Robert Godby Roger Coupal. August Preliminary Analysis. Correspondence: Robert Godby University of Wyoming

Transcription

1 A Comparison of Clean Power Plan forecasts for Wyoming Coal: What impact do modeling choices make? Estimated Impacts on Wyoming Coal Production, Employment and State Revenues of the Clean Power Plan using EIA Simulation Data. Robert Godby Roger Coupal August 2015 Preliminary Analysis University of Wyoming Correspondence: Robert Godby

2 Introduction: On May 22, 2015 the Energy Information Administration (EIA) released their analysis of the potential impact on the U.S. economy, and specifically on energy production outcomes within the United States the U.S. Environmental Protection Agency s (EPA) proposed Clean Power Plan rules could have (see EIA (2015a)). The analysis was completed at the request of the United States Congress using the rules proposed in June The EIA study was conducted using the National Energy Modeling System (NEMS), an integrated simulation model of the U.S. Energy economy, which generates predictions regarding production, consumption, imports, conversion and imports of energy in the United States economy. 2 The EIA analysis was conducted utilizing assumptions consistent with the EIA s Annual Economic Outlook 2015 (AEO2015), and utilizes the same report s Reference case as the baseline from which impact scenarios of the Clean Power Plan can be compared. 3 The EIA analysis was the second detailed analysis to consider the possible implications of the Clean Power Plan using NEMS, as the Rhodium Group also conducted such an analysis in 2014 (see Larson et al. (2014)). 4 The purpose of this paper is to present additional analysis to determine the potential impact the CPP could have on Wyoming, the nation s largest coal producing state, which produces approximately 40 percent of the nation s coal, and as much as the next six coal-producing states combined. Insights from this analysis are useful as the EIA simulations present additional detail not available in the earlier studies, and suggests not only how coal production in the country could be affected for changes in the implementation of the program, but also how these decisions could affect energy economies and other fuel sources in the United States. Impacts on Wyoming coal production based on the Larson et al. analysis, were detailed in Godby et al. (2015a). That report presented the potential impact of the CPP on the Wyoming economy and Wyoming state revenues using projections of the CPP developed by Larson et al., and highlighted how these impacts are very sensitive to implementation choices made by states, particularly with respect to energy efficiency use and cooperation efforts, including the use of emissions trading. Those results are compared to results derived using the new EIA projections using the same impact and revenue models used in Godby et al. (2015b). As in the previous study, the results are sensitive to implementation decisions, and differences between the two analyses the importance of timing of emission reduction efforts, and how changes in underlying energy markets could affect coal generation outcomes. They also highlight how sensitive complex projections of potential impacts of a regulatory plan like the CPP are to modeling assumptions and differences in model implementation. 1 These rules were only preliminary, with final rules to be released August 3 rd, Full analysis of the final rules had not yet been completed by any organization at the time of writing. The specific Congressional request can be found in EIA (2015a), Appendix A, page A description of NEMS can be found at accessed June 5, See EIA (2015b). The reference case reports NEMS projections through 2040 under only existing laws both in place and enacted, and does not include the potential impacts of the EPA s Clean Power Plan (CPP). 4 The Rhodium Group is one of a few private operators of the NEMS model and is a non-partisan private research organization providing policy analysis and consulting to the public and private sectors.

3 Previous Projections and Impact Estimates for Wyoming The Clean Power Plan (CPP) as proposed in 2014 defines a set of state-specific emissions rate targets for carbon dioxide (CO 2 ) from existing and new or modified fossil-fueled electricity generation plants. 5 To meet these goals for existing plants (the so-called 111(d) portion of the rules) the EPA proposed a building block approach, which included utilizing (i) heat rate improvements, (ii) re-dispatch of power generation from coal to less intensive combined-cycle natural gas, (iii) re-dispatch of power generation from coal or natural gas to renewable or carbonfree nuclear generation, or (iv) demand-side use of energy efficiency. 6 The EPA s proposed rules defined a set of declining intermediate target emissions rates by state beginning in 2020 that would eventually result in a 30 percent reduction by 2030 in total US CO 2 emissions from power plants relative to While states were assigned specific emission rate targets, because these targets differed across states and regions, the proposed rules explicitly allowed for trade or cooperation between states to meet their combined goals jointly. Further, the timing of reductions could differ from EPA s interim goals between 2020 and 2029, provided the emissions reductions over that period averaged to those required in the proposed rules. Leaving choices regarding how emission reductions would be made to the industry and states, the degree of cooperation states might engage in and the timing of reductions was a principle that the EPA described as a flexible approach to regulating CO 2. 8 Table 1: Design of Larson et al. CPP Regulatory Policy Scenarios Considered National cooperation Regional cooperation Energy Efficiency efforts allowed Scenario 1: National Cooperation with Energy Efficiency (EE) Scenario 3: Regional Cooperation with Energy Efficiency (EE) Energy Efficiency efforts not allowed Scenario 2: National Cooperation without Energy Efficiency (EE) Scenario 4: Regional Cooperation without Energy Efficiency (EE) The Larson et al. analysis of the CPP focused on how national energy outcomes and impacts of the Clean Power Plan depend on two fundamental state policy choices: the use of energy efficiency, and the use of cooperative efforts across states to meet emissions targets collectively, such as regional carbon trading and credit systems. To investigate these potential impacts, Larson et al. conducted simulations using NEMS as four policy treatments, as shown in Table 1. 9 They also conducted a baseline reference case similar to EIA s AEO2014 Reference case. 10 For 5 This analysis does not incorporate changes to the Clean Power Plan in the final rules released August 3, The new rules increase the targeted CO2 emission reduction in the original proposal from 30 percent to 32 percent by 2030, changed state targets and changed the timeline of reductions to begin in 2022 instead of The energy efficiency building block was removed from the final rule, however energy efficiency measures may still be used for compliance in the final CPP rules. 7 States would propose specific implementation plans that would meet an interim goal defined as the average of target emission rates defined for each state from 2020 to 2029, and a final 2030 final emission rate target. 8 The rules were termed flexible because they did not define specific technologies and plant-level outcomes. 9 The scenarios did not include heat rate improvements (building block 1). 10 The Larson et al. baseline assumed the EPA s new and modified plant rules proposed in 2014 (the so-called 111(b) rules in the Clean Power Plan) were implemented but is otherwise similar to the AEO2014 reference case.

4 all compliance scenarios, Larson et al. assumed states comply with the relevant targets in each year. The analysis did not investigate how results would change if states chose alternative compliance paths. In the energy efficiency scenarios, all states were assumed to implement energy efficiency programs to the levels EPA assumed could occur in its Technical Support Documents issued with the proposed rules in 2014 and used by the agency to set specific state targets. Alternatively, in the no-energy efficiency scenarios, only existing energy efficiency programs were assumed to occur in any state. Cooperation in the scenarios ranged from states engaging in only regional cooperation, defined by their local electricity market regions, or by allowing national cooperation, effectively assuming a national carbon emission trading market. 11 All four scenarios are extreme conditions used to show the potential of each policy option, recognizing that, in reality, outcomes would likely be intermediate cases of each policy choice at the national level. Larson et al. s concluded from the national results of their policy exercise that energy market and emissions impacts will depend significantly on how states choose to implement the CPP, that the impact of the CPP by region varies, and that the current abundance of shale gas allows the use of natural gas generation as the most cost-effective means of meeting the CPP requirements. Their simulations showed increased natural gas use was the primary means by which states were projected to meet CPP targets nationally, with coal generation being curtailed as less carbonintensive natural gas generation was substituted for it. The analysis also showed that increased use of shale gas could have profound regional effects for both natural gas producers, and for coal producing regions whose demand would be reduced as coal-generation was curtailed. Godby et al. (2015b) further investigated the regional impacts of the Larson et al. simulations, concentrating on how the simulation outcomes in Table 1 would impact energy production in the state of Wyoming as the largest coal producing state in the country, and fifth largest natural gas producing state. 12 Using an economic impact model of Wyoming s energy sector developed for the study, the implied impacts on output, employment and labor income were estimated. These were then also used to estimate the impact coal and natural gas market changes could have on state government revenues using a revenue model under the four policy scenarios considered. Wyoming coal production projections from 2012 to 2040 from the Larson et al. simulations, along with the AEO2014 Reference Case are shown in Figures 1 and As shown in the figure, coal production projections in Wyoming are highly sensitive to policy implementation nationally - whether energy efficiency is used by all states or not, and whether cooperation is minimal This allowed impact analysis of the potential impact of the proposed 111(b) rules alone. The reference cases in AEO2014 and Larson et al. s analysis differ little in outcomes. 11 Electricity market regions used were those defined by the North American Electricity Reliability Corporation (NERC). 12 A description of these estimated impacts for Wyoming coal production can be found in Godby et al., Potential Impacts on Wyoming Coal Production of EPA GHG Proposals, Electricity Journal, May and 2013 results are calibrated to actual production levels realized in those years. The 2014 and 2015 projections were further calibrated for 2014 and interpolated for 2015 in Godby et al. (2015) as the original simulation results from Rhodium and AEO2015 for those two years overestimated actual production levels.

5 among states and limited to local electricity markets, or whether cooperation occurs at a national level. Outcomes shown in Figure 1 are further discussed in Godby et al. (2015a). Comparison of the coal production scenarios presented in Figures 1 and 2 suggests the regulatory impact of the EPA s 111(d) proposal on coal generation, and by implication Wyoming coal production, would be severe, with maximum declines by scenario ranging for over 32 percent to over 51 percent relative to 2012 production levels. Godby et al. Coal Production Impact of 111(d) Annual Wyoming Production (Mill. Short Tons) AEO2014 Reference 111d - National w/ EE 111d - National w/o EE 111d - Regional w/ EE 111d - Regional w/o EE Figure 1: Godby et al. Wyoming Coal Production Projections across Regulatory Scenarios 0% -10% -20% -30% -40% -50% -60% Rhodium 111(d) Coal Impact Summary compared to 2012 Output -4.8% -8.8% -8.6% -11.1% -13.2% -19.9% -30.5% -31.5% -30.3% -32.2% -33.2% -40.2% -44.0% -42.2% -45.0% -48.0% National Cooperation w/ee Regional Cooperation w/ EE National Cooperation w/o EE Regional Cooperation w/o EE Figure 2: Godby et al. Coal Production Changes across CPP Scenarios Relative to 2012

6 For Wyoming and for the nation as a whole, the results from these projections clearly suggested that allowing GHG reductions to be accomplished by a wider set of sectors in the economy, as is the case when energy efficiency is included as a policy response, reduces the pressure to reduce emissions from coal. Energy efficiency potentially allows more cost-effective CO 2 reductions to occur elsewhere in the industrial and household sectors of the economy, thereby reducing the impact of the proposed rules on coal generation and therefore coal production. Increased cooperation has a similar though reduced effect through most of the projection period, with the period from 2025 to 2029 being the exception. Using the production outcomes described in Figure 2, impacts on the state economy for the proposed regulations were estimated in Godby et al. (2015b). They found that even in the best case scenario, a loss of over 2 percent of jobs could occur in Wyoming compared to employment levels in Other scenarios exhibit greater losses. Godby et al. (2015b) also detailed Wyoming region-specific impacts, finding that overall, the impacts of the proposed regulations in the worst case would be especially destructive to the economy of the Powder River Basin region, where over 97 percent of the state s coal is mined, and where almost one job in ten would be eliminated. 2.00% 0.00% -2.00% 111(d) Employment changes due to Coal Impact (Statewide) 2.00% 0.00% -2.00% 111(d) Employment changes due to Gas Impact (Statewide) -4.00% -4.00% 111d - National w/ EE 111d - National w/o EE 111d - National w/ EE 111d - National w/o EE 111d - Regional w/ EE 111d - Regional w/o EE 111d - Regional w/ EE 111d - Regional w/o EE Figure 3: Godby et al. Coal and Natural Gas Employment Effects due to CPP Impacts Complicating the analysis of the economic impact of the EPA 111(d) rules on the Wyoming economy is the potential to use fuel-switching between coal and natural gas as a compliance option. Since Wyoming is a large producer of natural gas, potentially mitigating effects in the natural gas sector must be accounted for to determine the overall effect of the Clean Power Plan regulations on the state. Using Larson et al. s projections to compute Wyoming-specific natural gas production changes, Godby et al. estimated the total potential combined effects of the 14 Using employment levels in 2012 this was equivalent to the loss of over 10,000 jobs. These included coalproduction and support sector job losses, as well as those estimated due to the reduction in incomes these losses would create in the state in other state sectors. They did not include additional job losses that may be caused due to state revenue losses and reductions in state spending that could occur to balance the state s budget.

7 proposed carbon regulations across the natural gas and coal sectors were estimated. Results indicated that despite the fact that CPP rules are stimulative for the Wyoming natural gas sector, employment losses in the state s coal industry are not offset by the additional natural gas production caused by more stringent carbon regulation. This is shown in Figures 3 and 4. Godby et al.: Total 111(d) Employment Effect 0.0% -0.5% -1.0% -1.5% -2.0% -2.5% -3.0% -3.5% -0.3% -0.5% -0.3% -0.5% -1.0% -1.1% -1.2% -1.3% -1.3% -1.8% -1.6% -1.6% -2.5% -2.7% -2.5% -3.2% 111d - National w/ EE 111d - National w/o EE 111d - Regional w/ EE 111d - Regional w/o EE Figure 4: Total Regulatory Employment Impact across Coal and Natural Gas Sectors in Wyoming for 111(d) implementation compared to 2012 Employment levels. Comparing employment changes, the negative impact to the state from reduced coal production is approximately two to four times larger than the positive natural gas employment effects. Across cases the total impact on statewide employment ranges from a 2.5% to a worst-case decline of 3.2% in total employment relative to the baseline in The impact of the regulations alone would be expected to lead to a contraction in statewide economic activity and employment absent any other offsetting economic growth in the state economy. Total Coal Tax Revenue Change from Reference 0.0% -10.0% -20.0% -30.0% -40.0% -50.0% -60.0% -70.0% -6.5% -6.5% -6.6% -13.6% -16.1% -22.9% -31.9% -36.6% -38.3% -37.1% -48.7% -45.6% -54.9% -56.7% -58.3% -63.1% 111d - National w/ EE 111d - National w/o EE 111d - Regional w/ EE 111d - Regional w/o EE Figure 5: Total Coal Economy Tax Revenue Changes from Reference based on Godby et al. Projections

8 The potential impacts of proposed carbon regulations on Wyoming state revenues are also severe. Wyoming has no state income or corporate profit taxes and derives a majority of its incomes from state energy revenues. Considering changes in state revenues based on the Godby et al. projections due only to coal market changes, the most favorable cases of proposed CPP carbon regulations are those in which energy efficiency is used as a compliance strategy, as shown in Figure 5. When combining the estimated revenue impacts from coal and natural gas production changes the results are less straightforward. Combined coal and gas effects on state revenues are shown in Figure 6. Tax revenues continue to decline in all policy implementation cases, but in contrast to the economic impacts of the proposed regulations, the largest total state revenue declines (combined coal and natural gas production induced revenue changes) occur in the policy cases in which energy efficiency is allowed. In other words, those cases leading to the highest coal production are worst for state revenues. The fact that the total revenue outcomes under the energy efficiency cases are worse for the state than those without, in direct contradiction to the interests of the general Wyoming economy, is caused by differential rates of tax applied to each commodity, and differing commodity price outcomes in each scenario. Total Energy Tax Revenue Change from Reference 10.0% 0.0% -10.0% -20.0% -30.0% -40.0% -50.0% 0.2% -0.7% -9.8% -11.6% -10.7% -13.9% -11.6% -22.2% -25.0% -35.7% -36.3% -36.1% -41.2% -41.2% -46.5% -40.7% 111d - National w/ EE 111d - National w/o EE 111d - Regional w/ EE 111d - Regional w/o EE Figure 6: Total Energy Economy Tax Revenue Changes from Reference based on Godby et al. Projections The general finding found previously in Godby et al. (2015b), however, remains. The stimulative effects that proposed CPP rules might have on natural gas production do not offset declining coal production effects. The higher tax revenues due to natural gas do not offset the losses to coal-tax revenues caused by the regulation. Overall, proposed carbon regulations result in a predicted decline in the state s combined coal and natural gas revenues of between 36% and 46%, by 2030, as shown in Figure 6. EIA Projections and Resulting Wyoming Impact Estimates

9 The EIA s NEMS simulations of the proposed CPP rules released in May 2015 differs in design and in implementation from the Larson et al. analysis used to estimate impact results in Godby et al. (2015). The EIA analysis includes what EIA considers a base policy case and thirteen policy side-cases to determine the sensitivity of outcomes to alterations in assumptions under which the CPP is implemented. The differences in the design of scenarios considered by Larson et al./godby et al. and EIA reflect the fact that the former effort was a simulation experiment meant to estimate the potential range of impacts that two specific policy choices could have on CPP rules-induced outcomes, while the EIA analysis, in response to a Congressional request, was meant to describe how the CPP might actually impact the US economy. The set of scenarios that EIA presents imitate how they presume the program would actually be implemented, while the numerous side-cases are meant to investigate the sensitivity of scenario outcomes to changes in assumed market and technical considerations. There are few directly comparable scenarios between the two studies. For those that seem comparable, there are some important differences in how EIA treats energy efficiency and defines as cooperation compared to Larson et al. Table 2 summarizes how selected EIA scenarios compare to Larson et al./godby et al. s on the basis of energy efficiency and degree of state cooperation. Table 2: Design of EIA CPP Regulatory Policy Scenarios Considered Simulation Treatment No Energy Efficiency Energy Efficiency Regional Cooperation EIA Base Policy Case X x EIA Extended Policy X x Case EIA No Energy x x Efficiency EIA Extended Energy X Efficiency 1 EIA National X Cooperation EIA Limited Regional X x Trade National Cooperation Energy efficiency sensitivity to price incentives is assumed higher than in other treatments that include energy efficiency being used as a compliance option (see text). Regional cooperation limited to be no greater than in Reference case (see text). Figure 7 shows how the EIA anticipates the impact the CPP would have on Wyoming coal production under what they consider the most likely policy implementation case (their base policy case). For convenience the four Godby et al. Wyoming production cases are also shown. The base policy case is implemented utilizing the CPP rules as proposed in 2014 while otherwise utilizing assumptions consistent with the AEO2015 Reference case. Figure 7 also describes how the AEO2015 reference case differs from AEO2014. The EIA analysis also describes a policy extension case in which it is assumed the CPP rules rises to a 45 percent reduction in CO2 emissions relative to 2005 levels in the power sector between 2030 and The purpose of x 15 The Clean Power Plan in the initial proposal and final rule only defines emissions standards through 2030.

10 this case is to show how anticipation of increased stringency in carbon emission rules beyond 2030 under the CPP could affect market and energy production and consumption outcomes both within the timeframe of the original proposal from 2020 to 2030, and afterward. Annual Wyoming Production (Mill. Short Tons) AEO2015 Reference Case Godby et al. Reference EIA Policy Extension Case Godby et al. - National w/o EE Godby et al. - Regional w/o EE AEO2014 Reference EIA Base Policy Case Godby et al. - National w/ EE Godby et al. - Regional w/ EE Figure 7: EIA CPP Base and Policy Extension Scenario Outcomes for Wyoming Coal compared to Godby et al. Outcomes Several differences are immediately obvious. First, the underlying Reference case becomes more favorable to coal in the AEO2015 forecast, especially from 2020 to Much of the difference between AEO2014 and AEO2015 reference outcomes can be attributed to changes in baseline assumptions regarding energy prices, specifically oil price changes presumed through the simulation, and the impact these have on natural gas production and prices. Results are also impacted to a lesser extent by changes in assumptions regarding coal market conditions. Comparison of the oil price assumptions used in AEO2014 and AEO2015 is illustrated in Figure 8. The later Reference projection includes the low world oil prices that occurred in late 2014 that were unanticipated in the previous forecast, and how EIA projects they will persist over time. These oil price declines in world oil markets have impacts on natural gas production levels in the EIA AEO2015 simulations. Because natural gas is often a co-product of oil-plays in the United States, and the fact that oil prices impact natural gas and its by-products, declines in oil prices reduce natural gas production and increase natural gas prices. These price increases result in less use of natural gas use in the production of electricity in the period from 2020 to 2028, and greater use of coal.

11 Brent Crude Price/bbl (2013 $) $160 $140 $120 $100 $80 $60 $40 $20 $0 Projected Outcomes Source: EIA AEO2015 AEO2015 AEO2014 Figure 8: EIA Brent Crude Oil Price Assumptions in AEO2014 and AEO2015 Periods before and after these years actually experience lower natural gas prices than projected in AEO2014 due to other dynamics in the natural gas market, especially increased production in the northeastern United States. 16 Natural gas prices for electricity generation, along with Wyoming minemouth coal prices are shown in Figure 9. Changes to AEO2015 assumptions shown in Figures 8 and 9 in part explain the changes in Wyoming coal production between the AEO 2014 and AEO2015 Reference cases shown in Figure 7, where coal production in Wyoming averages five percent higher from 2020 to 2032 in the AEO2015 projection than in the previous reference case. 17 Natural Gas Price (2013$/mmbtu) $9 $8 $7 $6 $5 $4 $3 $2 $1 $ $45.00 $40.00 $35.00 $30.00 $25.00 $20.00 $15.00 $10.00 $5.00 $0.00 Minemouth Coal Price (2013$/short ton) AEO2015: Natural Gas (2013$/mmbtu) AEO2015: Wyoming PRB coal (2013$/short ton) AEO2014: Natural Gas (2013 $/mmbtu) AEO2014: Wyoming PRB Coal (2013$/short ton) Figure 9: AEO2014 and AEO2015 Delivered Natural Gas and Wyoming Coal Prices 16 AEO2015 natural gas projections assume continued high production in the Marcellus and Utica natural gas production areas, delaying the the impact of changes in world oil prices until the 2020s. 17 For a summary of the differences in assumptions and results between AEO2014 and AEO2015, see Annual Energy Outlook 2015, Appendix E, (accessed June 18, 2015).

12 The second obvious difference between EIA and Godby et al. is apparent in the differences between coal output projections for Wyoming in the CPP scenarios shown. EIA s base and extended policy cases suggest much more modest reductions in Wyoming coal output than Godby et al. through The worst EIA decline in this period occurs in the mid-2020s and results an approximate 25 percent decline relative to 2012 production levels. Even in the extended case when emissions reductions are increased to 45 percent by 2040, coal output remains above the worst of the Godby et al. outcomes. While part of the reason for these differences is the previously noted changes to the reference scenario natural gas and coal costs, differences are also due to significant changes in simulation assumptions and in the implementation of the simulations, some of which are outlined below. Other important changes in simulation assumptions include the costs of renewable technologies, which are assumed lower in the AEO2015 case, especially in later years of the simulation. This effect is important when combined with the fact that the EIA assumed that states intentionally delay emissions reductions early to lower the cost of the regulation, missing the early targets and then in later years they imposing greater reductions to ensure the average EPA interim target is met. Larson et al. s simulations used to estimate Wyoming impacts in Godby et al. assume interim EPA reductions targets were met. This steeper glidepath assumption results in an important outcome. While both plans meet the 2030 goal and interim average targets of the CPP, the EIA simulation allows greater early use of coal than the previous studies presume, and later in the simulation period when renewables are cheaper, greater reductions are made using these technologies than in those studies. 18 In short, EIA presumes a lesser role for natural gas and a greater one for renewables than earlier studies, and for this reason, coal production is higher throughout all EIA simulations. 19 The third difference between EIA and previous outcomes is shown in Figure 10. While the Larson et al./godby et al. results still hold more energy efficiency is better than less for coal production, and greater cooperation across states increases coal output, EIA outcomes exhibit far less sensitivity to either policy choice. The base and policy extension cases, which include energy efficiency measures and presume power trading as a form of cooperation are between the scenario extremes, however, differences in any scenario rarely exceed 25 million tons per year when compared to others. As shown in Figure 1, differences in earlier scenario results could be as high as 100 million tons in a given year. The reason for these differences is in part due to differences in the definition and implementation of energy efficiency and cooperation. Across cooperation cases, EIA and Rhodium simulations define national cooperation very similarly, implementing a national trading system to minimize emission reduction costs. As shown in Figure 10, the Godby et al. national cooperation with energy efficiency case is very close in outcome to the EIA national cooperation case, especially when one observes the underlying reference case differences. Similarly, the Godby et al. regional trading with EE case 18 Other reasons also matter to a lesser extent and are not detailed here. One is that EIA simulations allow heat rate improvement to occur. The impact of this change, however, is small and can be evaluated in the heat rate sensitivity scenarios EIA provides. 19 EIA (2015a) includes a sensitivity case in which the EPA s interim target glidepath is met.

13 is similar to the limited trade case from EIA that includes energy efficiency efforts, once underlying reference conditions are considered. Where scenarios differ most is in the energy efficiency cases. Annual Wyoming Production (Mill. Short Tons) EIA vs. Selected Godby et al. Coal Production Outcomes AEO2015 Reference Case EIA Base Policy Case Godby et al. - National w/ EE EIA Policy with limited trade EIA w/o EE AEO2014 Reference EIA Policy Extension Case Godby et al. - Regional w/ EE EIA Policy with National Cooperation EIA w/ high EE Figure 10: Comparison of EIA and Selected Godby et al. Wyoming Coal Production Outcomes 0.0% -10.0% -20.0% -30.0% -40.0% -50.0% -60.0% EIA Coal Production Changes Relative to 2012: 111(d) -3.2% -4.1% -3.0% -7.1% -3.3% -3.0% -4.5% -13.3% -15.7% -13.2% -14.4% -16.0% -23.1% -20.5% -17.4% -19.9% -23.2% -22.0% -21.9% -26.9% -23.6% -22.7% -27.1% -25.6% Base policy Policy extension Policy with national cooperation Policy with high demand side energy efficiency compliance Policy with no demand side energy efficiency compliance Policy with limited trade Figure 11: EIA Coal Production Changes across CPP Scenarios Relative to 2012 levels Comparison of Figures 2 and 11, especially the cases with high with energy efficiency and without energy efficiency compliance (all of which presume regional cooperation) indicate

14 implied impacts in the previous Godby et al. cases are as much as twice the impact the EIA simulations predict. Part of the reason is that in the underlying Larson et al. simulations used in Godby et al., the EPA s estimated levels of energy efficiency possible by states and used to define their respective targets in the 2014 CPP policy proposal are imposed as constraints to be met in the energy efficiency cases. Many states argued these levels of energy efficiency were both not cost effective and overly optimistic. 20 EIA approached the estimation of possible energy efficiency standards quite differently, relying on an internal study that examined energy efficiency deployment sensitivity and state incentives. The EIA then modeled assumed specific subsidies for energy efficiency were used by states and modeled the resulting energy efficiency responses. Their modeling resulted in reduced levels of energy efficiency deployment relative to EPA estimates, and therefore lower levels of energy efficiency use than in the previous simulations. This combined with the fact that the EIA simulations imposed emissions reductions later than was used in the previous analyses (the steeper glidepath), and assumed lower renewable costs resulted in coal output remaining higher than in the earlier analyses. Due to the greater renewable deployment in the reduced energy efficiency use cases, EIA results also predicted higher coal use for generation. Overall, these results indicate the complex interactions that could potentially occur when baseline assumptions regarding natural gas pricing and coal costs are altered, when assumed compliance paths are changed and when compliance measures like energy efficiency are actually modeled. The resulting outcomes can have significant effects on implied coal production and therefore coal-generation outcomes. The EIA analysis suggests the Clean Power Plan as proposed in 2014, while having serious implications for future coal output, may have far less serious consequences than some studies have suggested. These reduced impacts then change the economic impacts of the regulation. Figures 11 and 12 describe the employment impacts of the clean power plan on Wyoming through 2030, considering both impacts on coal and natural gas production. The model used is the same impact model used to produce the estimates in Figures 3 and 4. Figure 12a shows how higher coal production projections result in less employment loss compared to Figure 3. Figure 12b, however, presents a surprise. Where the Godby et al. outcomes predict a switch from coal to natural gas as the primary means of meeting the CPP, EIA projections presume by 2030 that renewables are the primary pathway to meeting reduction targets. Natural gas provides only an interim solution or relegated to backstop for renewables in the EIA results and resulting Wyoming natural gas production does not increase to the degree it did in Godby et al. The result is very limited employment gains in that sector. 20 See for example Wyoming s comment submitted to the EPA by the Wyoming Public Service Commissioner Al Minier ( accessed August 25, 2015).

15 Combining both coal and natural gas impacts in the EIA simulations results in the estimated total employment effects in Godby et al. being broadly similar to those implied by the EIA results. The two analyses only differ only in how the regulations impact the natural gas and coal sectors. Godby et al. s worst case is slightly more adverse at 3.2% employment loss overall compared to 2012, but otherwise both simulation results imply a 2.4% to 2.9% decline in Wyoming employment due to the CPP regulations. Where in the Rhodium analysis natural gas increases offset a portion of the coal employment losses, in the EIA scenarios, natural gas has a far lower offsetting impact, and by 2030 natural gas impacts are increasing employment losses. 2.0% EIA: Coal 111(d) Employment Effect 0.0% -2.0% -4.0% -0.2% -0.3% -0.2% -0.2% -0.2% -0.3% -1.4% -1.1% -1.6% -2.0% -1.4% -1.5% -1.6% -2.5% -2.2% -2.1% -2.5% -2.6% -2.4% -2.4% -2.5% -2.3% -2.7% -2.5% Base policy Policy extension Policy with Policy with national limited trade cooperation Policy with high demand side energy efficiency compliance Policy with no demand side energy efficiency compliance Figure 12a: Coal Employment Effects using EIA Projections EIA: Natural Gas 111(d) Employment Effect 2.0% 0.0% -2.0% -4.0% 0.8% 0.1% 0.2% 0.5% 0.3% 0.7% 0.7% 0.8% 0.1% 0.2% 0.1% 0.1% -0.3% -0.2% -0.2% -0.5% -0.2% -0.3% -0.1% -0.3% -0.4% -0.2% -0.3% -0.1% Base policy Policy extension Policy with national cooperation Policy with high demand side energy efficiency compliance Policy with no demand side energy efficiency compliance Policy with limited trade Figure 12b: Natural Gas Employment Effects using EIA Projections

16 EIA: Total 111(d) Employment Effect 0.0% -0.5% -1.0% -1.5% -2.0% -2.5% -3.0% -3.5% -0.1% 0.0% -0.1% 0.0% -0.1% -0.1% -0.7% -1.1% -0.8% -0.7% -0.8% -0.8% -2.4% -2.4% -2.8% -2.7% -2.7% -2.4% -2.8% -2.9% -2.8% -2.6% -2.6% -2.8% Base policy Policy extension Policy with Policy with limited national trade cooperation Policy with high demand side energy efficiency compliance Policy with no demand side energy efficiency compliance Figure 13: Combined Coal and Gas Employment Effects using EIA Projections As might be anticipated from the EIA production projections that vary much less across scenarios, state revenue outcomes are far less volatile using the EIA simulations. All scenarios have comparable coal revenue losses by 2030, ranging from 38.1 percent to 44.1 percent. These are shown in Figure 14 and are significantly less than the worst Godby et al. case scenarios shown in Figure 5, where losses of more than 60 percent are potentially predicted. 0% -10% -2.1% EIA 111(d) State Revenue Changes from Coal in Wyoming -4.2% -3.3% -3.6% -3.2% -4.5% -20% -30% -21.0% -24.6% -16.4% -21.7% -23.5% -24.9% -40% -50% -36.1% -38.1% -38.0% -36.1% -38.2% -36.1% -38.4% -42.4% Base policy Policy extension National cooperation High demand side EE -39.1% -41.4% No demand side EE -40.7% -44.1% Limited trade Figure 14: Total Coal Tax Revenue Changes from Reference based on EIA Projections

17 0% -3% -5% -8% -10% -13% -15% 0.1% EIA 111(d) Total Energy Tax Revenue Change from Reference 1.2% -11.3% -12.9% 0.9% 0.4% 0.3% 0.62% 0.1% -2.13% -2.50% -0.30% -12.8% -11.9% -11.4% -13.3% -14.0% -14.1% -13.2% -13.3% Base policy Policy extension National cooperation High demand side EE No demand side EE -0.2% -1.38% -12.9% -14.3% Limited trade Figure 14: Total Energy Tax Revenue Changes from Reference based on EIA Projections Overall, state government revenues from energy implied by the EIA results also prove far less volatile and not nearly as severe as previously estimated. Because across EIA scenarios the losses to coal are not as severe, and natural gas production is predicted to be far less volatile by scenario than the Godby et al. cases, total energy revenue losses are predicted to be near 14 percent by 2030, as shown in Figure 15. This is comparable to the least costly Godby et al. case previously estimated, and less than a third of the worst case estimated. As in the economic impacts, revenue outcomes are less sensitive to scenario, but suggest significant losses in revenue relative to a no-regulation alternative. Conclusions Comparison of EIA simulation results regarding the Clean Power Plan s potential implications for Wyoming coal production to previous studies indicates the estimation of potential impacts of these regulations on the state s economy prove very sensitive to assumptions and the modeling differences. Comparison of the Godby et al. (2015b) results more recent projections from the EIA suggest that differences in outcomes can be attributed to three major considerations: design of the simulation experiments, the assumptions used in deriving projected results - especially assumptions regarding economic conditions in energy markets and technology costs, and in the implementation of the simulation programming used. By design, Larson et al. s policy experiment from which the Wyoming estimates in Godby et al. are based was meant to emphasize the potential implications of policy choices facing states regarding energy efficiency and cooperation. Results reflect this, with results highly sensitive to the policy decisions. Further, those simulations imposed EPA s energy efficiency assumptions in their 2014 Clean Power Plan on the simulations, which may have overstated the use of such measures for compliance. EIA results are far less volatile, in part because of the lack of differences in scenario assumptions. Meant to describe how the agency thought the regulation would most likely be implemented, differences in scenario conditions are more subtle and reflect

18 sensitivity testing of results to changes in specific assumptions. Most EIA scenarios have regional cooperation comparable to that in the Larson et al./godby et al. studies and also include some degree of energy efficiency. Energy efficiency efforts are not as high as those assumed previously because they are constrained by cost considerations in the model and alternative choices in energy investment. Because assumptions regarding renewable generation costs are more favorable in the EIA simulations, energy efficiency compliance efforts are limited and renewable energy expenditures are substituted for energy efficiency where they are computed to be more cost-effective. The EIA study also presumes that compliance efforts by states occurs later than the Larson et al. simulations presumed. Overall, the EIA results indicate the complexity in estimating potential impacts for new regulations, and the sensitivity of results when baseline assumptions regarding natural gas pricing and coal costs are altered, when assumed compliance paths are changed and when compliance measures like energy efficiency effort are actually modeled. Final EIA projections suggest that the CPP impacts on coal production in Wyoming, and therefore the United States will be far less impacted than initial analyses suggested. This occurs in part because of much more favorable energy market conditions for coal based on 2015 market conditions compared to those in Natural gas production is reduced due to recent events in world oil markets that are assumed to be persistent for many years. Developments in renewable generation have also reduced those costs and the result is less substitution to natural gas. This is further aided by an assumption that compliance efforts would occur later, which further reduces negative pressures on coal demand. Overall, while severe for coal production in Wyoming, simulation estimates suggest reductions in output in the order of 25 percent may be more reasonable than earlier suggestions of the potential for 50 percent losses. Employment outcomes, however, seem robust to the changes in simulation design. In both Godby et al. and new estimates based on EIA simulations, the state of Wyoming is estimated to face a reduction of as much as 3 percent employment relative to 2012 levels by the time the CPP is fully implemented. Job losses from the Clean Power Plan could therefore be as high as 11,000, combining impacts on the coal and natural gas industries. Unlike the projections in Godby et al., EIA coal employment impacts are more modest, reflecting the reduced impact these simulations estimate for Wyoming coal output. However, unlike in Godby et al., the regulatory stimulus natural gas production experiences is not present in the EIA projections, and therefore does not offset coal employment losses. This results in the EIA-based impacts on overall employment being very similar to those in the previous study. What differs is the sectoral impact, not the overall one. State revenue outcomes are far less volatile. Coal revenue losses are not as great in the EIA projections and overall revenue losses are not as volatile as in the earlier Godby et al. analysis. Across scenarios, EIA results overall energy revenues are predicted to experience an approximate 14 percent loss relative to a case without regulation. Like the employment outcomes estimated, the state of Wyoming is still impacted significantly over time by the proposed regulations despite the fact that the EIA modeling avoids some of the much larger losses previous projections suggested could be possible under particular circumstances.

19 Given the final rule changes to the CPP announced in August 2015, however, can results presented here inform how Wyoming impacts are likely to occur? The 2015 final rule has several important changes. It is stricter, aiming at a 32 percent reduction in CO2 emissions by 2030 instead of the original plan s 30 percent. Additionally, the program begins later and imposes a steeper glidepath on required reductions. Energy efficiency, while still a potential compliance strategy, is no longer a primary tool the EPA used to define state goals. Given these outcomes, one could presume that the 2015 rules would be more likely to cause states to choose compliance paths similar to those in the EIA simulations relative to those in previous studies that assumed compliance occurred earlier, in accordance with the timeline suggested in the earlier rule proposal, with greater use of energy efficiency measures, and using natural gas more intensely to meet compliance goals. The EIA simulations delay early reductions - sparing coal somewhat, and rely less on natural gas and energy efficiency, and more on renewable energy as a compliance pathway. This is all consistent with the final rules announced in August Further, with more recent oil, and lower renewable energy and coal cost estimates, updated oil market predictions, the EIA estimates are also more consistent with current conditions. This suggests that the EIA modeling may more realistically portray how the Clean Power Plan will be implemented across states even it if was based on earlier rules and therefore may present a better guess at what future outcomes for electricity production, coal production in Wyoming, and the impacts on that state s economy might be. References EIA (2015a), Analysis of the Impacts of the Clean Power Plan, May 2015, accessed June 2, EIA (2015b), Annual Energy Outlook 2015 with projections to 2040, April 2015, accessed June 2, EIA (2015c), Annual Energy Outlook 2014 with projections to 2040, April 2014, accessed June 2, EIA (2015d), Analysis of Energy Efficiency Program Impacts Based on Program Spending, May 2015, Accessed August 14, Godby, R., Coupal, R., Taylor, D. and Considine, T. (2015a) Potential Impacts on Wyoming Coal Production of EPA GHG Proposals, Electricity Journal, Vol. 28:5, June 2015, pp Godby, R., Coupal, R., Taylor, D. and Considine, T. (2015b) The Impact of the Coal Economy on Wyoming, report prepared for the Wyoming Infrastructure Authority, February, (accessed August 25, 2015).

20 Larsen, J., Ladislaw, S., Ketchum, W., Melton, M., Mohan, S., and Houser, T., (2014) Remaking American Power, Rhodium Group and Center for Strategic and International Studies, released November See accessed June 8, Minier, Al (2014) Letter to the Honorable Gina McCarthy on behalf of the Wyoming Public Service Commission, PSC Letter Number , November 21, accessed August 25, 2015.