ERCOT market design options—a summary of E3's November 2022 report
This note summarizes the “Assessment of Market Reform Options to Enhance Reliability of the ERCOT System,” a report commissioned by the PUCT from energy consulting firm E3 and released on November 10, 2022. The PUCT is currently seeking comments from the public on this report.
Key Takeaways
· All six of the reform proposals result in the retention or construction of natural gas-fired generation that would otherwise not be part of the grid by 2026. The LSERO, FRM and PCM are technology-neutral and enable batteries and renewables to earn credits. The BRS and DEC will likely exclusively benefit natural gas.
· All of the reform proposals pay for this additional capacity by assigning costs for these resources through the sale of “credits” from generators to load-serving entities (LSEs). LSEs will pass these costs on to their end customers, leading to an increase of total system costs of 2-4%.
· The FRM and PCM create centralized markets for the exchange of credits. The LSERO and DEC require buyers and sellers to contract on a bilateral basis.
· E3 recommends the PUC adopt the FRM, a mechanism that has precedent in the U.S. The PUCT staff are soliciting feedback on the PCM, which is as-yet untested nationally.
· Qualitatively, E3 sees the LSERO and FRM as having the most potential to address extreme weather events, at the cost of having high administrative complexity and a long implementation timeline.
Overview
In the aftermath of 2021’s Winter Storm Uri, the Texas Legislature passed a law, Senate Bill 3, mandating that the PUCT take steps to ensure the reliability of the grid during extreme weather conditions. E3’s report outlines six reform mechanisms that could enable ERCOT to better withstand extreme weather events in the future. The PUCT will likely end up choosing a version of one of the six. The current debate is about which mechanism is best.
E3’s also report discusses a technical reliability standard of “1-in-10 Loss of Load Expectation (LOLE),” which means that the system should not have a loss of load of any size or duration for more than 1 day every 10 years. In other words, this is a definition of the goalpost that has to be reached by the system (the PUCT could pick a different goalpost). The 1-in-10 LOLE is a commonly accepted reliability standard in the U.S. For context, Uri resulted in 3 consecutive days of load shed in ERCOT.[1]
The six proposals in the E3 report are: 1) Load Serving Entity Reliability Obligation (LSERO), 2) Forward Reliability Market (FRM), 3) Performance Credits Mechanism (PCM), 4) Backstop Reliability Service (BRS), 5) Dispatchable Energy Credits (DEC), and 6) a hybrid of the Backstop Reliability Service and Dispatchable Energy Credits (BRS + DEC). The LSERO, FRM, and PCM share a number of conceptual similarities and a couple of key differences, which I discuss below. E3’s report recommends the FRM as the best option out of the six, while it appears that the PUCT staff may be favoring the PCM, or at least wants more robust analysis of the PCM. It appears that E3 thinks the DEC and BRS models have some major drawbacks, which I touch on in brief.
Similarities between the LSERO, FRM, and PCM
All three of these mechanisms introduce a new requirement on Load Serving Entities[2](LSEs) to purchase “credits” proportional to their share of electric load during periods of net peak load (i.e., what proportion of the electricity market are they serving when supplies are tightest, since this is the time period when reliability is most at risk). These credits are called “reliability credits” in the case of the LSERO and FRM, and “performance credits” in the case of the PCM. In principle, however, they are similar—they represent a wealth transfer from LSEs to generators which incentivizes generators to build more power generation facilities, which would be used to provide reliable power supply in times of system stress. These additional facilities would likely be money losers in the absence of reliability or performance credits. LSEs will likely pass these costs on to their customers, and E3 estimates that implementation of any of these three mechanisms would increase total system costs by ~2%.
E3 sees all three of LSERO, FRM and PCM as meeting a 1-in-10 LOLE standard, compared to a LOLE of 1.25 days per year in a “status quo” situation. This appears to be mostly achieved through the retention of 5,630 MW of natural gas-fired generation that would have otherwise been uneconomical to operate (and would have left the grid) in the absence of credit payments. As a point of reference, E3’s baseline assumption is that roughly ~38,000 MW of solar, wind and battery storage will be added to the grid in the next four years regardless of which reliability mechanism is ultimately chosen. These resources will also be eligible for credit payments under the LSERO, FRM and PCM, but likely at lower rates than gas-fired generation. Although natural gas appears to benefit the most from the LSERO/FRM/PCM mechanisms, all three are explicitly technology-neutral.
Differences between the LSERO, FRM and PCM
The biggest difference between the PCM and the other two is that the PCM is backward-looking in how it assigns performance credits to generation resources, while the LSERO and FRM are forward-looking.
The PCM works as follows: ERCOT reviews grid performance for the preceding calendar year[3]and finds the 30 hours during which the system was under the greatest stress. The LSEs are then assigned a requirement to purchase performance credits based on their share of load during those 30 hours, and generation resources in turn are assigned performance credits to sell based on how much power they supplied during those same 30 hours. The buying and selling of credits happens through a centralized market, and the price is determined by a demand curve set by ERCOT that it believes will achieve a 1-in-10 LOLE. LSEs will likely pass the cost of those performance credits on to their end customers in some form.
E3 points out a risk factor about the PCM, which is that it is a novel mechanism and has not been tested in the United States before. However, E3 believes that the PCM would be easier to administer than the LSERO or FRM.
In my view, the biggest risk of the PCM is that it assumes that the near future will look like the recent past. For example, if 2025 turns out to be a very hot year with the 30 worst hours of system reliability being 6-8pm in July and August, that will create incentives for generators to build generation that performs well from 6-8pm in July and August. But 2026 might turn out to be a cold snap year, with something like Uri, in which the 30 worst hours are a sudden freeze for 3 days in February. It’s possible that resources designed to perform well from 6-8pm in July and August would also perform well during a February freeze. However, if the PCM had existed in 2020, I’m not sure that it would have prevented loss of load during Uri—the generation resources that performed well during the worst 30 hours of 2020 may have been natural gas plants that failed during Uri because of a lack of weatherization. E3 addresses this risk in its report, noting that the PCM may fail to account for a resource’s true reliability value, which would only be apparent during an extreme weather event.
The LSERO and FRM
By contrast, the LSERO and FRM are forward-looking and use an accreditation process to assign reliability credits, which ERCOT could use to account for infrequent extreme weather scenarios. Under these mechanisms, ERCOT would make an assessment of the amount of resources needed in the future to meet a 1-in-10 LOLE standard.
The LSERO and FRM begin by assigning a “marginal effective load carrying capacity” (ELCC) to each generation resource. The ELCC measures each resource’s ability to deliver energy when system conditions are expected to be at their tightest—in other words, it is a forward-looking estimate of how a generation source will perform when grid conditions are bad, as opposed to a backward-looking “how did this power plant perform last year when conditions were bad” as the PCM would do. The LSERO and FRM then assign “reliability credits” to individual generation resources based on their respective ELCCs.
The LSERO and FRM then assign requirements to purchase reliability credits to LSEs based on their share of load during the periods of the highest reliability risk. As with the PCM, the requirement to purchase reliability credits represents a wealth transfer from LSEs (who will probably pass the costs on to their customers) to generators. This wealth transfer should incentivize generators to keep additional power generation available and ready to deploy in times of system stress that would otherwise be money-losing in the absence of reliability or performance credits.
One unquantifiable benefit that E3 identifies is the lower financing cost of generation resources for which the owner can forecast revenues. Since the LSERO and FRM are a forward-looking mechanism, investors may be better able to forecast credit-based revenues for their resources relative to the PCM. This would make it easier for reliability-credit producing assets to raise construction debt. Intuitively, this is the same reason that banks require you to show proof of income when you want to take out a mortgage—the steadier your income, the more comfortable the bank is lending against it because they have higher confidence you can pay it back. E3 suggests that the LSERO and FRM may result in more predictable revenues for generators than the PCM.
The main difference between the LSERO and the FRM is how the trading of reliability credits happens between LSEs and generators. In the LSERO, LSEs must shop around and contract bilaterally with generators to meet their reliability credit requirements. In the FRM, ERCOT would administer a centrally-cleared market where purchase and sale of reliability credits happens automatically through an auction. From a financial markets perspective, central clearing tends to make sense when the market in question has a relatively homogenous product and when movement away from bilateral clearing would improve overall liquidity. E3 points out that the FRM’s central clearing would increase administrative burdens for ERCOT; in my view this cost of central clearing would be balanced by a reduction in administrative burdens and improved liquidity for generators and LSEs trading in a homogenous product like reliability credits. (The PCM also creates a centralized clearing process for performance credits.)
Regardless of the market settlement mechanism, E3 sees both the FRM and the LSERO as being administratively complex and having a 2-4 year implementation timeline.
The BRS, the DEC, and the BRS/DEC hybrid
The BRS and the DEC are conceptually different from the prior three options. I explain them in brief below.
The BRS seems almost like a “strategic petroleum reserve” for the Texas electricity market. The BRS authorizes ERCOT to buy up[4]“backstop generation,” which are resources that are only allowed to be deployed when the grid is in physical scarcity. The costs for procuring these supplies would be allocated to LSEs based on their share of load, as with the prior three mechanisms.
According to E3, one drawback of the BRS is that the definition of what qualifies as a backstop resource is very specific—a resource must be able to dispatch for 8 consecutive hours for 3 consecutive days. A resource that can dispatch for only 7.5 hours would be totally ineligible, even though such a resource would obviously be able to contribute to aggregate system reliability. Furthermore, the BRS requirement is not technology-neutral—more or less only natural gas fired generation will be able to meet the 8 hours/3 days requirement, so the BRS effectively becomes a subsidy directly to natural gas.[5]
The DEC mechanism requires LSEs to procure “dispatchable energy credits” on an annual basis. As with the BRS, the definition of what qualifies as a dispatchable resource is very specific—a start time of 5 minutes or less, net heat rate of <9000 LHV Btu/kWh, and sustained generation for 48 hours. Basically only certain types of natural gas-fired facilities will meet these criteria. An eligible resource would earn a DEC when it sells power between 6-10pm on any day of the year. As with the other mechanisms, LSEs would be required to purchase DECs, with the amount set to 2% of their annual MWh load.
E3 identifies one major risk factor with the DEC scheme: because the requirements to earn a DEC are very specific, natural gas resources that do not qualify for DECs may turn into money losers and may exit the grid. Because of the additional revenues DECs represent, DEC-eligible resources may bid into the market at levels below their marginal cost during the DEC window. This would enable them to steal away market share that would likely have gone to non-DEC eligible natural gas, harming their economics and potentially reducing overall system capacity. In other words, the DEC scheme may have a serious unintended consequence that would blunt its positive impact on overall system reliability.
The DEC/BRS hybrid simply merges the two approaches, with a DEC scheme supported by ERCOT procuring additional backstop generation. As with all of the policy proposals, LSEs eat the cost and will likely pass them on to their customers.
High-level conclusions
Overall, all of the reform proposals result in an increase in overall system cost through payments made from LSEs to generators (which will likely be passed on to customers). These payments will go to incentivize extra resources to be available during periods of high system need that would otherwise be uneconomical to operate absent the payments. E3 expects these marginal resources to be exclusively natural gas in the case of the BRS and the DEC, and largely but not exclusively natural gas in the case of the LSERO, FRM and PCM. E3 expects these latter three to result in the preservation of 5,630 MW of gas-fired generation that would otherwise have left the grid by 2026 (outcompeted by lower-cost solar and wind). Note that E3’s base case forecast already assumes ~40,000 MW of solar, wind and batteries to enter the grid by 2026 regardless of the reform proposal chosen.
[1] https://ferc.gov/media/february-2021-cold-weather-grid-operations-preliminary-findings-and-recommendations-full
[2] A LSE is a municipal utility, retail electric provider, or other firm that provides electric service to end users, such as households and industrial customers
[3] This is actually a “compliance period,” which is defined as a calendar year by the Texas Administrative Code.
[4] I don’t think ERCOT would actually own the power plants. ERCOT would form contractual agreements with individual generation resources that would be required to supply energy when called upon.
[5]Although E3 forecasts that natural gas resources would benefit from the LSERO, FRM and PCM, these mechanisms are technology-neutral, meaning that other types of generation could also receive payments if they can contribute to system reliability in some way.