LFM Standard Product Design — Model A vs B Recommendations for DSOs

An analysis of the Model A and Model B variants for Sweden’s three Ei-approved Flexibility Market products — LFM-h, LFM-p, and LFM-e — with pros/cons and DSO recommendations grounded in Swedish and European evidence. See Source - Energiföretagen Förteckning Standardiserade Marknadsprodukter (2025) for full product specifications.

What Model A and B mean — and why the distinction is different per product

The A/B choice is not the same question for the availability products (LFM-h, LFM-p) as for the energy activation product (LFM-e). They must be read as two separate design decisions.

For LFM-h and LFM-p (availability/capacity products):

The critical point: Model A is not “accept or decline.” FSPs still submit energy activation bids and the lowest bid wins selection. The distinction from Model B is what the competition is on — in Model A, FSPs compete on their activation price while the availability rate is non-negotiable; in Model B, FSPs compete on the capacity/availability price itself. (Source - Energiföretagen Förteckning Standardiserade Marknadsprodukter (2025))

For LFM-e (energy activation):

Pricing distinction to keep in mind: In SWITCH market data and similar sources, the “avg avail. price” figure represents the DSO-offered availability rate — what the DSO posts per MWh (LFM-h) or MW (LFM-p) of reserved capacity. Under Model A this rate is DSO-set and non-negotiable; under Model B it would be the competitive clearing price. In both models, FSPs also submit energy activation bids — and under Model A it is specifically these activation bids that determine selection (lowest activation bid wins). The availability rate and the activation bid are two separate prices and should not be conflated. (Flexibility Market › LFM-h / LFM-p / LFM-e — Ei-approved standardized products (December 2025))

LFM-p — the period availability product

LFM-p procures availability over a multi-day or seasonal period, priced in kr/MW. It is the product closest to a traditional capacity market: the DSO pays for committed capacity regardless of whether it activates.

Model A (fixed period price) — recommended

The core insight from Swedish experience: the decisive finding from CoordiNet and subsequent seasons is that energy-only revenue is insufficient to sustain FSP participation. Activation occurs in only 6% of hours (2024/25 data); revenue is heavily weather-dependent; and year-to-year temperature variance means the same portfolio might activate 3× more in one winter than the next. A known, weather-independent capacity fee is what makes the FSP business case calculable. LFM-p Model A is the structural response to this finding. (Source - CoordiNet D4.7.2 Swedish Demonstration (2022), Source - FlexAbility Delrapport 2 (2025))

Complementary evidence from Palm et al.: FSPs report not knowing what price to set on their own flexibility. In a competitive availability auction (Model B), this uncertainty translates directly into either strategic high-bids or non-participation. A DSO-set price removes the bidding problem entirely for FSPs who simply need to decide whether the fee is worth accepting.

Practical track record: LFM-p’s equivalent (Säsongstillgänglighet in SWITCH) shows that when the DSO commits to a fixed seasonal rate, some markets — NÖ Skåne regional (100% fill at 236,000 SEK/MW) and Kallhäll (100% fill) — clear fully. The same markets have zero or near-zero Tillgänglighetsordrar (hourly) clearing, indicating FSPs value the revenue certainty of the period product when the hourly market is unreliable. (Source - SWITCH Marknadsdata (info.switchmarket.se, 2026))

Limitation of Model A: The DSO must set a price without a competitive reference. Setting too low means no supply; too high means overpaying with no pressure to improve. Current public price reference points in Sweden: 192,000–236,000 SEK/MW as standard rate (E.ON markets, 2024–26); 320,000–377,600 SEK/MW for Södra Skåne (impact factor-adjusted). JämtFlex quoted 500–800 SEK/MWh availability compensation. The range is wide enough to require local calibration.

Model B (competitive period auction) — appropriate only at high liquidity

Model B becomes useful once a market has enough independent FSPs to produce genuine competition. The Netherlands ENW experience is the warning case: thin competitive auctions produce strategic bidding at price caps (reported availability prices of ~11,000 EUR/MW — an outlier driven by a handful of FSPs gaming the auction). The same risk applies in Sweden: with only 3–7 FSPs per market area and limited price benchmarks, Model B’s theoretical efficiency advantage disappears. No current Swedish market appears to have sufficient depth. The markets cited in stakeholder interviews (Sassone et al. 2025) as having viable competitive dynamics are GB (UKPN, Northern Powergrid) and the Netherlands post-GOPACS maturation — not Sweden at current scale.

LFM-h — the hourly availability product

LFM-h procures availability per specific delivery hour up to D-2, priced in kr/MWh. It requires the DSO to forecast congestion several days ahead and issue targeted hour-specific calls.

Model A vs Model B — same conclusion, additional operational consideration

The same A/B logic as LFM-p applies. Model A is appropriate for DSOs at current market maturity; Model B requires competitive depth that does not yet exist.

The additional design question for LFM-h is whether to use it at all before having adequate forecasting capability. LFM-h is operationally more demanding than LFM-p because:

• The DSO must identify specific future congestion hours (not just a seasonal need)
• Callout timing is tight: availability must be called off at D-2 18:00 (or earlier — E.ON moved practical callout to D-2 10:30 in 2025/26 to allow aggregators to incorporate LFM-h into BRP planning). (Source - BeFlexible D5.2 Demo Planning and Deployment 2 (2025))
• Revenue per FSP is more variable hour-to-hour — less effective as a revenue floor than LFM-p

Recommendation: Use LFM-p as the primary availability instrument; add LFM-h as a supplementary product once the DSO has reliable day-ahead demand forecasting and has established FSP relationships. Running both simultaneously is valid — SWITCH data shows some markets use Säsongstillgänglighet (LFM-p equivalent) and Tillgänglighetsordrar (LFM-h equivalent) in parallel to serve different FSP types.

LFM-e — the energy activation product

LFM-e procures energy activation either D-1 (Auction 1, closes 09:30 D-1) or intraday (Auction 2, closes H-2). It can be used standalone or as the dispatch leg of an LFM-h/p commitment.

Model B (pay-as-bid) — recommended for now

Pay-as-bid is the European norm: the EC VITO 2025 study found pay-as-bid used across virtually all operational EU LFMs. This convergence is not accidental — it is the rational response to thin markets. (Source - EC LFM Specification and Design Criteria (VITO, 2025))

Why pay-as-bid limits risk in thin markets: With only 3–7 FSPs per area, marginal pricing (Model A) gives the most expensive accepted bidder market power — their bid sets the clearing price for all. Under pay-as-bid, each FSP receives only their own bid price, limiting the leverage of any single high-cost resource.

The strategic bidding counterargument: Pay-as-bid gives FSPs an incentive to shade bids upward — they won’t reveal true marginal cost if they expect to be accepted anyway. This is the classic criticism. Palm et al.’s finding that FSPs don’t know what price to set their flexibility at partially mitigates this in practice: bid inflation from genuine uncertainty is less harmful than bid inflation from calculated market power. Publishing a DSO willingness-to-pay indication (as ~40% of EU LFMs do) further limits strategic behaviour while preserving pay-as-bid.

Model A (pay-as-cleared) — appropriate once markets deepen

Pay-as-cleared becomes efficient when there are enough competing FSPs to produce genuine price discovery. At that point the dominant strategy (bidding true marginal cost) produces the socially optimal procurement cost. The transition from Model B to Model A for LFM-e should track market depth, not a calendar date.

VCG as a third option: Energiforsk 2026-1151 (Source - Energiforsk 2026-1151 Effektauktioner med Värmepumpar (2026)) demonstrated that VCG (Vickrey-Clarke-Groves) auctions theoretically outperform both alternatives in thin markets by incentivising truthful cost revelation. However, VCG is not currently supported by SWITCH or NODES, its settlement mechanism is complex to implement, and it generates such low absolute payout levels in simulated thin markets (1,600–16,000 SEK/year for heat pump portfolios) that no commercial aggregator could justify the infrastructure cost. VCG is a medium-term research direction, not a near-term operational option.

The LFM-p vs LFM-h choice: a prior design decision

Before reaching the Model A/B question, DSOs must choose between LFM-p and LFM-h as the primary availability product. These serve different congestion profiles:

The forecasting burden differs between the two products only on the availability commitment side: LFM-p requires estimating a seasonal need rather than specific hours, which is a less precise task. LFM-h adds a D-2 availability callout step that requires the DSO to identify which hours will be congested several days ahead.

However, both products carry the same activation forecasting requirement. Once capacity is standing (whether LFM-p seasonal or LFM-h hourly), the DSO must still forecast accurately when to trigger LFM-e energy activation. Activating on every available hour quickly becomes more expensive than the grid investment it was meant to defer — the cost-effectiveness case for flexibility depends on targeting the hours where congestion actually materialises. A DSO without reliable activation forecasting will over-procure energy, eroding the value case regardless of which availability product it chose.

LFM-p is therefore the simpler starting point in that it removes one forecasting step (the D-2 hourly callout), not in that it removes the need for forecasting altogether. Both products require the same operational forecasting discipline to be financially feasible for the DSO.

Summary recommendation matrix

Suggested starting combination for DSOs new to flex procurement: LFM-p Model A + LFM-e Model B. This provides the revenue predictability FSPs need to enter the market while limiting worst-case activation price outcomes. Add LFM-h once internal forecasting is ready.

National best practice vs local variation

The case for national convergence on model structure

Fragmentation has a direct cost: aggregators operating across multiple DSO markets must build separate operational workflows for different auction mechanics. SWITCH and NODES already create two parallel commercial tracks; adding A/B variation within each would multiply the complexity. FSPs who lack price benchmarks (a documented barrier — Palm et al.) benefit from knowing that the market mechanism is the same regardless of which DSO they serve. The NC DR will impose harmonization requirements within a few years; starting to converge now reduces transition cost.

The case for local variation within the standardized framework

The right product mix genuinely depends on the congestion profile. Chronic structural congestion (urban transformer district with persistent winter peak) calls for LFM-p and a higher fixed price; seasonal intermittent congestion calls for LFM-h. Hässleholm and Bålsta — the two most active Swedish markets by different metrics — have completely different underlying grid dynamics and would rationally use different product combinations.

Liquidity also varies enough that the Model A→B transition threshold will be reached at different times in different areas, if ever. Mandating a uniform model at the national level risks locking in the wrong design for markets at opposite ends of the maturity spectrum.

Conclusion: harmonize the structure; preserve local pricing and product mix

The Energiföretagen standardization already achieves the correct balance: it standardizes product names, attributes, and interoperability (enabling FSP participation across platforms) while leaving Model A/B as a per-DSO design choice. That is the appropriate architecture.

Near-term national default that most DSOs should gravitate toward: LFM-p Model A + LFM-e Model B. This combination has the strongest empirical support from Swedish and European experience at current market maturity. DSOs with more active markets (and higher liquidity) can diverge toward Model B for availability once that threshold is genuinely met — which in Sweden means a meaningful shift in FSP depth, not just market age.

The deeper caveat: product design choices operate within constraints that no model selection resolves. Two structural barriers dominate:

1. The charity problem: DSOs earn no regulated return on flex procurement under the pre-2028 CAPEX regime. The TOTEX reform (Ei RP5, 2028–2031) is the real structural enabler. Until then, DSO engagement depends on individual commitment rather than aligned incentives. (Source - FlexAbility Delrapport 5 (2025))

2. The TSO subscription two-equilibrium problem: Svk’s temporary subscription grant policy sets a de facto price ceiling. When Svk grants subscriptions readily, DSO willingness to pay drops to near zero regardless of product design; when it denies them, DSOs pay any price for the marginal MWh. Market design cannot resolve this without TSO-DSO coordination reform. See TSO-DSO Coordination — The Central Design Problem.

A DSO with the optimal Model A/B combination but no resolution to these two structural issues will still find their market undersupplied or unnecessary in the wrong season.

Related pages

• Flexibility Market — Full Swedish and EU LFM landscape; product specifications; FSP barrier analysis
• TSO-DSO Coordination — The Central Design Problem — The subscription mechanism and two-equilibrium problem in depth
• Why Swedish Local Flex Markets Are Thin — Structural Causes — Eight root causes of thin liquidity
• The Flexibility Provider Base — Structure, Barriers, and the Aggregator Constraint — Supply-side structure and barriers
• DSO Flexibility Valuation — Methods and Swedish Evidence — DSO willingness to pay and the CAPEX bias
• Baseline Methods — Verification and settlement methodology (separate from but linked to product design)