The June 2024 PJM Base Residual Auction cleared BGE zone capacity at $466.35/MW-day, more than six times the prior year's $73. For BGE residential customers on Standard Offer Service, that auction translates directly into a retail generation rate impact. The path it took from the wholesale market to the residential bill is the case study.
Maryland regulators did not let the cost flow through proportionally across the delivery year. The Maryland Public Service Commission directed BGE to recover the incremental capacity cost only in six shoulder months: September through November 2025, and March through May 2026. December, January, and February were exempted. The June through August summer months were also excluded.
For a community solar developer or DG investor underwriting against BGE retail rates, the question this raises is not whether to model the capacity cost. It is how to read a tariff that has buried the cost inside a single bundled line item that does not move smoothly. BGE prices its residential generation service through Rider 1, with no capacity component disclosed. The shoulder-month rate pattern reflects the underlying capacity recovery, but the tariff itself never says so.
This paper walks through the auction result that triggered the rate action, the PSC's May 29, 2025 letter order that designed the shoulder-month recovery, why BGE's tariff structure obscures the wholesale-to-retail relationship, and how Sitara's component-level decomposition recovers the capacity layer from the bundled rate. The chart at the center of the analysis isolates the capacity contribution from energy across 29 months of actual BGE filed rates and reveals a structurally different rate profile than a flat escalator would imply.
The methodological argument is straightforward. A flat escalator applied to a bundled rate cannot distinguish between an energy market shift and a regulatory cost recovery decision. A component model can. For any firm modeling retail rate trajectories across the second half of the decade, that distinction matters.
The 2024 auction and the BGE zone result
The June 2024 PJM Base Residual Auction for the 2025/26 delivery year cleared BGE zone capacity at $466.35/MW-day. The prior year's auction had cleared BGE zone at $73/MW-day. The 6.4x increase reflected a combination of generation retirements, queue attrition, and resource-class ELCC adjustments that compressed available capacity across the PJM footprint, with the locational deliverability constraints in BGE producing one of the higher clearing prices in the RTO.
For a fully bundled retail customer on Standard Offer Service, the $466.35/MW-day capacity result flowed through to retail generation rates as an incremental cost layer. The Maryland Public Service Commission was the body responsible for designing how that cost would be recovered over the June 2025 through May 2026 delivery year.
The PSC had options. The conventional approach would have been proportional recovery across all twelve months, with the capacity adder showing up as a uniform per-kWh contribution to the bundled generation rate. The path the PSC chose was different.
The May 29, 2025 PSC letter order
The PSC issued a letter order on May 29, 2025, directing BGE to recover the 2025/26 incremental capacity cost only in six specified months. Capacity cost recovery in September, October, and November 2025. No incremental recovery in December 2025, January 2026, or February 2026. Recovery again in March, April, and May 2026. The peak summer months of June through August 2025 also did not receive the incremental adder; those months had been priced under a separate tariff filing covering the start of the delivery year.
The result is a rate profile that steps up sharply for three fall months, drops back to a baseline level for three winter months, and steps up again for three spring months. The capacity layer adds roughly $32.50/MWh in the months where it is recovered, and zero in the months where it is not.
The PSC characterized its own intervention as extraordinary in the order, signaling that the shoulder-month design was a response to the specific 2025/26 auction outcome rather than a permanent rate-setting approach. The implication is that future delivery years could see different cost recovery structures depending on the auction outcomes and the PSC's case-by-case judgment about bill impact management.
For a community solar developer modeling forward bill credit values against BGE retail rates, that judgment becomes a source of forecasting uncertainty distinct from the underlying wholesale market.
What the bundled tariff hides
BGE prices residential Standard Offer Service through Rider 1 as a single bundled generation rate. There is no separate line item for capacity. The customer's bill shows a generation charge denominated in cents per kWh. That charge embeds energy procurement cost, capacity, ancillary services, and any other generation-related cost component the utility is recovering through SOS.
From the filed tariff alone, an observer sees Rider 1 jump by approximately $32.50/MWh in shoulder months and return to a baseline level in non-shoulder months, with no explanation of what is driving the pattern. The tariff does not say "this is the capacity component." It says "this is the SOS rate for this month."
For a developer using a flat escalator approach, this structure is a problem. A flat escalator takes a recent rate level, applies an annual growth rate (typically 2-3%), and projects forward. There is no mechanism within the flat escalator framework to distinguish between a baseline energy charge and a temporary capacity adder layered on top. The model treats the rate as one number.
If the model is built off a shoulder-month observation, it overstates the long-run rate trajectory by treating the capacity step-up as the new permanent baseline. If the model is built off a non-shoulder-month observation, it understates the rate by ignoring the capacity recovery entirely. Either error compounds over a 20-year project life.
The fundamental problem is that the bundled rate does not behave like a continuous variable. It behaves like a base rate plus a regulatory overlay, and the overlay is on a separate cadence than the base.
The decomposition
The chart below shows BGE residential generation rates from January 2024 through May 2026, decomposed into energy and capacity components. The rate pattern in the chart is BGE's actual filed Rider 1 across 29 months. The attribution of the shoulder-month step-ups to capacity is derived from the May 29, 2025 PSC letter order, BGE's tariff filings, and the 2025/26 BRA clearing price.
Three features of the decomposition are worth highlighting.
First, the thin capacity layer visible in the months before September 2025 is not absent. It reflects the prior delivery year's BRA clearing price of $73/MW-day, scaled by the same wholesale-to-retail ratio that produces the 2025/26 capacity contribution. The decomposition method applies one ratio consistently across both years, anchored in the documented 2025/26 PSC treatment and applied backward to prior periods where the same generation rate construct was in effect.
Second, the energy layer itself steps up beginning in September 2025, from approximately $100/MWh to approximately $110/MWh. That movement is independent of the capacity treatment. It reflects energy market forwards, the underlying SOS procurement cycle, and seasonal adjustments. A bundled-rate observer sees only the combined movement and cannot tell that two different drivers are operating simultaneously.
Third, the shoulder-month pattern is sharp. The capacity layer adds roughly $32/MWh in September, October, and November 2025; falls to zero in December 2025 through February 2026; and returns to roughly $32/MWh in March, April, and May 2026. The bundled rate moves in lockstep with the capacity treatment because the PSC's intervention is the dominant driver of the shoulder-month differential.
The decomposition does not require any proprietary data. The inputs are publicly available BGE tariff filings, the publicly available PJM BRA clearing prices, and the publicly available PSC letter order. The work is in reading the regulatory record carefully and applying a consistent attribution framework.
Why this matters for forecasting
The methodological argument for component decomposition is not that bundled rate forecasting is wrong in every period. It is that bundled rate forecasting is unable to distinguish between two structurally different signals: a market-driven move in the underlying cost of energy, and a regulatory choice about how to recover a specific cost layer.
For BGE specifically, the 2025/26 delivery year is a clear case where the two signals diverged. The energy layer rose modestly. The capacity layer rose sharply. The PSC then chose to concentrate the capacity recovery in six months rather than spread it across twelve. A bundled-rate model cannot reconstruct any of those decisions. It can only observe the final blended output.
For community solar underwriting, the practical consequence is that the bill credit value calculation needs to track the underlying rate components, not the bundled output. A bill credit calculated against a smoothed average misses the within-year volatility entirely. A bill credit calculated against a peak-shoulder month overstates the annual revenue. A component model produces a forward curve where the energy layer follows its own trajectory, the capacity layer follows the BRA clearing path adjusted for any regulatory smoothing decisions, and the bundled retail rate emerges as the sum.
For project finance specifically, the component approach also enables scenario analysis that a bundled model cannot support. If a developer needs to stress-test against an upside or downside capacity scenario, the component model allows the capacity layer to be shifted independently of the energy layer. A bundled model has to be shifted as a single number, which conflates two distinct risks and produces stress tests that are difficult to interpret.