

s c

(3)

where st c is the share of new capacity contributed by technology t

among the set of technologies, T . The variable Ut is the utility of capacity t and is a function of multinomial parameter (α), returns to capacity (pt

c), technical efficiency changes for new capacity (at c), and

taxes on new capacity (tt c).

The analysis here necessarily relies on projections of technological contributions to generation in 2030. Peters and Hertel (2017) provides full documentation of the model equations, parameters (all of which are adopted here unless otherwise stated), as well as through validation exercises.5 The validation exercises demonstrate confidence, as mea-

Table 1 Possible future states for scenario-based planning: low vs. high gas price; CPP vs. no CPP implementation.

5 The full model and a sample experiment file are included in the supplemental material.

J.C. Peters Energy Policy 106 (2017) 41–47

43

sured by correlation, in the model's ability to reasonably predict observations of utilization and total generation from 2002 to 2012. The validation is extended for total CO2 emissions from 2007 to 2014 in Fig. 2 in Section 4.

The output of electricity sector model are the contributions from each technology to total electricity generation (in GWh). The CO2 emissions from combustion and CO2 equivalent from fugitive methane emissions (measure in GWP) flow directly from generation by each technology. The CO2 emissions from combustion is given by the following equations.

CO E Q2 = ⋅t t t g

(4)

where CO2t is the level of CO2 emissions from technology t, Et is the emission rate for technology t, and Qt

g is the level of electricity generation from technology t. The level of GWP from spillover of fugitive methane emissions is given by the following equation.

GWP r ρ η q= ⋅(1 − )⋅ ⋅ gas g

(5)

where r is the fugitive emission rate, ρ is the reduction in emissions from policy, η is the GWP conversion factor for methane based on time scale of analysis, and qgas

g is the level of receipts of methane for electricity production in metric tons. Gas receipts are directly derived from demand for gas power. That is, if gas power expansion increases gas fuel demand by 10%, fugitive emissions from gas infrastructure increases 10%. The proportionality assumption simplifies the problem, and may be an accurate estimate; DOE (2015) found that gas infrastructure expansion may be mitigated in part by regionally distributed gas power demand and gas resource supply with sufficient existing network capacity.

3. Data and scenario assumptions

The model's base year is 2007, just prior to the sharp decline in natural gas prices following the shale gas boom in order to capture both utilization and capacity expansion mechanisms of electricity generation adjustment. The model is shifted to future years from the 2007 base year given shifts in population, fuel prices (EIA, 2016a), labor costs (BLS, 2015) and technological change. Observed values are used to project 2007–2014, while additional assumptions are necessary to project beyond the observable range, 2015–2030.

The business-as-usual (BAU) scenarios use EIA Annual Energy Outlook 2016 projections of annual net generation (EIA, 2016b). The model endogenously determines the sector-level technological change necessary to meet this demand with the accompanying projections of population and income The CPP scenarios use this endogenously

derived technological change variable to project electricity generation while still preserving price effects from the CPP policy. Table 2 shows that for both HG- and LG- scenarios, technological change is positive, as expected.

Table 3 shows that differentiating assumptions between the four scenarios. Coal and oil prices are assumed to remain unchanged from 2014 levels out to 2030 in each scenario. The high (2007) and low (2014) gas price are $7.11/MMBTU and $5.00/MMBTU, respectively. These gas prices are accompanied by the corresponding coal capacity retirement rate that is observed with that gas price. Due to the lack of available data on fuel price sensitivity of retirements, Peters and Hertel (2017) recommend combining gas price with the observed coal retirement rate at that price level and report relatively low sensitivity to this assumption. Prior to 2008, coal capacity retired at a rate of 1.0% of total capacity, while the rate increased to 2.0% after the decline in gas prices – reflecting the expected change in economic lifetime. These price and retirement rate combinations are adopted for these projec- tions.

In the face of policy uncertainty and to make the CPP policy implementation tractable, we implement the CO2 tax on combustion required to drive the full 32% emissions by 2030 beginning in 2014 (the last year of available price data in EIA (2016a)). The CO2 tax would

Fig. 2. Million metric tons (MMT) of CO2 emissions from combustion in each of the four scenarios from 2015 to 2030. The model is used to project technological contributions to the observed total generation using observed fuel prices, subsidies, and technological efficiency from 2007 to 2014 to lend confidence in the model's output (black data points). The model performs well for gas, coal, and oil power, which account for 99.4% of total CO2 emissions from combustion in 2007, but poorly for "other" power, which account for the remaining 0.6%. The blue line is the CPP target, which is satisfied in the HG-CPP and LG-CPP scenarios in 2030. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Table 2 Projections of important variables from 2015 to 2030 in the model (percentage change from 2007 baseline). Qg from the EIA AEO 2016 is used to project electricity demand for the no CPP scenarios (i.e. HG-BAU and LG-BAU). Total sector efficiency, AHG BAU

g − and

ALG BAU g

− , are derived from their respective BAU scenarios and are used to project electricity demand with the CO2 tax to meet the CPP in the HG-CPP and LG-CPP scenarios, respectively.

Income Population O & M Qg AHG CPP g

− ALG CPP g

−

2015 2.06 6.85 13.35 −1.60 3.89 7.86 2016 2.73 7.73 14.31 −1.18 4.73 8.76 2017 3.39 8.61 15.27 0.21 4.40 8.45 2018 4.06 9.48 16.24 0.76 5.10 9.19 2019 4.72 10.36 17.20 1.80 5.20 9.33 2020 5.39 11.24 18.16 2.11 6.18 10.38 2021 6.05 12.11 19.13 2.62 6.91 11.17 2022 6.72 12.99 20.09 3.40 7.32 11.62 2023 7.39 13.87 21.05 4.49 7.38 11.71 2024 8.05 14.74 22.02 5.49 7.54 11.90 2025 8.72 15.62 22.98 6.34 7.88 12.27 2026 9.38 16.50 23.94 7.27 8.12 12.54 2027 10.05 17.37 24.91 8.14 8.44 12.89 2028 10.71 18.25 25.87 8.98 8.78 13.27 2029 11.38 19.13 26.83 9.80 9.15 13.67 2030 12.04 20.00 27.80 10.43 9.73 14.29

J.C. Peters Energy Policy 106 (2017) 41–47

44

need to be $24.75 and $6.50 per metric ton of CO2 for the 2007 and 2014 gas price scenarios, respectively. The EIA does not report significant differences in generation from gas power when comparing rate-based versus mass-based allowance trading CPP implementations (Martin and Jones, 2016).

Low and high estimates of fugitive emission rate (1.1% and 5.6%), policy effectiveness (0% and 40% reductions), and GWP time horizon (100-year and 20-year) are combined to produce eight projections for GWP from fugitive emissions for each scenario (Fig. 3). Fugitive emissions are assumed to increase in fixed proportions to electricity production.

4. Results and discussion

Fig. 2 shows that generation changes, as measured by their contributions to total emissions, are captured reasonably well, as measured by the correlation between projections and observations, for coal, gas, and oil power over the observable range (2007–2014) in response to the sharp decline in gas prices. The model does not perform well for 'other' power, but these various technologies con- tribute less than 0.6% of total emissions from the electricity sector. The four scenarios are projected annually to 2030 using the assumptions in

Table 3. The results here project greater reductions than some other studies

because coal power is expected to continually decline in the face of regulation and inexpensive gas. Many other models exploring electri- city generation futures in response to gas prices predict that coal power reaches a relatively high long-run equilibrium production level in the future (e.g. Arora and Cai, 2014; Shearer et al., 2014), but do not present any evidence that lends confidence to the predictive ability of these models. The model used here reports results similar to recently updated projection from the EIA Annual Energy Outlook 2016 (EIA, 2016b). Also, with every year, the EIA Annual Energy Outlook projections seem to converge toward the Peters and Hertel (2017) projections that capture the interdependency between utilization and capacity mechanisms.

Inexpensive gas, along with increased efficiency and less total demand, drove electricity sector CO2 emissions down over 14.2% from 2005 levels by 2014. The LG-BAU scenario projects this same trend to continue to 28.4% reductions by 2030 – almost to the extent of the CPP reduction target of 32%. Even if gas prices rebound back to 2007 levels (HG-BAU scenario), coal capacity retirements and renewable capacity expansion are still expected to drive down sector-wide emissions 21.3%. Therefore, both gas price scenarios present optimistic views of achieving the CPP, but additional incentives are still required. The decline in gas price pre- and post-shale gas boom helps reduce the policy intervention required to meet the CPP target from $24.75 per metric ton of CO2 tax to a more modest $6.50 per metric ton of CO2.

Both CPP scenarios reach the 32% reduction in electricity sector CO2 emissions by 2030. As expected, the LG-CPP scenario achieves the target largely by substituting coal power for gas power. Generation from gas increases 40.5%, while coal power declines 46.8%. Non-fossil fuel power, largely wind and solar, increases 77%. The HG-CPP instead meets the reduction target through renewable growth. Both gas and coal generation decline – 10% and 35.8%, respectively. Non-fossil fuel power increases by 86.6%, again driven by wind and solar, which supports the conclusion that inexpensive gas may delay renewable penetration. Interestingly, this conclusion comes with the caveat that the high gas price also means that coal power remains a larger share of total electricity production.

Accounting for GHG emissions is complicated by several emission- specific factors, including the different degradation timelines of differ- ent emissions. The United States intends to use the 100-year GWP values (US Department of State, 2016) for the Paris Agreement, while some researchers instead advocate for the 20-year GWP, which would ultimately result in much higher GHG leakage rate (Howarth, 2014). Therefore, in addition to reporting the low and high fugitive methane emissions cases with and without effective containment policy, the results report the 20-year and 100-year GWP in terms of CO2

Table 3 Assumptions on fuel prices, coal retirement rate, and CPP policy for each scenario.

Fig. 3. Cumulative million metric tons (MMT) of CO2e from combustion by source (bars) and various from assumptions on methane leakage (data points) in 2030. The y- axis is appended to focus on leakage assumptions. Oil and other combustion are included but contribute minor amounts to total MMT CO2e in 2030 – ranging from 39.8 MMT in the HG-CPP to 45.6 MMT in the HG-BAU scenario.

J.C. Peters Energy Policy 106 (2017) 41–47

45

equivalences (CO2e). Fig. 3 captures the opposing perspectives of the debate of whether natural gas can be an effective "bridge fuel".

Looking through the lens of high fugitive methane emissions and a 20-year GWP (i.e. the most vertical of the data points in Fig. 3), inexpensive natural gas leads to lower CO2 emissions in the electricity sector; however, the CO2 reduction is offset by increasing fugitive methane emissions from substitution toward gas power. In fact, the total GWP of combustion and fugitive emissions, absent the CPP, increases due to inexpensive gas under these assumptions, and are fairly similar with the CPP (as shown comparing HG-BAU with LG- BAU and LG-CPP). In other words, from this perspective, we are better off without the shale boom and the CPP than with the shale boom even with the CPP. The conclusion from this viewpoint is that inexpensive gas is counter-productive in terms of climate mitigation – supporting Howarth et al. (2011) and Howarth (2014). Also, this viewpoint would conclude that policy is a greater determinant of GWP outcome, and the CPP should be met with renewables rather than substitution toward gas in order to reduce US GWP at all – supporting Bistline (2014) and Shearer et al. (2014).

The opposing perspective begins with the assumption of low fugitive emissions or with 100-year GWP. From this perspective, the reductions from CO2 combustion are only modestly offset by fugitive emissions. Further, the low price of natural gas means that meeting the CPP requires less stringent mitigation policy (i.e. $6.50 versus $24.75 per metric ton of CO2). Imposing the HG-CPP carbon tax in the LG- CPP regime (represented by the LG-CPP+ scenario) nets greater total GHG emissions than the HG-CPP scenario – leading to the conclusion that natural gas is an effective bridge fuel. The LG-CPP+ scenario can be thought of as treating inexpensive gas as a windfall reduction on top of a target that would have had to be met without the existence of inexpensive gas. The scenario leads to both greater fuel-switching from coal power to gas power than the LG-CPP scenario and greater renewable penetration than the HG-CPP scenario.

The results here add clarity into the different starting assumptions that drive much of the debate around the question of whether natural gas is a bridge fuel to a low-carbon future or merely a distraction. Many researchers focusing on electricity find natural gas is an effective bridge fuel, while researchers focused on the broader economy find the opposite conclusion. Further, researchers that assume low emissions or compare GWP over a longer time horizon rather than high fugitive emissions with a shorter time horizon, would see natural gas a cost- effective way of meeting climate mitigation targets in the short-run. The results presented here indicate that both arguments are correct. Rather than add fuel the debate, the consensual path forward would be to ensure fugitive emissions are low.

The analysis begs the question: what scenario track are we on now? and what do we think that state-of-world will be in 2030? As of 2016, gas prices are comparable to 2014 prices, and the future of the CPP remains uncertain – placing us in either the LG-BAU or LG-CPP scenario, which have similar results with respect to spillover from methane emissions. In terms of fugitive emission rate, here is no clear consensus that emissions from shale gas wells are significantly greater than conventional gas (Burnham et al., 2011). Studies tend hover closer to the low fugitive estimate (Brandt et al., 2014) with some researchers sharply critical of the higher estimates (e.g. Cathles et al., 2012). In terms of GWP, the United States indicates that it will use the 100-year GWP values to calculate CO2 equivalent emissions (US Department of State, 2016). Therefore, perhaps the best estimate of 2030, right now, would be the low, no policy, 100-year estimate in the LG-BAU and LG-CPP scenario.

In the LG-BAU and LG-CPP scenario with our “best guess” of emissions rate, CO2 emissions fall 28.4% and 32.0% from 2005 levels and total 100-year GWP falls 26.9% and 30.5%, respectively – signaling modest spillover effects. Spillover into the Paris Agreement is defined here as the difference between US GWP changes from the CPP with and without accounting fugitive emissions. Ignoring fugitive methane, with

low gas prices the CPP would reduce nationwide 100-year GWP 9.6% from 2005 levels by 2025 (a large contribution to the Paris Agreement goal), but only between 9.4–8.2%, depending on fugitive emission rate and policy effectiveness, when accounting for the fugitive methane – a modest spillover of 0.2–1.4%.

However, assuming high fugitive emissions, ineffective policy, and 20-year time horizon, the LG-CPP scenario spillover could be as high as 4.4%, which supports legitimate concerns about over-reliance on fuel- switching for climate change mitigation. Effective policy that would reduce fugitive emissions 40% would decrease the spillover effect from 4.4% to 3.6% – a significant amount when viewed as a portion of nationwide GHG emissions.

5. Conclusion and policy implications

The results here show that fuel-switching toward inexpensive gas power from high-emitting coal power is an effective way to meet the CO2 reduction target of the CPP, but policy may be required to drive deeper climate change mitigation. Further, large growth in gas power can adversely impact more comprehensive policies like the Paris Agreement due to spillover from fugitive gas emissions upstream. The extent of the spillover depends on various assumptions on leakage rate, future policy effectiveness, and time horizon for the GWP calculation. Depending on these assumptions, natural gas can be seen as a potential bridge fuel or a distraction. The actual rate of fugitive methane emissions remains elusive. While most of the estimates fall within the range used here, some are much higher (e.g. Schneising et al., 2014) and further research on fugitive emissions will help narrow the band of uncertainty.

Regardless of assumptions on fugitive emissions and one's perspec- tive on natural gas as an effective bridge fuel, the consensual conclusion demonstrated here is that it is important to ensure that fugitive methane emissions are low. This calls for advanced techniques to measure the scale and sources of fugitive methane emissions along gas infrastructure. If these fugitive emissions are on the high-end of rates explored here and can be easily remedied, then policy could go a long way to avoid spillover and increase the effectiveness of the CPP in meeting nationwide GWP objectives of the Paris Agreement.

While this research assumes a nationwide emissions rate, Brandt et al. (2014) indicate that there may exist low emitters and "super- emitters". Identifying the source of fugitive emissions along the super- emitter infrastructure then offers great promise in reducing nationwide emissions. The US gas infrastructure is expansive and may grow in the face of increased gas power. Therefore, more research is needed on the scale and sources of fugitive emissions as well as on technologies that can be used to both identify and reduce emissions. With a clear picture on cost and effectiveness, we can answer whether fugitive emissions should be included within a larger-scoped CPP policy or addressed through additional policy already underway, such as New Sources Performance Standards (2016).

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at http://doi:10.1016/j.enpol.2017.03.039.

References

Arora, V., Cai, Y., 2014. US natural gas exports and their global impacts. Appl. Energy 120, 95–103.

Bistline, J.E., 2014. Natural gas, uncertainty, and climate policy in the US electric power sector. Energy Policy 74, 433–442. http://dx.doi.org/10.1016/j.enpol.2014.08.017.

Brandt, A.R., Heath, G.A., Kort, E.A., O'Sullivan, F., Pétron, G., Jordaan, S.M., Tans, P., Wilcox, J., Gopstein, A.M., Arent, D., Wofsy, S., Brown, N.J., Bradley, R., Stucky, G.D., Eardley, D., Harriss, R., 2014. Methane leaks from North American natural gas systems. Science 343 (6172), 733–735. http://dx.doi.org/10.1126/science.1247045.

Bureau of Labor Statistics (BLS), 2015. Labor Productivity and Costs By Industry Tables. Technical Report, Bureau of Labor Statistics.

J.C. Peters Energy Policy 106 (2017) 41–47

46

Burnham, A., Han, J., Clark, C.E., Wang, M., Dunn, J.B., Palou-Rivera, I., 2011. Life- cycle greenhouse gas emissions of shale gas, natural gas, coal, and petroleum. Environ. Sci. Technol. 46 (2), 619–627.

Cathles, L.M., Brown, L., Taam, M., Hunter, A., 2012. A commentary on “The greenhouse-gas footprint of natural gas in shale formations” by R.W. Howarth, R. Santoro, and Anthony Ingraffea. Clim. Change 113 (2), 525–535. http://dx.doi.org/ 10.1007/s10584-011-0333-0.

Clean Power Plan. Carbon Pollution Emission Guidelines for Existing Stationary Sources: Electric Utility Generating Units, 80 (205) Federal Register 64552–64964 (Oct. 23, 2015) (to be codified at 40 C.F.R. pt. 60).

Davis, S.J., Shearer, C., 2014. Climate change: a crack in the natural-gas bridge. Nature 514, 436–437. http://dx.doi.org/10.1038/nature13927.

US Department of Energy (DOE), 2015. Natural Gas Infrastructure Implications of Increased Demand from the Electric Power Sector. United States Department of Energy. February 2015.

US Department of State, 2016. Second Biennial Report of the United States. Under the United Nations Framework Convention on Climate Change.

US Energy Information administration (EIA), 2016a. Electric Power Annual 2014. Technical Report, US Energy Information Administration, Updated February 2016.

US Energy Information administration (EIA), 2016b. Annual Energy Outlook 2016. Technical Report, Early Release. US Energy Information Administration, DOE/EIA- 0383(2016).

US Environmental Protection Agency (EPA), 2014. Inventory of U.S. GHG Emissions and Sinks: 1990-2012. EPA 430-R-14-003. Washington, D.C.: EPA. Accessed October 2014. http://www.epa.gov/climatechange/Downloads/ghgemissions/US-GHG- Inventory-2014-Main-Text.pdf.

Golding, S.D., Boreham, C.J., Esterle, J.S., 2013. Stable isotope geochemistry of coal bed and shale gas and related production waters: a review. Int. J. Coal Geol. 120, 24–40.

Harrison, J., Horridge, M., Jerie, M., Pearson, K., 2014. "GEMPACK manual," GEMPACK Software, ISBN 978-1-921654-34-3.

Howarth, R.W., Santoro, R., Ingraffea, A., 2011. Methane and the greenhouse-gas footprint of natural gas from shale formations. Clim. Change 106, 679–690. http:// dx.doi.org/10.1007/s10584-011-0061-5.

Howarth, R.W., 2014. A bridge to nowhere: methane emissions and the greenhouse gas footprint of natural gas. Energy Sci. Eng. 2 (2), 47–60. http://dx.doi.org/10.1002/ ese3.35.

Kerr, R.A., 2010. Natural gas from shale bursts onto the scene. Science 328 (5986), 1624–1626.

Levi, M., 2013. Climate consequences of natural gas as a bridge fuel. Clim. Change 118 (3–4), 609–623.

Lyon, D.R., 2016. Chapter 3 – methane emissions from the natural gas supply chain. In: Kaden, Debra, Rose, Tracie (Eds.), Environmental and Health Issues in Unconventional Oil and Gas Development 2016. Elsevier, 33–48. http://dx.doi.org/

10.1016/B978-0-12-804111-6.00003-0. Martin, L., Jones, J., 2016. Effects of the Clean Power Plan. Issues in Focus for Annual

Energy Outlook 2016. U. S. Energy Inf. Adm., (6/20/16)⟨https://www.eia.gov/ forecasts/aeo/cpp.cfm⟩.

Myhre, G., Shindell, D., Bréon, F.-M., Collins, W., Fuglestvedt, J., Huang, J., Koch, D., Lamarque, J.-F., Lee, D., Mendoza, B., Nakajima, T., Robock, A., Stephens, G., Takemura, T., Zhang, H., 2013. Anthropogenic and natural radiative forcing. In: Stocker, T.F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S.K., Boschung, J., Nauels, A., Xia, Y., Bex, V., Midgley, P.M. (Eds.), Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Nisbet, E.G., Dlugokencky, E.J., Manning, M.R., Lowry, D., Fisher, R.E., France, J.L., Michel, S.E., Miller, J.B., White, J.W.C., Vaughn, B., Bousquet, P., 2016. Rising atmospheric methane: 2007–2014 growth and isotopic shift. Glob. Biogeochem. Cycles 30 (9), 1356–1370.

New Sources Performance Standards. Oil and Natural Gas Sector: Emission Standards for New, Reconstructed, and Modified Sources, 81(107) Federal Register 35824– 35942 (June 3, 2016) (to be codified at 40 C.F.R. pt. 60). https://federalregister.gov/ a/2016-11971.

Peters, J.C., Hertel, T.W., 2017. Achieving the Clean Power Plan 2030 CO2 target with the new normal in natural gas prices. Energy J. 38 (5), 23–50. http://dx.doi.org/ 10.5547/01956574.38.5.jpet.

Schaefer, H., Fletcher, S.E.M., Veidt, C., Lassey, K.R., Brailsford, G.W., Bromley, T.M., Dlugokencky, E.J., Michel, S.E., Miller, J.B., Levin, I., Lowe, D.C., 2016. A 21st- century shift from fossil-fuel to biogenic methane emissions indicated by 13CH4. Science 352 (6281), 80–84.

Schneising, O., Burrows, J.P., Dickerson, R.R., Buchwitz, M., Reuter, M., Bovensmann, H., 2014. Remote sensing of fugitive methane emissions from oil and gas production in North American tight geologic formations. Earth'S. Future 2 (10), 548–558. http://dx.doi.org/10.1002/2014EF000265.

Schwietzke, S., Sherwood, O.A., Bruhwiler, L.M., Miller, J.B., Etiope, G., Dlugokencky, E.J., Michel, S.E., Arling, V.A., Vaughn, B.H., White, J.W., Tans, P.P., 2016. Upward revision of global fossil fuel methane emissions based on isotope database. Nature 538 (7623), 88–91.

Shearer, C., Bistline, J., Inman, M., Davis, S.J., 2014. The effect of natural gas supply on US renewable energy on CO2 emissions. Environ. Res. Lett. 9 (9), 1–8. http:// dx.doi.org/10.1088/1748-9326/9/9/094008.

van der Mensbrugghe, D., Peters, J.C., 2016. Volume preserving CES and CET formulations. Presented at In: Proceedings of the 19th Annual Conference on Global Economic Analysis. GTAP Resource 5070. Washington D.C., USA, June 2016.

Weber, C.L., Calvin, C., 2012. Life-cycle carbon footprint of shale gas: review of evidence and implications. Environ. Sci. Technol. 46 (11), 5688–5695.

J.C. Peters Energy Policy 106 (2017) 41–47

47