What the Heck is ‘Enhanced Price Formation’ in PJM?
FERC recently closed the comment period on the U.S. DOE’s proposal to provide a guaranteed rate of return for generators in competitive markets that have 90 days of fuel on site. Kudos to PJM Interconnection for filing terrific comments pushing back on the DOE/FERC proposal, asserting it would undermine competitive markets and seeks to solve a problem not supported by facts.
As an alternative, PJM requested that FERC direct it to embark on energy market price formation reforms. In fact, PJM had promoted price formation reforms earlier this year. And, although DOE provided no technical discussion of why price formation reform was needed, it was the number one recommendation from it’s grid reliability report, and various price formation FERC dockets were listed in DOE’s proposed resiliency rule.
Last week, PJM released its proposal for enhancing energy price formation, including changing the way locational marginal prices (LMP) are set, and revising shortage pricing.
Let’s start by acknowledging that price formation concepts and proposed solutions are highly esoteric and require advanced degrees in economics, mathematics, or similar fields to truly understand.
In a separate blog, I explore some key questions that immediately arise from this proposal. Here, I’ll do my best to explain the proposed changes to LMP, as I currently understand them (and within blog-length).
In extremely simple terms, current rules require the market price of energy to be set by the cost of the last generation unit needed (i.e. the marginal unit) to provide power to serve customer demand. This is called marginal pricing and it is a well-known economic concept.
In the past, units (mostly natural gas-fired) that can easily increase and decrease power (i.e. are flexible) usually had high costs and operated on the margin. Now, with cheap and plentiful shale gas, the flexible natural gas units that are allowed to set price have lower costs, therefore reducing market prices.
At certain times and locations, the market price becomes too low for some large and inflexible (i.e. can’t easily ramp power up and down) generators to operate, yet PJM still needs them to provide power. If these resources were flexible, they would just incrementally ramp down power production. But, inflexible generators typically can’t (due to operational constraints) – or are unwilling to (due to economics) – ramp down. Typically, inflexible units are nuclear or coal technologies.
If PJM still needs the power from a large inflexible unit to meet customer demand, it can ask other (usually smaller) flexible resources to ramp down in order to keep the inflexible unit operating. Since the market price is not signaling these flexible resources to turn down, PJM can request them to reduce output, then pay side payments to compensate.
On the other hand, PJM can keep all these resources online (if needed) and pay the inflexible unit side payments cover its additional costs not covered by the energy market price.
These side payments, called “uplift”, are problematic because they are real costs not incorporated into market prices. Uplift payments can inhibit investment, lack transparency, and make it harder for those in the market to predict or manage costs (e.g. hedge against transmission constraints).
The solution PJM proposes is to let all resources set market price. This seems reasonable, right? Turns out it is more complex.
Per marginal cost principles, the energy market is supposed to reflect the short-term variable costs of generator operations, which are expected to increase as output increases (a principle called the convex condition). This enables additional power resources to be dispatched by PJM based on lowest cost to meet demand.
However, for large, inflexible units that can only be dispatched in chunks (think large blocks of power rather than increments), costs may initially be high then actually decrease as output increases within the block, a phenomenon called non-convexity. These high initial costs can be related to, for example, economies of scale, fixed start-up, no load costs, etc. In this situation, marginal pricing logic wouldn’t work.
As a result, PJM proposes moving to an extended LMP (eLMP) framework to address the non-convexity by having a two stage dispatch model. The first stage would clear for dispatch in a security-constrained manner (e.g. accounting for conditions that impact the transmission system). The second run would be for price, where flexible and inflexible units could compete to set price. PJM states that certain fixed costs (e.g. start-up) would be considered in addition to variable costs, and a convex relaxation method (which I don’t fully understand due to mathematical complexity) would be used to allow flexible and inflexible units to set market prices.
PJM estimates this proposal will annually increase energy market prices by $3.50/MWh, reduce uplift by $80 million, add $20 million in cost to compensate resources for following dispatch, and reduce capacity market costs between $1.2 – $2.2 billion. The net impact would be $440 million – $1.44 billion or a 2%-5% increase in annual costs. These costs do not seem to include impacts related to scarcity pricing, a part of PJM’s proposal that is not covered in this blog.
PJM further states these changes would avoid future capacity costs to load if its current 24% reserve margin dips near its 16.6% installed reserve margin (due to resource retirements from low revenues), and capacity costs rise.