In April 1998, a British newspaper reported that, to integrate the U.K.'s traffic patterns with the rest of Europe's, all vehicles would drive on the right side of the road. To ensure a smooth implementation of this major initiative, the report continued, the change would be made in phases. For the first six months, the regulation would apply only to buses and trucks.

The discerning reader would have picked up the specious logic in the anecdote. It was, indeed, an April Fool's Day prank, but it illustrates that there are logistical obstacles involved when restructuring long-established practices. Important transitions must be planned all the way through to their conclusion--for there are no do-overs.

The U.S. power grid--arguably the biggest machine on Earth--delivers electricity across nearly 7 million miles of transmission and distribution lines. Sitting atop this grid, and powering it, is a diverse generation fleet of conventional and renewable assets that use varied fuel sources. This fleet has undergone a tremendous transformation by making the largest fuel switch, to natural gas from coal. At the same time, the industry faces more energy efficiency and demand responses, increasing penetration of renewable energy sources, and states' evolving transmission policies.

Even as these changes progress, a confluence of factors is now increasing systemic risk in power markets, in our opinion, which in turn is having a negative impact on credit quality. Unlike our British traffic anecdote, these risks manifested in dramatic fashion in the ERCOT last month. While the isolated nature of ERCOT's grid, among several other factors, cascaded into large scale blackouts that have caused significant economic harm and human toll, we think these risks could surface in other markets as well. Here, we discuss developments that have contributed to these systemic risks. While data informs, stories tend to be more enduring as they stick. We will provide data but also build a narrative that supports our analysis.

It's All About Firm Power

Any discussion about fuel sources in power generation often becomes controversial. Our focus, however, is on firmness (i.e., reliability) of power. While we will indeed discuss how the intermittency of renewable power contributes to systemic risk, we first underscore the importance of firm power by providing some perspective on how power generation from different fuels performed in past winter events.

History does not repeat itself, but it often rhymes.  In January 2014, an extremely cold weather system dubbed the polar vortex forced PJM Interconnection grid outages at rates three times higher than expected. Although mechanical failures stemming from punishing cold caused winterization-related forced outages, a significant proportion came from fuel unavailability (Chart 1). For a frame of reference, plants with natural gas as their primary fuel accounted for 29% of the total installed capacity megawatts (MW), and coal-fired plants represent 41% (retirements since have changed this mix). There were three significant PJM outages: one on Jan. 7 and two on Jan. 29. The former took out about 40 gigawatts (GW), or 22%, of all available generation in the region, a major contingency event.

Chart 1

The polar vortex was unique but underscored that the then-prevailing capacity market pricing system did not spur investments in reliable generation. Before the capacity performance (CP) scheme in PJM, we believe the capacity market construct effectively penalized generators that contemplated reliability investments by not compensating them. There were two specific problems. First, generation owners were not allowed to include the cost of firm fuel supply in their supply offers; firm fuel supply helps mitigate resource interruption risk. Second, the PJM capacity market excused from penalties any outages due to fuel-supply interruptions. So, generation owners that made these sometimes-costly reliability investments became less competitive than generators who did not. The result was a perverse, yet rational, decision by generators to underinvest, which resulted in a higher outage factor. It was only a matter of time before large-scale outages culminated in the contingency event of Jan. 7.

The CP construct implemented after the polar vortex went some way to remedy that. In a commentary published in 2016, we calculated that, all else equal, based on the full historic average penalty, the CP product auction bids of combined cycle gas-fired generation assets would increase by about $50 per MW-day to subsume the risk of performance penalties (see "U.S. Electric Capacity Markets Update: Are Generators Picking Up Nickels In Front Of Steamrollers?", published April 14, 2016). We do not think that risk is being reflected in today's PJM capacity market pricing. However, with milder winters, new supply, flat to declining load (higher reserve margins), and energy efficiency, the PJM system has also not faced a significant weather stress since.

Milder events can also be harbingers of impending major ones.   A winter storm that may be overlooked--but one that came close to major outages--is the Northeast cold snap in December 2017. The extended cold spell spiked demand for natural gas, and supply constraints (which often happens in New England) raised natural gas prices dramatically. However, the New England market is different than ERCOT in that it has plenty of dual-firing capability. These older assets with oil-fired capabilities are expensive but serve as physical options. From Dec. 1 until the cold spell began, oil and coal plants contributed less than 1% of the energy generated by New England power plants. Between Dec. 26 and Jan. 8, 2018, oil contributed an astonishing 27% (Charts 2 and 3).

Chart 2

Chart 3

During those two weeks, New England generators burned through about 2 million barrels of oil. That's more than twice the oil used by New England power plants during all of 2016. Oil inventories declined from 70% of capacity to 20%. A blackout was averted only because the weather let up. Yet, diversity in generation sources, and storage of fuel supply helped--as did a quiet, solid run by a nuclear unit.

A root cause report by the independent system operator (ISO) in its June 2018 newswire specifically noted that the rising number of outages and poor response rate of the region's power plant fleet could be traced to several factors, including:

• Not having firm contracts for fuel delivery.
• Not keeping up with plant maintenance.
• Some natural-gas-fired generators with dual-fuel capability (the ability to burn oil stored on-site if natural gas unavailable) no longer maintaining that capability.
• Natural-gas-fired generators having trouble getting natural gas during cold weather, when the fuel is in high demand for both heating and power generation.
• Oil- and coal-fired power plants not stockpiling enough coal or oil to operate for extended periods because holding costs of the fuel is expensive, and they were rarely dispatched.
• Older oil- and coal-fired power plants that rarely operated often having mechanical problems when they tried to start.

A pay-for-performance scheme implemented soon after the winter event (June 2018) is intended to reward or penalize generation resources. Between June 2018 and June 2021, resources with a capacity obligation could be penalized $2,000/MWh for failing to meet their obligation during energy shortfalls, while resources that overperformed relative to their obligation could receive $2,000/MWh of additional revenue. This performance payment rate has increased to $5,455/MWh for the 2021 auction (delivery period June 2024).

The ISO-wide demand has since declined in New England, partly from milder weather but also from secular demand destruction as industrial load has declined some. As a result, capacity prices for the past 3-4 years have not reflected much, if any, scarcity premiums. We believe--and this is important--there are generation units currently participating in the market that do not have the capability of responding fast enough to a scarcity event.

That brings us to the storm that swept Texas in February.

The ERCOT 2021 Winter Storm

According to ERCOT's latest Seasonal Assessment of Resource Adequacy report (SARA) dated Nov. 5, 2020, it expected peak winter demand to be 57.7 GW based on based on normal weather conditions during peak periods from 2004 through 2018. Based on its analysis of the past two extreme events, after adjusting for expansion in the economy, ERCOT estimated an extreme--load of 67 GW. The last previous highest winter peak load was 65.7 GW (January 2018). The peak load that showed up with the stormfront was a colossal 70 GW. ERCOT later revised the peak load to 76 GW (without load shed).

ERCOT has about 82.5 GW of resources on the grid, after adjustments for capacity contribution. Of this, about 8.5 GW is typically under maintenance during the winter months (based on historical winter outage data compiled since 2017). Of the remaining 74 GW of capacity, 59 GW (80%) was expected to be supplied by ERCOT's conventional thermal and hydroelectric generation resources. Because ERCOT derates renewable resources during the winter and applies an average weighted capacity factor of about 27%, only about 7 GW of the state's 28 GW of wind power was available to meet winter peak demand.

The U.S. Energy Information Administration's Hourly Electric Grid Monitor showed that electricity net generation fell below ERCOT's day-ahead forecast demand shortly after midnight on Feb. 15, 2021, and persisted into Feb. 18, 2021. The mismatch between demand and day-ahead forecast demand quickly increased to at least 30 GW on Feb. 15 before narrowing to slightly less than 20 GW by Feb. 17, 2021.

At their lowest on Feb. 16,2021, the conventional thermal and hydro generation resources on ERCOT's grid supplied only 36.7 GW of capacity, 62% of their expected contribution. The shortfall in available capacity among the state's nuclear, coal, gas, and hydro plants limited ERCOT's ability to meet demand. While available wind generation capacity fell to 1.5 GW or just 17% of its expected 7 GW, wind capacity accounted for only a fraction of the system's total capacity shortfall--6 GW, or 17%, of the total of 36 GW of capacity lost.

Deregulation works best when competitive markets are left to decide the lowest cost reliable provider. But deregulation also means reliability rules are set not by law but by standards established by the market. The winter storm laid bare problems with outages to thermal and renewable generation, gas supply issues, challenges in implementing rolling blackout protocols, and failure to secure power to critical loads such as gas facilities. We note that while wind performed poorly, intermittent renewable resources were never expected to meet ERCOT's peak winter capacity needs. The surprise was that thermal generation units, considered firm sources of power, turned out to be unreliable.

The ERCOT winter event poses a credit dilemma. On one hand, an event that wipes out 25%-35% of aggregate estimated EBITDA is unfavorable for companies operating in ERCOT. Yet, the fact independent power producers (IPPs) survived such an extreme shock can also be viewed as testament to the robustness of their business models. From a ratings trajectory perspective, what is key are the bills moving through the Texas House of Representatives (House Bills 10 through 17) and whether the eventual implementation of reforms ensures such an event does not happen again.

The Grid Is In Transition

On March 31, President Joe Biden and the new administration released initial details of the $2.25 trillion American Job Plan--with details on clean energy and water infrastructure spending and incentives. An extension of renewable tax credits for 10 years is positive for clean energy companies/utilities for utility scale wind and solar, as well as residential solar.

The proliferation of renewables is having the desired effect of making the grid greener, but also resulting in other meaningful changes. Economically, the notable aspect of the renewables cost structure is that they have low (read: zero) variable costs, making them fully dispatchable.

This means renewables are fundamentally changing the shape of the dispatch curve. By way of explanation, let's assume the cumulative generation supply capacity along the horizontal X-axis of a graph and the marginal costs of these generation units along the vertical Y-axis. Then a plot of dispatch costs (or prices bid by the units in the day-ahead market) represents the generation supply stack. What's happening is that along the horizontal axis, the entry of new, near-zero marginal cost renewable resources has pushed the curve to the right. At the same time, reductions in marginal fuel costs of natural gas have lowered the slope of the curve. Not only has the curve flattened, it has effectively flipped. Less-flexible units, such as coal-fired and nuclear generation, formerly committed as base and/or mid-merit power supply are now more regularly the marginal resources needed to meet demand. With a flattened generation supply curve, energy prices are likely to stay low most of the time without the historical variation and price spikes.

What's more, the renewable cost curve is also swiftly declining; their overnight installed costs have declined rapidly to about $1,250 per kilowatt (KW) for wind and $1,100/KW for solar (from $3,000/KW, and above, a decade ago). As a result, fixed costs are also declining. The investment tax credit (ITC) for solar makes that effect even more potent. In a commentary last month, we presented that the levelized cost of energy (LCOE) from solar is approaching the low-$20 per megawatt hour (MWh) area (see "Going With The Flow: The Competitiveness Of Battery Storage Economics In the Power Sector", published Feb. 4, 2021).

The production tax credits (PTC) available to wind generation make its cost structure more significant. Wind power producers earn about $23/MWh of PTCs, allowing wind to dispatch even under negative locational marginal prices (LMP). Also, because the wind resource is variable, transmission is never built to handle 100% of the wind nameplate capacity. Something lower is reasonable. Consequently, in situations when wind output exceeds 70% or 80% of capacity, power prices fall due to congestion. Often, we observe negative LMPs because of the willingness of PTC-subsidized plants to bid negative prices and unwillingness of inflexible generators to back down (Chart 4).

Production tax credits are generally a good shield for wind generation in case of negative prices as they can bid up to their PTC values and earn revenue during negative priced hours. However, this becomes significant for solar assets, especially in areas such as ERCOT West with high rates of solar ceasing production when bids become $0/MWh due to high negative price frequency. This is where some mechanism such as equivalent demand forced outage rate (EFORd) or other reliability performance market signal could identify which assets providing the weakest reliability performance should back down.

Chart 4

From a credit perspective, declining power prices affect wholesale generators adversely as they not only lower current year power prices but also influence the forward power curve. Yet, contracting for low-priced renewable power benefits integrated companies as they can chose to serve their retail load by purchasing cheaper renewable power rather than generating it from owned generation units. This is the reason most integrated IPPs performed strongly in 2020 despite the pandemic.

'Hedgeability' Of This Power Market Has Come Under A Cloud

Many power purchase agreements (PPA) that renewable sponsors enter are "as-generated" or "unit-contingent" contracts. The off-taker agrees to purchase electricity and renewable energy certificates (REC) from the developer only when generated. The result is a backwardated forward power price curve as these energy contracts become part of the price discovery for forward years. In fact, if one were to mark the unhedged generation of an IPP such as Vistra Corp. (BB+/Watch Neg) to the forward curve in ERCOT for 2023, it will likely result in an EBITDA profile 15%-20% lower than in 2020. The markets have acknowledged the interruptible nature of renewable resources. Even as the forward power price curve is backwardated in markets such as ERCOT, it has consistently risen meaningfully about 4-5 months before the delivery year in each of the past years (Chart 5). Still, we think 2-3 years of strong prices are not enough of a signal for new generation projects to move forward with construction. This is partly why the unregulated power sector has lost its long-term investor.

Chart 5

With forward prices lower, finding high-quality off-take via conventional PPA contracting has become difficult. In general, there is very little liquidity in most forward markets for power past 2-3 years, making it difficult to hedge out for 12-20 years--the typical contract duration for a renewable PPA. With this long a tenor, generators and off-takers have too wide a bid-ask spread to reasonably settle at a price. This becomes relevant because traditional PPAs have the concept of mechanical availability percentages or guaranteed energy production thresholds. These pertain to total annual delivery volumes required of a site under its PPA. However, these thresholds are conservatively set and an icing event of two or three days would not be as material a driver to a site's performance under those traditional requirements. Moreover, conventional or bilateral PPAs may also have exclusions for weather events, such that even if the sites were not operating, extreme weather events would be deemed Force Majeure and would not impact the annual thresholds.

What some renewable developers experienced during the extreme winter storm in Texas that required procuring replacement power/financial settle (see "Not All Hedging Is Created Equal" and "Systemic Risk 3: Shape Risk In Renewable Contracts" below) is different from contracting terms under traditional PPAs.

From a credit perspective, with growing uncertainty in cash flows, lending against these projects would be more expensive.   After the winter event, our credit reviews of IPPs will assess a company's or project's hedging practices under more significant downside stresses. The amount of collateral and overall liquidity requirements to operate under these systemic risks will rise, especially now as they have manifested in such dramatic fashion in ERCOT. This is a material credit risk. It could slow credit momentum for power companies and, more broadly, increase costs for new generation investments in the sector.

Clean, Firm, And Cheap Power

We think interruptible power has limitations and can weigh negatively on market economics. While as-produced renewable contracts deliver cheap and clean power, they do not deliver firm power. In the process, they could discourage baseload generation, especially those with relatively low variable operation costs but high fixed costs.

For instance, all solar units produce at the same time. This means that as solar installed capacity goes up, its effective load carrying capability (ELCC) goes down. This is because ELCC values depend on when electricity shortages are most likely, and the type and quantity of resources already on the grid have a significant impact on the likely timing of electricity shortfalls. As the grid adds more solar plants, it reaches a point where they prevent daytime reliability issues so effectively that the remaining reliability challenges move into the evening hours when solar can't produce. At this point, adding more solar does very little to prevent electricity shortages, and the unit's ELCC often falls to 0%.

Replicating a firm renewable energy contract is illustrative. We've seen as-produced solar power contracts in the low-$20/MWh area. By the time firming shape is included around generation, resource adequacy payments, RECs, and a risk premium, it all adds up. The Geysers--a network of 725 MW of geothermal assets owned by Calpine Corp. (BB-/Stable) in California--are probably among the best forms of firm renewable energy in the country. These contracts are not contracted at $25/MWh; not even close. These contracts are significantly above market. We estimate those power contracts are north of $55-$60/MWh (no tax credit benefits).

This is why co-location of storage with solar and wind is gaining importance. Based on our analysis (see battery economics below), we arrived at an LCOE value that adds about $10-12.50/MWh to the LCOE ranges of stand-alone photovoltaic solar, or about a 50% premium to stand-alone solar levelized costs. We think adding large, utility-scale storage to the grid will do a lot to absorb the low-cost power and redistribute it to peak hours. This is the type of economic signal markets were designed to provide.

Not only does adding storage provide a potential firm power product, it has meaningful up-front benefits from qualifying for ITCs. Storage effectively doubles the capital costs of a solar project, 26% of which can immediately attract ITC benefits and avail modified accelerated cost recovery system depreciation. We believe adding storage offers significant growth prospects to companies such as NextEra Energy Inc. (A-/Stable/--) because it already has the advantage of established interconnections. Similarly, Fluence, the AES Corp. (BBB-/Stable) battery energy storage joint venture with Siemens AG (A+/Negative), has disclosed that it expects to increase its revenues by 45% each year over the next five years. We're seeing a significant increase in storage projects joining the interconnection queue (Chart 6)

Chart 6

We expect there will be spending on investments to the U.S. grid itself in recommendations that will follow. About 70% of the national grid is 25 years or older, and it faces harsher weather conditions. With ERCOT's significant winter event, we think all regulators will likely increasingly consider the firmness of power delivery. All options for firming power are somewhat costly, whether a regional transmission organization chooses to invest in renewable energy with the related transmission, fuel infrastructure with long-term contracts, or further measures to reduce demand for wholesale electricity and natural gas. However, inaction also comes at a cost, including greater risks to reliability and higher emissions when it's more economical to burn oil than natural gas. In other words, a region can pay for its fuel-security risks periodically, in spiking wintertime prices and potential energy shortages, or it can pay the costs proactively and avoid reliability risks by investing in infrastructure, firming renewables, firmer fuel contracts, and other reliability incentives.

In this regard, infrastructure could include further efficiency measures, electric transmission and new renewable energy resources, storage facilities for liquid fossil fuels, and gas pipeline infrastructure. We also expect reforms for winterization of the gas industry to become an increasing point of focus and could be a credit positive for the power companies. Power and gas companies may also consider negotiating into their contracts how force majeure conditions are defined.

Diversity of fuel sources would be under discussion, even as it is controversial.   When reliable base-load generation is a key difference that separated a power shortfall in ERCOT from business as usual in the Midwest, regulators, among others, in every state will likely take note. For instance, in our winter event example, a subsequent New England ISO briefing noted that electricity produced by the Millstone nuclear station during the December 2017 cold spell was equivalent to what could be produced by about 880,000 barrels of oil. The fact that Dominion Energy Inc.'s (BBB+/Positive) owned Millstone nuclear unit's sturdy run potentially warded off a Northeast blackout is not lost on us. Unlike other assets, one cannot inject nitrogen into nuclear units (to prevent rusting, etc.) to mothball them for future use. Once retired, they must be decommissioned as their fixed costs are high. There are no comebacks.

We think sometime soon, technology will get to power that is entirely clean, firm, and cheap. But the grid is still transitioning, and it's not quite there yet. Among clean, firm, and cheap power, it appears that now the choice is any two.

Perhaps it is time for the grid to get firm--about power delivery and ensuring such delivery.

Related Research

• Going With The Flow: The Competitiveness Of Battery Storage Economics In the Power Sector, Feb 4, 2021
• Independent Power Producers Update: The Greatest Risk Is Not Taking One, Dec. 6, 2017
• Merchant Nuclear Power Update: American Beauty Or American Reality?, Sep. 13, 2017
• Power Market Update: Knowledge Speaks But Wisdom Listens, May 23, 2017
• U.S. Electric Capacity Markets Update: Are Generators Picking Up Nickels In Front Of Steamrollers?, April 14, 2016
• Texas's Long, Hot Summer Makes Power Retailers Sweat, Oct 6, 2011