Creative Destruction and Your IPP Portfolio: Flat Demand & Capacity Additions Likely to Erode Gas & Coal Capacity Factors Through 2019

admin
Print Friendly, PDF & Email

______________________________________________________________________________

Eric Selmon Hugh Wynne

Office: +1-646-843-7200 Office: +1-917-999-8556

Email: eselmon@ssrllc.com Email: hwynne@ssrllc.com

SEE LAST PAGE OF THIS REPORT FOR IMPORTANT DISCLOSURES

_________________________________________________________________________________________

June 7, 2017

Creative Destruction and Your IPP Portfolio:

Flat Demand & Capacity Additions Likely to Erode Gas & Coal Capacity Factors Through 2019

Over the last five years, the output of the U.S. gas fired generating fleet has risen by 370 million MWh or 36%, and now exceeds that of the nation’s coal fired power plants. Much of this gain is due to increased hours of operation at existing plants; the capacity factor of the combined cycle gas turbine (CCGT) fleet increased from 48% to 56% over 2011-2016. Looking forward, however, we believe that IPP investors should focus on two less appealing facts: U.S. power demand remains stagnant, and the output of conventional power plants is declining under pressure from the rapid growth of renewable generation. Marginal gas and coal fired generators will be further pressured by large planned additions of new, highly efficient CCGTs. In this note, we assess how these trends will affect existing generators in each of the five U.S. regional transmission organizations (RTOs) that comprise the bulk of the deregulated power market. We conclude that the capacity factor of existing CCGTs will fall markedly over 2016-2019, with the largest output declines likely to occur in CAISO, the New York ISO and ISO New England. In PJM, both the existing coal fired and gas fired fleets should also see significant drops in power output. Generators in ERCOT, with its robust demand growth, should see smaller declines. Most exposed are DYN and CPN, with potential impacts on EBITDA from reduced output from existing generation of -12% and -7% by 2020. VST, with its generating fleet entirely in Texas, should fare best, with a maximum decline in EBITDA of -1% in 2019, after which we expect output to rise. Finally, we are removing EXC and PEG from our preferred list for hybrid electric utilities; the overhang of concerns about power markets and power generators will limit these stocks’ ability to outperform the sector.

Portfolio Manager’s Summary

  • We expect flat power demand and the rapid growth in renewable generation to erode the share of gas and coal in total U.S. power output, curtailing the capacity factors of existing fossil fuel plants (Exhibit 1). Capacity factors will fall further due to large additions of new, highly efficient CCGT plants, pushing more expensive gas and coal fired generating units up the supply curve and reducing their hours of operation.
  • We expect the output and capacity factor of the existing CCGT fleet to fall markedly, with the largest output and most prolonged declines likely to occur in CAISO (down 28% over 2016-2021), the New York ISO (down 23% over 2016-2019) and ISO New England (down 21% over 2016-2022) (Exhibit 2). We expect the output PJM’s existing CCGT fleet to fall by 14%, and that of its coal fleet to fall by 15%, by 2021.
  • In the early years of the next decade, the absence of planned additions of new CCGT capacity, combined with the expected retirement of nuclear, coal and gas fired steam turbine plants, should stabilize the capacity factors of existing CCGT fleets in PJM and the NY ISO, but we expect capacity factors to continue to fall in CAISO and ISO New England.

Exhibit 1: Forecast Capacity Factors of Existing Exhibit 2: Forecast Peak Drop in Power Output of Power Plants (2017-2022) Existing Power Plants (Off of 2016 Base)

Source: SNL, SSR estimates and analysis

  • Only ERCOT, with its robust growth in power demand, is likely to follow a different trajectory from the other RTOs. We expect the output of ERCOT’s existing CCGT fleet to fall by 10% through 2019; thereafter, power demand growth will drive output back almost to 2016 levels by 2021 and higher in 2022.
  • These expected declines in the power output of existing generating fleets are unlikely, in our view, to be offset by increases in power prices and spark spreads, at least through 2019. While the spark spreads implied by the forward prices of power and gas are flat to rising, we expect new, zero marginal cost wind and solar plants, and new, highly fuel efficient CCGTs, to bring down system marginal costs and thus power prices, eroding not just the output but the revenue and generation gross margin of existing plants.
  • Beyond 2019, however, we see the potential for tightening reserve margins in ERCOT and, to a lesser extent, CAISO to be reflected in more frequent price spikes during high demand hours (see Exhibit 15).
  • In ERCOT, the tight reserve margins we anticipate in 2020 and beyond coincide in our forecast with a recovery in the capacity factor of the existing CCGT fleet (see Exhibit 1). The early years of the next decade could thus see ERCOT’s existing generating fleet, with its higher variable cost of operation, setting the price of power during an increasing percentage of hours, as well as a rising number of scarcity hours when demand approaches the limit of available capacity. The two trends could contribute to a potentially material increase in power prices in ERCOT from 2020 on. The longevity of the recovery could be limited, however, by the generation overbuild that has occurred after prior price spikes in ERCOT.
  • In CAISO, we do not believe the impact of tightening reserve margins on prices will be as a great. First, the power output of CAISO’s existing CCGT fleet is expected to fall by 28% through 2021 (Exhibit 2); with the output of the higher variable cost existing CCGT fleet constrained by the generation of new, lower cost entrants — zero variable cost solar and wind farms, as well new CCGTs with materially lower heat rates — the marginal cost of supply in CAISO is unlikely to rise. Second, CAISO is unlikely to allow reserve margins to fall to the levels our forecast would suggest; in California, load serving entities are required to prepare long term power procurement plans to ensure adequate generation capacity to meet forecast load. Utilities are therefore likely to offer PPAs to ensure the timely construction of new peaking plants or additional storage.
  • Which independent generators are likely to be most adversely affected?
    • Given their exposure to the most vulnerable power markets (CAISO, PJM, the New York ISO and ISO New England; see Exhibit 16), we expect Dynegy (DYN) and Calpine (CPN) to suffer the largest revenue losses. We estimate that by 2020 the decreased output of the existing generating fleets of these two companies could reduce EBITDA by 12% and 7%, respectively, from 2016 levels, with bigger declines possible if prices and spark spreads should decline further to reflect the new generation sources.
    • By contrast, the power plants owned by Vistra (VST) are located entirely in Texas, and over 60% of this capacity is base load nuclear and coal. Given the smaller and shorter anticipated decline in the capacity factor of the existing CCGT fleet in Texas, its subsequent expected recovery, and the possibility of a marked tightening of reserve margins in ERCOT in 2020 and beyond, we believe VST to be the best positioned of the competitive generators, with a maximum decline in EBITDA of -1% by 2019 and increasing generation after that.
  • We are removing Exelon (EXC) and PSEG (PEG) from our preferred list for the hybrid electric utilities, even though the impact from reduced generating output will be limited for them. In our view, their increased merchant generating footprint with new CCGT capacity coming online and the likely overhang of concerns about power markets and power generators will limit the ability of these stocks to outperform the sector.

Exhibit 3: Heat Map: Preferences Among Utilities, IPP and Clean Technology

Preferences Among Utilities, IPPs and Clean Technology
Sector Weighting Favorites Concerns
Regulated Electric Utilities Overweight AEP, XEL ALE, SCG
Hybrid Electric Utilities Neutral NEE D, ETR
IPPs Underweight CPN, NRG
Renewables Underweight
Yieldcos Neutral NEP

Source: SSR analysis

Contents

Page

Summary and Conclusions 3

Company Impacts 8 How Do Our Expectations Compare with the Market’s? 10

Methodology 11

Market Analyses 12

Summary and Conclusions

The last five years have seen a radical change in the composition of U.S. power output, with coal fired generation falling by ~500 million MWh, or 28%, from 2011 through 2016, while gas fired generation increased by ~370 million MWh or 36%. (See Exhibits 4 and 5.) Coal’s share of total U.S. power generation fell from 42% in 2011 to 30% by 2016, while that of gas rose from 25% in 2011 to 34% in 2016, surpassing that of coal. (See Exhibits 6 and 7.) Much of this shift reflected the re-dispatch of the existing fleet of fossil fuel plants: over the last five years, the average capacity factor of U.S. coal fired plants fell from 60% to 53%, while that the nation’s combined cycle gas turbine fleet rose from 48% to 56%.

Exhibit 4: Change in U.S. Generation by Energy Exhibit 5: Change in U.S. Generation by Energy Source (TWh, 2011-2016) Source (%, 2011-2016)

___________________________________________________________________________________________________________________________________________________________________________________________

Source: Energy Information Administration, SSR analysis

Exhibit 6: Generation by Energy Source, 2011 Exhibit 7: Generation by Energy Source, 2016

___________________________________________________________________________________________________________________________________________________________________________________________

Source: Energy Information Administration, SSR analysis

For gas fired generators, the years of rising power output may now be behind them. The forces driving the increase in gas fired generation over the last five years are one-time events that cannot be repeated in the years ahead: The substitution of gas fired for coal fired generation after gas prices fell well below $4.00/MMBtu in late 2011, followed by the retirement of ~15% of U.S. coal fired capacity in 2015-2016 in response to the EPA’s Mercury and Air Toxics Standards.

Looking forward, we believe that IPP investors should focus on two less appealing facts:

  1. Total power output of the United States is stagnant.
  2. The output of conventional power plants is falling.

Over the last five years, total U.S. power output decreased by 1%, continuing a broad pattern of stagnation that has been evident since 2006. U.S. power output in 2016 was essentially unchanged from its level 10 years before, even as real GDP has increased by 14%. As the nation’s power output has stagnated, the share attributable to conventional power generation resources has declined: the combined output of the hydroelectric, nuclear, coal, gas and oil fired power plants in the U.S. has fallen by 6% over the last ten years, and has decreased from 98% to 92% of total U.S. power output (see Exhibit 8). The falling share of conventional generation is attributable to the rapidly rising output of new renewable power plants, particularly wind and solar: over the last ten years, non-hydro renewable generation increased by almost 250 million MWh, or from 2% to 8% of total power output, while conventional generation has declined by over 230 million MWh.

Exhibit 8: Total U.S. Generation by Energy Source (Millions of MWh)

___________________________________________________________________________________________________________________________________________________________________________________________

Source: Energy Information Administration, SSR analysis

We expect this pattern to continue over the next five years: net power demand (gross power demand net of self-generation) will remain flat; the ongoing growth of renewable generation will continue to erode the share of fossil fuel generation in total U.S. power output; as result, coal and gas fired generation will decline, curtailing the capacity factors of existing fossil fuel plants. Darkening this outlook are large planned additions of combined cycle gas turbine (CCGT) capacity. These new, highly efficient CCGTs, whose fuel consumption is on average 10% lower than that of the existing CCGT fleet, will enter the power supply curve below existing gas fired power plants, pushing these older, more expensive units up the supply curve and reducing their hours of operation. Higher cost coal fired power plants, specifically those burning Appalachian coal, will also be pushed up the curve and see their capacity factors fall. In summary, the existing fleet of fossil fuel plants will be squeezed between stagnant power demand and the rising capacity to supply this demand from lower variable cost units, reflecting the entry into operation of new wind and solar plants and new, more fuel efficient CCGTs.

In this note, we will assess the impact of these trends on the output and capacity factors of gas and coal fired power plants in the five principal regional transmission organizations (RTOs) covering the primary de-regulated power markets of the United States: PJM Interconnection (PJM), the Electricity Reliability Council of Texas (ERCOT), the California Independent System Operator (CAISO), the New York ISO (NYISO) and ISO New England (ISO-NE). (See Exhibits 9 and 10 for a comparison of the relative size of these markets.) As explained in the section entitled “Methodology” below, we have compared the growth in power demand forecast by each RTO (see Exhibits 11 and 12 for a comparison of historical and forecast growth rates in power demand by region) with the increase in generation expected from planned capacity additions[1]. Where new generation exceeds the expected growth in power demand, we have assumed that the output of the highest cost existing units (the marginal, price setting generators) falls commensurately.

Exhibit 9: Total Power Generation by RTO Exhibit 10: 2016 Generation by RTO

(Millions of MWh) (% of Combined Total)

___________________________________________________________________________________________________________________________________________________________________________________________

Source: SNL, SSR analysis

 

Exhibit 11: Electricity Demand Growth by RTO Exhibit 12: Electricity Demand Growth by RTO


___________________________________________________________________________________________________________________________________________________________________________________________

1. With the exception of CAISO, forecast growth in electricity demand reflects the load forecasts of the RTOs themselves. For CAISO, we have forecast power demand growth based on the experience of the RTO over the last five and ten years.


Source: ERCOT, ISO New England, New York ISO, PJM Interconnection, SNL, SSR estimates and analysis

 

The results of our analysis are summarized in Exhibits 13 and 14. Over the next three years (2017-2019), increases in low cost generation from planned additions of wind, solar and new CCGT capacity outpace the expected growth in power demand. As a result, we expect the marginal, price-setting units in each of the five RTOs (CCGTs in most markets, but a mix of CCGTs and coal fired power plants in PJM) to experience significant declines in power output and capacity factors through 2019 (see Exhibit 13). In CAISO and ISO New England we expect the capacity factor of the existing CCGT fleet to continue to fall through 2021, reflecting the stagnant power demand combined with continued growth in new CCGT and renewable generation capacity. In the New York ISO and PJM we expect the capacity factor of existing CCGTs to stabilize beginning in 2020, as planned retirements of nuclear and coal fired power plants materially reduce the supply of lower cost generation. We expect a robust recovery in the output and capacity factor of the existing CCGT fleet only in ERCOT, due to continued rapid growth in power demand and a lack of planned capacity additions.

Exhibit 13: Forecast Capacity Factors of Existing Exhibit 14: Forecast Drop in Power Output of Power Plants (2017-2022)(1) Existing Power Plants (Off of 2016 Base) (1)

___________________________________________________________________________________________________________________________________________________________________________________________

1. Estimates are for the average expected power output and capacity factors of existing combined cycle gas turbine (CCGT) plants. For PJM, we also present our estimates for the average expected power output and capacity factors of PJM’s existing coal fired fleet

Source: SNL, SSR estimates and analysis

Exhibit 14 presents the forecast percentage decline in the power output of the marginal, price setting units in each of the five regions, as well as the year in which this drop in output reaches its maximum. We expect the capacity factor of the existing CCGT fleets to fall markedly, with the largest and most prolonged output declines likely to occur in CAISO, where we forecast the generation of the existing CCGT fleet will fall by 28% over 2016-2021; the New York ISO, where we expect a 23% drop in CCGT generation over 2016-2019; and ISO New England, where the generation of the existing CCGT fleet could fall by 21% over 2016-2022. We expect smaller, but still very material declines in PJM, with the output of PJM’s existing CCGT fleet expected to fall by 14%, and that of its coal fleet to fall by 15%, by 2021. Only ERCOT, with its robust growth in power demand, is likely to follow a different trajectory. We expect the output of ERCOT’s existing CCGT fleet to fall by 10% through 2019; thereafter, power demand growth will drive output back almost to 2016 levels by 2021 and higher in 2022.

These expected declines in the power output of existing generating fleets are unlikely, in our view, to be offset by increases in power prices and spark spreads, at least through 2019. In competitive power markets, the prevailing price of electricity reflects the variable cost of generation at the last unit dispatched to meet demand. New, zero marginal cost wind and solar plants, and new, highly fuel efficient CCGTs, push higher cost fossil fuel plants up the supply curve; to the extent that the highest cost fossil fueled units are no longer required to meet prevailing demand, power prices fall to reflect the variable cost of supply of the lower cost units that remain. By way of example, new CCGTs have heat rates of ~6.7 MMBtu/MWh, some 10% below the average heat rate of ~7.4 MMBtu/MWh of the existing CCGT fleet. Thus for each hour that a new CCGT supplants an existing one as the marginal, price-setting unit on the system, the marginal cost of supply and thus the price of power will be 10% lower than would otherwise be the case. As we discuss in the section that follows, therefore, we believe that the rising spark spreads reflected in the forward power and gas curves are likely optimistic.

Beyond 2019, we see some potential for tightening reserve margins in ERCOT and CAISO to be reflected in more frequent price spikes during high demand hours. Adequate reserve margins (the excess of dispatchable capacity over peak demand) allow for the continued, uninterrupted supply of power even when large power plants or transmission interconnections fail, or when prolonged heat waves drive a surge in air conditioning demand. Conversely, when reserve margins are tight, such events may cause a scarcity of available capacity during peak demand hours. Because power supply curves are very steep at high level demand, scarcity conditions can be reflected in power price spikes.

Historically, system planners have sought to maintain reserve margins of 15% or more. In each of the five RTOs, we expect this target to be achieved through 2019, and in three RTOs (PJM, the New York ISO and ISO New England[2]) we expect reserve margins to remain well above 15% through 2022. In ERCOT and CAISO, however, we expect reserve margins to fall below 15% by 2020, and significantly so by 2022 (see Exhibit 15).

Exhibit 15: Historical and Forecast Reserve Margins by RTO


___________________________________________________________________________________________________________________________________________________________________________________________

Source: ERCOT, ISO New England, New York ISO, PJM Interconnection, SNL, SSR estimates and analysis

In ERCOT, the tight reserve margins we anticipate in 2020 and beyond coincide in our forecast with a recovery in the capacity factor of the existing CCGT fleet; we expect this to surpass 2016 levels in 2022 (see Exhibit 13). The early years of the next decade could thus see ERCOT’s existing generating fleet, with its higher variable cost of operation, setting the price of power during an increasing percentage of hours, as well as a rising number of hours when demand approaches the limit of available capacity as reserve margins fall. The two trends together could contribute to a potentially material increase in power prices in ERCOT from 2020 on.

However, we are cautious as to the longevity of such a rebound in prices, as previous spikes in ERCOT have been rapidly followed by overbuilds of new capacity. We note that several GWs of planned CCGT capacity are ready to start construction in ERCOT, having received all necessary permits. ERCOT’s own forecast of reserve margins, which includes planned CCGTs that have received all permits but have not started construction or received financing, shows a 16.8% reserve margin in 2022.

We expect that CAISO will also see a tightening in reserve margins beyond 2019, reflecting the planned retirement of ~9.0 GW of gas fired steam turbine peakers and 1.4 GW of existing CCGT capacity (see Exhibit 31). However, we believe the impact on prices of CAISO’s tightening reserve margins will be mitigated by other factors. First, CAISO by this time will be abundantly supplied with zero variable cost renewable generation as well as increased generation from new, fuel efficient CCGTs. Planned additions of new wind and solar capacity in CAISO are expected to add 8 million MWh of renewable generation by 2019, rising to 14 million by 2022 (see Exhibits 32 and 33). New CCGTs are expected to add a further 1 million MWh of low cost generation by 2019 and 7 million MWh by 2022. In the context of a market with flat power demand, the result is that the output of CAISO’s existing, higher cost CCGT fleet is expected to fall by 9 million MWh by 2019 and by 19 million MWh by 2022 – a drop of 28% (see Exhibits 32 and 33). With the output of the relatively higher cost existing CCGT fleet constrained by the generation of new, lower cost entrants, the average marginal cost of supply in CAISO is unlikely to rise.

Second, CAISO is unlikely to allow reserve margins to fall to the levels our forecast would suggest. In California, load serving entities are required to prepare long term power procurement plans to ensure access to adequate capacity to meet peak demand ten years into the future. The planned retirements of California’s gas fired steam turbine plants, reflecting the high cost of compliance with state regulations prohibiting the use of once through cooling water systems, have been anticipated for years. Solicitations to replace the capacity of these plants, either with new generation capacity or electric energy storage, are likely to be put out by California’s investor owned and municipal utilities, who historically have provided financial support for the construction of new capacity by signing long term power purchase agreements with independent generators.

Company Impacts

Based on (i) our estimates of the likely decline in power output of the existing CCGT fleets in each of the five RTOs, as well as the expected decline in generation of the existing coal fired fleet in PJM, and (ii) the exposure that each of the competitive generators has to these regional fleets, we have estimated for each of the competitive generators the expected change in their power output, relative to a 2016 base, for each year through 2020.[3] (See Exhibit 16). Based on currently prevailing forward power prices in each of these markets, we have estimated the impact of these changes in power output on the revenues and earnings of the competitive generators, again compared to a 2016 base.[4] (See Exhibit 17.)

Exhibit 16: Estimated Reduction in Annual Generation by Company (1)

___________________________________________________________________________________________________________________________________________________________________________________________

1. Percentages represent the expected increase or decrease in each company’s generation from existing power plants in a given year relative to that company’s total power output in 2016.  The percentages do not represent annual, sequential declines in generation, nor do they include the output from new power plants coming online during the period.

Source: Company reports, SNL and SSR analysis

Exhibit 17:  Estimated Earnings Impact of Forecast Declines in Generation (1)

___________________________________________________________________________________________________________________________________________________________________________________________

1. Percentages represent the expected increase or decrease in each company’s earnings in a given year relative to that company’s earnings in 2016.  The percentages do not represent annual, sequential declines in earnings.

Source: Company reports, SNL and SSR analysis

Among the independent generators, which are likely to be most adversely affected by the marked and prolonged decline in the capacity factors of existing gas and coal fired fleets in CAISO, PJM, the New York ISO and ISO New England? Given their exposure to these four RTOs (see Exhibit 18), we expect Dynegy (DYN) and Calpine (CPN) to suffer the largest revenue losses. We estimate that by 2020 the decreased output of the existing generating fleets of these two companies could reduce EBITDA by 12% and 7%, respectively, from its 2016 levels (see Exhibit 17 above). We note also that, in the longer term, the oversupply of energy in CAISO suggests that NRG Yield (NYLD) will face a dramatic decline in margins and cash flows when the contracts on its 1,755 MW of California CCGTs expire in 2023.

By contrast, the power plants owned by Vistra (VST) are located entirely in ERCOT, and over 60% of this capacity is base load nuclear and coal (see Exhibit 19). Given the smaller and shorter anticipated decline in the capacity factor of the existing CCGT fleet in Texas, its subsequent expected recovery (see Exhibits 13 and 14), and the possibility of a marked tightening of reserve margins in ERCOT in 2020 and beyond (see Exhibit 15), we believe VST to be the best positioned of the competitive generators, with a maximum decline in EBITDA of -1% by 2019 and increasing generation after that (see Exhibit 17).

Exhibit 18: CPN, DYN, NRG: Vulnerable Exhibit 19: VST’s All-ERCOT fleet is skewed

markets account for 50-65% of capacity toward base load nuclear and coal (MW)

___________________________________________________________________________________________________________________________________________________________________________________________

Source: SNL and SSR analysis

The impacts on NRG Energy (NRG) are also minor, reducing EBITDA by 1.3% by2021 after restructuring GenOn or by 1.8% even if the restructuring of GenOn fails. This is due to NRG’s relatively small CCGT fleet, its large coal and nuclear output in ERCOT and the effort in recent years to retire or convert to natural gas the least economic coal fired power plants in PJM. As a result, in PJM, NRG’s largest market by generating capacity, a majority of its revenues are from capacity revenues, which are not dependent on generation output rather than energy.

Among hybrid electric utilities with large merchant generation fleets the impacts are muted, with potential impacts on earnings more dependent on what happens to future power prices. For Exelon (EXC), with an EPS reduction of only 0.5% by 2020, this reflects the dominance of nuclear generation in its fleet, which leaves it more exposed to power prices than other generators, as the cost of nuclear fuel does not move with the general energy pricing environment, as gas and, to a lesser extent, coal do. At FirstEnergy (FE), with a 2% EPS reduction by 2020, the fleet is primarily nuclear and coal in PJM, but the coal fleet’s output has already declined significantly in recent years and capacity payments have become a larger percentage of earnings. PSEG (PEG) is the most exposed of the hybrids, experiencing a potential decrease in EPS of 2.2% by 2019 from its existing fleet because their fleet is now primarily a mix of nuclear and CCGTs.

How Do Our Expectations Compare with the Market’s?

Our expectations for the supply/demand dynamics in the principal U.S. power markets, which we set out in greater detail below, are generally more pessimistic for independent generators than the expectations implicit in forward market prices. In particular, the spark spreads implied by the forward price curves for peak hour electricity and natural gas suggest that generation gross margins per MWh are likely to rise through 2021 in CAISO, ERCOT, the New York ISO and ISO New England, with margins weakening only in PJM (see Exhibit 20). By contrast, we expect large planned additions of new renewable and CCGT capacity to suppress the output of existing fossil fueled power plants in each of these markets, with the single exception of ERCOT, driving down the marginal cost of supply and with it power prices and spark spreads.

Second, our forecast of regional reserve margins suggests that all five RTOs should be able to maintain reserve margins of 15% or more through 2019 and that three (PJM, the New York ISO and ISO New England) should be able to do so through at least 2022 (see Exhibit 15). We note, moreover, that California’s regulatory framework requires load serving entities to maintain long term power procurement plans to ensure adequate capacity is available to meet forecast peak demand; we therefore expect the state to act to avoid the erosion of reserve margins that would result from the scheduled retirement of its gas fired steam turbine plants in 2020 and the Diablo Canyon nuclear plant in 2024. Only in ERCOT do we see the potential for a tightening reserve margin to be reflected in a material increase in scarcity pricing.

To the extent that the stock prices of CPN, DYN, NRG and VST reflect current forward price curves for power and gas, and thus the expectations for spark spreads discussed above, our assessment of these stocks would likely be consistent with the market’s only in the case of VST, whose fleet, we agree, should benefit from a tightening supply/demand balance in ERCOT. Our expectation for the revenues and margins of CPN, DYN and NRG, by contrast, are likely more pessimistic than the market’s.

Exhibit 20: Spark Spreads Implied by the Forward Prices Curves for Peak Hour

Electricity and Natural Gas ($/MWh)(1)

___________________________________________________________________________________________________________________________________________________________________________________________

1. New York ISO spark spreads are for New York Zone J. Spark spreads in other zones of New York are much lower, but follow a similar pattern

Source: Bloomberg and SSR analysis

Methodology

A basic assumption of our analysis is that dispatch order of regional generation fleets will not change through 2022. This assumption is predicated on the current forward prices for coal and gas. Other than normal seasonal fluctuations in the price of natural gas, the forward price curves for coal and gas are broadly flat (see Exhibit 21); the market’s expectation, in other words, is that the relative prices of these fuels will remain roughly constant through at least the end of 2018. The critical implication for our analysis is that the relative cost of operating coal and gas fired power plants will also remained unchanged. To the extent coal or gas fired power plants are the marginal, price-setting units in power markets today, current forward prices suggest that they will remain so in the future. Based upon our estimate of the variable operating costs of coal and gas fired power plants in the various RTOs, given currently prevailing forwards, as well as the capacity factors of the coal and gas fired fleets in these power markets, we have assumed that CCGTs are the marginal, price setting units during the bulk of hours in CAISO, ERCOT, New York ISO and ISO New England, and that coal and CCGTs are both on the margin during the bulk of hours in PJM.

Exhibit 21: Forward Price Curves for Coal and Natural Gas, Expressed in $/MMBtu

___________________________________________________________________________________________________________________________________________________________________________________________

Source: NYMEX, SNL and SSR analysis

Given our assumption that the dispatch order of regional generation fleets will not change, we estimated future shifts in the composition of generation in the five U.S. RTOs, and assessed the implications for the output and capacity factors of existing power plants, based on the following assumptions:

  • To estimate peak load and power demand growth, we have relied upon the long-term forecasts of the RTOs themselves. In the case of CAISO, where power demand forecasts are not available, we have assumed no growth in power demand, reflecting the stagnation of regional power demand over the last five and ten years. (See Exhibit 11 for a comparison of historical and forecast growth rates in power demand by region).
  • Our estimates of the growth of renewable generation are based upon the expected power output of the renewable power projects under construction or in development today. Our analysis suggests that the output of this backlog of renewable power projects will be required to meet the targets set by existing state renewable mandates. We have assumed that the capacity factors of new wind and solar projects is equivalent to that of wind and solar projects in the same region that have entered service over the last three years.
  • Similarly, we have included in our forecast the generation from new combined cycle gas turbine power plants. In PJM and ISO New England, we have assumed that CCGT projects that are currently under construction or in advanced development and have cleared in capacity auctions, will enter operation as scheduled. In CAISO, we have assumed that CCGT projects that are currently under construction or in advanced development and have entered into a PPA with a utility, will enter service as scheduled. In ERCOT and NYISO, we have assumed that only the CCGT projects that are currently under construction will enter into service as scheduled. We have assumed that the capacity factors of new CCGT projects is equivalent to that of CCGTs in the same region that have entered service over the last five years.
  • We have also modeled the impact of announced retirements of nuclear, coal and gas fired generators, again assuming that the units to be retired operate at capacity factors equal to the average of their peers in the region.

Finally, we have modeled the power output of fossil fueled power plants according to the following rule. To the extent that power demand growth exceeds the expected growth in generation from new renewable and conventional power plants, we have assumed that the output of the marginal, price-setting units on the system will rise to fill the gap. Conversely, if the planned growth in generation exceeds the growth in power demand, we assume that the output of the marginal units in the region will fall to keep the regional demand and supply for power in balance.

We should highlight two important limitations of the methodology outlined above. First, because the number of power plants in construction or under development falls over time, and particularly so for renewable projects with their shorter development and construction cycles, our estimates of capacity additions in later years will likely fall short of the mark. Our forecasts may therefore under-estimate the extent to which the output and capacity factors of existing fossil fuel plants could continue to erode over time. Second, like most forecasts, ours assumes that past trends will continue. Should power demand growth accelerate rather than continue to stagnate, or the relative prices of coal and gas shift markedly, outcomes could be radically different from those forecasted here.

Market Analyses

In the sections that follow, we review the supply/demand dynamics of each of the five RTOs, first focusing on changes over 2016-2019 and then taking a longer view out to 2022. In these regional analyses, we review the expected growth of power demand, the likely increase in generation attributable to new capacity additions, and the loss of generation from the planned retirement of nuclear, coal and gas fired power plants. We then assess the combined impact of these factors on the output and capacity factors of existing fossil fueled power plants.

In summary, our expectations are that over the next three years (2017-2019), increases in low cost generation from planned additions of wind, solar and new CCGT capacity outpace the expected growth in power demand. As a result, we expect the marginal, price-setting units in each of the five RTOs (CCGTs in most markets, but a mix of CCGTs and coal fired power plants in PJM) to experience significant declines in power output and capacity factors through 2019 (see Exhibit 13). In CAISO and ISO New England we expect the capacity factor of the existing CCGT fleet to continue to fall through 2021, reflecting the stagnant power demand combined with continued growth in new combined cycle gas turbine and renewable generation capacity. In the New York ISO and PJM we expect the capacity factor of existing CCGTs to stabilize beginning in 2020, as planned retirements of nuclear and coal fired power plants materially reduce the supply of lower cost generation. We expect a robust recovery in the output and capacity factor of the existing CCGT fleet only in ERCOT, due to continued rapid growth in power demand and a lack of planned capacity additions.

PJM Interconnection

Exhibits 22 and 23 present announced capacity additions and retirements in PJM, by energy source, for the periods 2016-2019 and 2016-2022, respectively. Exhibits 24 and 25 present our forecasts for PJM of the change in generation by energy source for the same periods.

Over the next three years (2017-2019), ~15.2 GW of new CCGT capacity is scheduled to come on line in PJM (see Exhibit 22). If these plants enter operation as scheduled, and achieve capacity factors equal to the average of recent CCGT projects in PJM, their combined output can be estimated at ~81 million MWh in 2019 (see Exhibit 24). In addition, new wind and solar generation capacity is expected to add another ~7 million MWh of generation by 2019, bringing the total increase in generation over the next three years to some 88 million MWh.

By contrast, we expect power demand growth over this period, based on PJM’s own load forecast, of only some 15 million MWh. (Even this may be optimistic; while PJM is forecasting demand growth, electricity consumption in the RTO has declined over both the past five and ten years.) In addition, ~12 million MWh of nuclear generation is expected to be lost with the planned retirement in 2019 of Exelon’s Three Mile Island nuclear power plant. We expect that through 2019 another ~3 million MWh of generation will be lost due to the planned retirement of gas-fired, steam turbine peakers, and a further ~3 million MWh to be lost due to the retirement of coal fired power plants.

The imbalance between the output of the planned new CCGT and renewable power plants (~88 million MWh) and the need for additional generation (~33 million MWh) is ~55 million MWh. We therefore expect that by 2019 we will see a commensurate drop in the output of PJM’s marginal, price setting generators, which today comprise a mix of CCGTs and coal fired power plants (see Exhibit 24).

Looking out to 2022, we expect the expansion of PJM’s CCGT, wind and solar fleets to continue, with planned capacity additions over the period 2016-2022 totaling ~17.4 GW of CCGTs, 2.6 GW of wind capacity and 0.6 GW of solar. Again assuming these new plants come on line as scheduled and operate at capacity factors equal to those of recent projects deploying similar technologies in PJM, by 2022 the output of these new units can be expect to approach ~96 million MWh from the CCGTs, 8 million MWh from the wind farms, and 1 million MWh from the solar power plants, for a total of 105 million MWh of new generation. By contrast, PJM’s estimate of net power demand growth over 2016-2022 is only 13 million MWh. Adding to the need for new generation will be the loss of ~12 million MWh in generation from the Three Mile Nuclear nuclear plant, which is scheduled to be retired in 2019; the loss of ~20 million MWh from the retirement of coal fired power plants; and the loss of a further ~3 million MWh from the retirement of various gas fired, steam turbine peakers. We estimate the total need for new generation, therefore, at ~48 million MWh. Given this imbalance between the need for new generation and the output of PJM’s planned capacity additions, we expect the generation of PJM’s marginal, price-setting units to fall by ~56 million MWh.

We have assumed that this reduction in the output of PJM’s marginal fossil fuel plants is absorbed by the existing CCGT and coal fired fleets in proportion to their 2016 generation. On this basis, we expect a 15% reduction in the output of PJM’s operating coal fired power plants, causing their capacity factor to fall from 50% in 2016 to ~42% in 2021, and a 14% reduction in the output of PJM’s existing CCGT fleet, reducing its capacity factor from 57% in 2016 to 49% in 2021.

 

Exhibit 22: PJM – Change in Generation Exhibit 23: PJM – Change in Generation Capacity, Capacity, 2016-2019 (MW) 2016-2022 (MW)

Exhibit 24: PJM – Change in Generation by Exhibit 25: PJM – Change in Generation by

Energy Source, 2016-2019 (Millions of MWh) Energy Source, 2016-2022 (Millions of MWh)

___________________________________________________________________________________________________________________________________________________________________________________________

Source: PJM, SNL, SSR estimates and analysis

ERCOT

The outlook for the supply and demand for power in ERCOT is radically different from that in PJM due to ERCOT’s much more rapid expected growth in power demand. Whereas we assume that electricity consumption in PJM will expand at 0.27% p.a., ERCOT’s Long Term Load Forecast sees electricity consumption rising at 1.8% p.a., or more than six times as rapidly. Despite a massive planned buildout of wind, solar and new CCGT capacity, therefore, the capacity factors of ERCOT’s existing CCGT fleet are expected to stop falling by 2019, to recover almost to 2016 levels by 2021, and to surpass them by 2022.

In other respects, the two markets are more similar: like PJM, we expect major new capacity additions to come on line over the next three years (2017-2019), including ~8.1 GW of wind capacity, 2.7 GW of CCGT capacity, 2.2 GW of solar and 0.6 GW of gas turbine capacity (see Exhibit 26). As in PJM, the output of this new capacity will likely far exceed the growth of electricity demand, eroding the output of the existing power plants on the system.

Assuming that the planned capacity additions come on line as scheduled and achieve capacity factors equal to those of recent projects deploying similar technologies in ERCOT, these capacity additions will together add ~44 million MWh to the total supply of electricity in ERCOT by 2019 (see Exhibit 28). This new generation is more than sufficient to offset the expected growth in electricity demand, which is estimated at only 21 million MWh over 2016-2019, as well as the expected loss of ~4 million MWh of generation from coal and gas fired steam turbine generators that are scheduled to be retired this year and next. Indeed, the imbalance between the output of the planned new power plants (~44 million MWh) and the need for additional generation (~25 million MWh) is such that we expect the output of ERCOT’s existing CCGT fleet to fall by ~18 million MWh by 2019. If correct, this estimate would imply a 10% reduction in the output of ERCOT’s existing CCGT fleet, causing its capacity factor to fall from 47% in 2016 to ~42% in 2019 (see Exhibits 13 and 14).

Over the period from 2020 through 2022, however, the additions of generation capacity that we can forecast with confidence are very small, and the forecast growth in electricity demand is expected materially to outpace the increase in generation from these projects. The planned retirement of additional coal fired units adds to the need for increased generation. (See Exhibits 27 and 29). As a result, the output of the existing CCGT fleet is expected to rise by ~16%, raising the average capacity factor of the fleet, by our estimate, from a low of 42% in 2019 to 49% by 2022.

Exhibit 26: ERCOT – Change in Generation Exhibit 27: ERCOT – Change in Generation Capacity, 2016-19 (MW) Capacity, 2016-2022 (MW)

Exhibit 28: ERCOT – Change in Generation by Exhibit 29: ERCOT – Change in Generation by

Energy Source, 2016-2019 (Millions of MWh) Energy Source, 2016-2022 (Millions of MWh)

___________________________________________________________________________________________________________________________________________________________________________________________

Source: ERCOT, SNL, SSR estimates and analysis

CAISO

Of the five regional power markets analyzed, we expect CAISO to see the longest and most severe decline in the output of its existing CCGT fleet. From 2016 through 2021, we estimate, California’s existing CCGTs will suffer a 28% decline in power output, driving the capacity factor of the fleet from 37% in 2016 to 26% by 2021. The primary factors driving this decline are continued rapid growth in solar and wind power, as California strives to meet its increasingly aggressive targets for renewable generation; planned additions of new CCGT and gas turbine capacity; and the stagnation of power demand.

Over the last five years, electricity demand in California has been flat, while measured over the last ten years it has contracted at an average annual rate of 0.4%. We assumed that CAISO will see no growth in electricity demand through 2022.

Over this period, California will suffer a significant loss of capacity, but a small loss of generation from the planned retirement of its gas fired steam turbine generators: 8.0 GW of this capacity is scheduled to be retired by 2022, eliminating some 2 million MWh of electricity supply (Exhibits 31 and 33). Of this total, 2.3 GW, generating ~1 million MWh, is to be retired over the next three years (2017-2019) (Exhibits 30 and 32). These plants, which were built over 40 years ago, have very high heat rates and, thus, operating costs, and serve today primarily as peakers. Their retirement is planned in response to the prohibitive cost of complying with California regulations phasing out once-through cooling systems for steam turbine generators by 2020.

The loss of output from these peaking plants will be more than offset by planned additions of wind and solar capacity: 2.3 GW of solar and 1.6 GW of wind are scheduled to come on line by 2019, rising to 4.1 GW of solar by 2022 (see Exhibits 30 and 31). If these plants are completed on schedule and achieve capacity factors similar to those of recent wind and solar projects in California, they will add ~8 million MWh of zero marginal cost renewable energy to the state’s supply of electricity by 2019 and ~14 million by 2022. Planned additions of new CCGTs will add a further ~7 million MWh by 2022. (See Exhibits 32 and 33.)

One key risk to the forecast of new solar capacity in CAISO is the potential of increased tariffs due to Suniva’s petition to the U.S. International Trade Commission for protection from imports under the Section 201 “safeguard” provisions of US trade law. This would affect solar installations across the country, but in the other regions solar installations have a small impact on the total generation. We have not tried to estimate the impact of higher tariffs starting in 2018 for two reasons: i) it is difficult to estimate because the strong state support for new renewables, including a bill that just passed the California Senate that increases California’s RPS requirements to 50% by 2020 and 100% by 2045, suggesting that installations may not decline significantly even if costs increase; and ii) we believe there is a significant probability that the petition will be rejected based on a legal analysis of the petition.

In the context of flat demand and minimal losses in output from the retirement of the state’s gas fired steam turbine plants, CAISO’s large planned additions of solar, wind and CCGT capacity will result in an excess of generation through 2022. Zero marginal cost renewable generation and low cost power from new CCGTs will suppress the output of the state’s existing CCGT fleet, which we expect to fall by ~9 million MWh over 2016-2019 and by ~19 million MWh over 2016-2022 (Exhibits 32 and 33). In percentage terms, this would represent a decrease in the output of California’s existing CCGT fleet of 28% through 2021, driving the fleet’s capacity factor down from 37% in 2016 to an estimated 26% by 2021 (see Exhibits 13 and 14).

Exhibit 30: CAISO – Change in Generation Exhibit 31: CAISO – Change in Generation Capacity, 2016-19 (MW) Capacity, 2016-2022 (MW)

Exhibit 32: CAISO – Change in Generation by Exhibit 33: CAISO – Change in Generation by

Energy Source, 2016-2019 (Millions of MWh) Energy Source, 2016-2022 (Millions of MWh)

___________________________________________________________________________________________________________________________________________________________________________________________

Source: CAISO, SNL, SSR estimates and analysis

New York ISO

The New York ISO expects net electricity demand to decline through 2022, falling at an estimated 0.4% p.a. as stagnant final demand is met in part through the growth of distributed solar generation. In this context, we expect the New York ISO’s large planned additions of wind power and new CCGT capacity to force a 23% decline in the output and capacity factor of the existing CCGT fleet by 2019. In the longer term, however, the planned retirement of Entergy’s 2.1 GW Indian Point nuclear power plant in 2020 and 2021 will eliminate 17 million MWh of generation. As a result, we expect a recovery in the output of the existing CCGT fleet, with generation rising by 19% from 2019 through 2021.

Over the next three years (2017-2019), planned wind projects are expected to add ~1.9 GW of wind capacity to the New York ISO, while planned additions of CCGT capacity are expected to total ~0.7 GW. Through 2022, additions of wind power and CCGT capacity are expected to rise to ~2.4 GW and 1.8 GW, respectively. A 0.2 GW increase of imports in the most recent capacity auction and a 0.1 GW addition to the state’s solar fleet are expected to further augment system resources. (See Exhibits 34 and 35.)

If these generation resources come on line as scheduled, and achieve capacity factors equal to those of similar projects built in the region in recent years, the New York ISO will enjoy an increase in generation of ~8 million MWh over 2016-2019 – even as electricity consumption falls by ~1 million MWh. To offset this surplus generation, the output of the existing CCGT fleet must fall by 9 million MWh (Exhibit 36), reducing its capacity factor from 43% in 2016 to 33% in 2019. (See Exhibits 13 and 14).

In 2020 and 2021, however, the planned retirement of Entergy’s 2.1 GW Indian Point nuclear power plant will eliminate 17 million MWh of generation (Exhibit 37). We expect the loss of Indian Point to allow a 19% recovery in the output of the existing CCGT fleet that we estimate will raise its capacity factor back from 33% in 2019 to 39% by 2021.

Exhibit 34: NY ISO – Change in Generation Exhibit 35: NY ISO – Change in Generation Capacity, 2016-19 (MW) Capacity, 2016-2022 (MW)

Exhibit 36: NY ISO – Change in Generation by Exhibit 37: NY ISO – Change in Generation by

Energy Source, 2016-2019 (Millions of MWh) Energy Source, 2016-2022 (Millions of MWh)

___________________________________________________________________________________________________________________________________________________________________________________________

Source: NYISO, SNL, SSR estimates and analysis

ISO New England

ISO New England expects net electricity demand to decline from 2017 through 2022, falling at an estimated 0.5% p.a. as stagnant final demand is met in part through the growth of distributed solar generation. Yet this gradual decline in net electricity demand will coincide with a large planned expansion of the region’s CCGT and gas turbine fleet, increased imports of hydroelectricity from Quebec, and new renewable generation capacity. Despite the planned retirement of Entergy’s 685 MW Pilgrim Nuclear Power Station in 2019, and the consequent loss of its 5 million MWh of generation, we foresee a 21% decline in the power output of New England’s existing CCGT fleet over 2016-2022, causing the fleet’s capacity factor to fall from 42% in 2016 to 33% by 2022 (see Exhibits 13 and 14).

Over the next three years (2017-2019), planned CCGTs are expected to add 2.4 GW of capacity to ISO New England, while new gas turbines, wind and solar power projects are expected to add ~1.0 GW of additional capacity (see Exhibit 38). If these new resources come on line as scheduled, and achieve capacity factors equal to those of similar projects built in the region in recent years, ISO New England will see generation increase by ~11 million MWh over 2016-2019, even as net electricity demand remains flat. While scheduled retirements of nuclear and coal fired capacity are expected to eliminate ~6.0 million MWh of supply by 2019, the output of ISO New England’s existing CCGT fleet must nonetheless fall by ~4 million MWh, or 8%, to offset the surplus of generation (see Exhibit 40). Looking out to 2022, by which time net electricity demand is expected to have declined by 4 million MWh and the output of the new CCGT fleet to have further increased, we estimate that the cumulative decline in the output of the existing CCGT fleet must rise to 11 million MWh (see Exhibit 41). This would bring the total drop in output of New England’s existing CCGT fleet to 21%, reducing the capacity factor of the fleet from 42% in 2016 to 33% in 2022 (see Exhibits 13 and 14).

Exhibit 38: ISO NE – Change in Generation Exhibit 39: ISO NE – Change in Generation Capacity, 2016-19 (MW) Capacity, 2016-2022 (MW)

Exhibit 40: ISO NE – Change in Generation by Exhibit 41: ISO NE – Change in Generation by

Energy Source, 2016-2019 (Millions of MWh) Energy Source, 2016-2022 (Millions of MWh)

___________________________________________________________________________________________________________________________________________________________________________________________

Source: ISO-NE, SNL, SSR estimates and analysis

©2017, SSR LLC, 225 High Ridge Road, Stamford, CT 06905. All rights reserved. The information contained in this report has been obtained from sources believed to be reliable, and its accuracy and completeness is not guaranteed. No representation or warranty, express or implied, is made as to the fairness, accuracy, completeness or correctness of the information and opinions contained herein.  The views and other information provided are subject to change without notice.  This report is issued without regard to the specific investment objectives, financial situation or particular needs of any specific recipient and is not construed as a solicitation or an offer to buy or sell any securities or related financial instruments. Past performance is not necessarily a guide to future results.

  1. Our estimates of the growth of renewable generation are based upon the expected power output of the renewable power projects under construction or in development today. Our analysis suggests that the output of this backlog of renewable power projects will be required to meet the targets set by existing state renewable mandates. We have assumed that the capacity factors of new wind and solar projects is equivalent to that of wind and solar projects in the same region that have entered service over the last three years.Similarly, we have included in our forecast the generation from new combined cycle gas turbine power plants. In PJM and ISO New England, we have assumed that CCGT projects that are currently under construction or in advanced development and have cleared in capacity auctions, will enter operation as scheduled. In CAISO, we have assumed that CCGT projects that are currently under construction or in advanced development and have entered into a PPA with a utility, will enter service as scheduled. In ERCOT and NYISO, we have assumed that only the CCGT projects that are currently under construction will enter into service as scheduled. We have assumed that the capacity factors of new CCGT projects is equivalent to that of CCGTs in the same region that have entered service over the last five years.
  2. In ISO New England, our forecast is significantly above the ISO’s forecast in 2022 (27.8% vs ISO New England’s forecast of 21.8%). Our forecasts assumes that the import capacity that cleared in the 2020/21 capacity auction (~1.2 GW) will continue to clear in future auctions, whereas the ISO New England removes imports after 2021 since they have not been firmly committed to the ISO.
  3. We did not include the impact of new generation under construction by a number of generators because plants are already in current expectations for the companies, we do not expect the output of those plants to change signifcantly versus expectations and the primary change versus expectations will be in the pricing of the power, an issue we do not address in this note. The expected new generating capacity for the companies under coverage are: CPN – 828MW of CCGT in PJM and 270 MW in ERCOT; EXC – ~2400 MW of CCGTs in ERCOT; NRG – 360 WM of peaking capacity in ERCOT and 898 MW in CAISO; and PEG – ~1200 MW of CCGTs in PJM and 485 MW in ISO-NE.
  4. In calculating the earnings impact, we ignored the effect of fuel and power price hedging because we assumed that, in light of the generally weak power markets, most hedges are in the money, companies will retain the benefit of their hedges and the impact of the lost generation output would, therefore, only be the market value of the output.
Print Friendly, PDF & Email