Flex›The Flexibility Provider Base — Structure, Barriers, and the Aggregator Constraint

The Flexibility Provider Base — Structure, Barriers, and the Aggregator Constraint

The demand side of Swedish flexibility markets — DSOs procuring congestion management services — receives most of the policy attention. The supply side is less examined. Yet understanding who actually provides flexibility, how they are organized, and what prevents more of them from participating is essential for assessing whether markets can scale. The evidence is consistent: the active FSP base in any Swedish flex market is small, dominated by a handful of sophisticated actors, and structurally dependent on an aggregator layer that is itself thin. DSO demand and pricing are not the binding constraints — SWITCH markets offer 3,000–3,500 SEK/MWh for availability and go uncleared in most areas because no FSPs are present to supply. The organizational and infrastructure conditions for large-scale FSP participation do not yet exist.

Who participates today

Participation numbers are small across every active Swedish market:

At the TSO/balancing market level, the picture is broader but still concentrated. There are 28 registered BSPs as of March 2026 (Source - Svk Leverantörer av Balanstjänster 2026). Of these, around 6 are primarily demand-side or storage aggregators: Flower Infrastructure Technologies, Ingrid Capacity, Entelios, Capalo AI, Mind Energy, and Vimab BESS. The rest are generators, traders, and utilities participating primarily on the supply side.

Six aggregators for a country that is supposed to develop 1,387–2,523 MW of distribution-level flexibility over the next decade is thin market infrastructure.

Resource types in the market

The FSP base is more varied than the small numbers suggest:

  • Industrial consumers — flexible production loads, backup generators, process heat. The most commercially sophisticated participants. Also the most constrained by core-business risk: any activation that disrupts production quality carries a risk cost that can be 10–50× the direct energy value (Source - FlexAbility Delrapport 5 (2025)).
  • Battery storage — home batteries (Flower, Ingrid Capacity, CheckWatt portfolios — CheckWatt alone aggregates 15,000+ home BESS installations totalling 300+ MW across the Nordics (Source - CheckWatt Website (2025-2026)), dedicated BESS operators (Vimab BESS). Technically well-suited for rapid response; constrained by FCR-D wear and warranty terms.
  • District heating and CHP — Stockholm Exergi participates at TSO level; BeFlexible documented 36 MW of district heating in multi-market operation. Significant capacity but requires coordination across energy carriers.
  • Heat pumps and EVs — the largest potential pool by unit count; require aggregation to participate at any scale.
  • Plannable local production — GENAB’s use of idle CHP and gas turbines as plannable local capacity is an underappreciated model: 1 MW of plannable production ≈ 1 MW of transmission capacity relief (Source - GENAB Nätutvecklingsplan 2025-2034 (2024)).

The aggregator as necessary infrastructure

The most structurally important finding from the research: most flexible resources cannot participate in markets directly. They need an aggregator. Palm et al., in 25 interviews across FSPs and PFSPs in CoordiNet Uppland and Skåne, found that FSPs who had an aggregator partner would not have participated without one (Source - Palm et al LFM Drivers and Barriers (2023)). The aggregator provides bidding, activation communication, baseline calculation, and settlement. Without this layer, the resource owner — a building manager, a factory energy coordinator, an EV fleet operator — faces complexity they cannot absorb alongside their primary job.

For residential resources this is essentially universal. 65% of home battery owners use an aggregator (Source - FlexAbility Delrapport 4 (2025)). The 35% who don’t are largely self-managing behind Nord Pool spot contracts or fixed tariffs.

The aggregator constraint has a specific implication: the size of the FSP supply pool is bounded above by the coverage and capacity of the aggregator layer. If there are six active aggregators in Sweden, and each can manage a few hundred assets, the total accessible FSP base is in the low thousands at most. Growing FSP supply past that ceiling requires either growing the aggregator sector or creating routes to direct participation that are currently absent.

Drivers of participation

The research literature identifies a counterintuitive finding: financial incentives are not the primary driver of early LFM participation. Activation volumes are too low and prices too uncertain to justify purely economic entry. Most FSPs joined despite an unproven business case (Source - Palm et al LFM Drivers and Barriers (2023)).

What actually drives participation:

The champion concept is the most important driver and the least discussed in policy. Organizational participation depends on a single individual inside the organization — someone with personal interest, a mandate to explore, and enough credibility to get internal sign-off. Without the champion, the organization doesn’t engage, regardless of price signals. With the champion, organizations with no obvious flexibility assets find reasons to participate. This finding emerged spontaneously in Palm et al.’s interviews and was not covered in prior LFM research. It has significant policy implications: a market that can only recruit participants organization by organization, one champion at a time, cannot scale through price signals alone.

Other drivers: the potential for future revenue as markets mature; sustainability and public relations value (especially for PFSPs — organizations that hadn’t yet joined but wanted to be associated with the concept); and the ability to influence market design through early participation.

Barriers to supply

The information barrier — hardest for PFSPs

The primary reason organizations don’t participate is not cost but comprehension. The challenge is not access to information but the difficulty of understanding the market and then explaining it internally well enough to secure organizational mandate and budget. Large organizations with multiple divisions — municipalities, industrial companies — face internal coordination problems where the commercial division doesn’t know what the technical division is doing on flexibility, and vice versa. Approximately 15 potential FSPs in Palm et al.’s study declined even to be interviewed, citing insufficient knowledge — the information barrier prevented research participation, let alone market participation.

BSP absence — the most damaging structural gap

The Electricity Balancing Guideline required a proper BSP (Balance Service Provider) role by December 2020. Svk’s May 2024 “implementation” is described by market actors as a “paper construction” — formal compliance without operational substance (Source - FlexAbility Delrapport 5 (2025)). Full implementation is now deferred to 2028.

The practical consequence: an aggregator managing resources across 10 electricity suppliers must run 10 separate BRP administrative processes instead of one. One aggregator reports spending 50% of working time on BRP management for a portfolio of ~500 MW. The cross-BRP problem means that every customer served by a different electricity supplier creates a separate coordination thread with a separate Balance Responsible Party.

One estimate: independent BSP with proper cross-BRP aggregation infrastructure would allow +300 MW to enter markets immediately — resources that are technically qualified but locked out by administrative friction. The BSP deferral was ranked the worst-deteriorating barrier between spring 2024 and autumn 2024 in FlexAbility 5’s workshop.

The TSO/LFM revenue gap

In the early CoordiNet markets (V2020/21), TSO balancing market revenue was approximately 5× higher than local flexibility market revenue at comparable volumes (Source - Palm et al LFM Drivers and Barriers (2023)). This differential drives rational capacity allocation toward TSO products for FSPs who can serve both. In FlexAbility 5’s workshop, 50% of active flexibility providers reported increased interest in Nord Pool participation as an alternative to local flex markets (Source - FlexAbility Delrapport 5 (2025)). For markets where FSPs already exist, this upward diversion of capacity is a real dynamic. For most SWITCH markets, it is secondary to the more fundamental problem: the FSPs are not present at all. You cannot divert capacity that hasn’t been recruited.

FCR-D wear cost escalation

In September 2023, Svk raised the required number of FCR-D activations from approximately 60/year to approximately 3,000/year — a 50× increase. For battery systems, this creates significant cell degradation that shortens battery lifetime. For industrial processes, this frequency of activation is incompatible with production schedules. FCR-D participation is now effectively closed to many assets that were previously active in it (Source - FlexAbility Delrapport 5 (2025)). There is also a moral hazard: aggregators managing third-party batteries earn revenue from FCR-D activations but do not directly bear the degradation cost — the asset owner does.

Short contracts block investment

Season-by-season contracts prevent FSPs from investing in measurement, control, and connectivity infrastructure. If the contract is a winter season, the payback horizon for a 100,000 SEK control system investment doesn’t close. FSPs in Palm et al. are explicit: “If you have short contracts, it will never be possible to make any investments.” This is particularly binding for heat pump and EV aggregation, where the enabling technology (smart meters, communication interfaces, home energy management systems) must be installed at the customer site.

D-1 timing and pricing difficulty

Bidding closes day-ahead. Information gathered between bidding and activation — weather, EV arrival patterns, production schedules — cannot be used. For real-time-dependent resources this creates forecast error that becomes baseline risk. Separately, FSPs are required to price their own flexibility, and pricing it correctly requires knowledge of the grid constraint geometry and the DSO’s activation threshold that most resource owners don’t have: “I have no idea what price to set on flexibility. It can differ several hundred percent when we internally discussed reasonable prices” (Source - Palm et al LFM Drivers and Barriers (2023)).

No BRP compensation mechanism

When an aggregator activates resources in a customer portfolio, the customer’s BRP is left with an imbalance it did not forecast. Sweden has not implemented a BRP compensation mechanism. This creates friction: retailers and BRPs resist aggregation of their customers’ resources because they face uncompensated imbalance exposure. Finland and Denmark have resolved this at the TSO level; Sweden is waiting for the centralt datahanteringsverktyg, which names this as a required function (Source - FlexAbility Delrapport 5 (2025)).

How a compensation mechanism would work — and why the design is contested

The mechanics are straightforward in principle. An aggregator shuts off a factory’s HVAC for an hour and sells the load reduction into a flex market. The factory’s BRP had planned for the factory to consume X MWh; it actually consumed X − ΔX. The BRP is now over-long and must sell the surplus in imbalance settlement, typically at a loss. The direct cost to the BRP is real and calculable.

A compensation mechanism requires the aggregator to pay this cost, calculated from a referensprofil — a counterfactual baseline of what consumption would have been without activation — multiplied by a reference imbalance price. This is the design Ei proposed in 2021 (Source - Ei R2021-03 Oberoende Aggregatorer) and is consistent with what Denmark and Finland have implemented.

The design problem is that aggregator activations are not always harmful to the BRP. Sometimes the BRP’s overall portfolio is already long in the opposite direction — it has underestimated consumption elsewhere and is short. The aggregator’s activation of this particular customer (reducing their load) then worsens the BRP’s position; but the same activation at a moment when the BRP is already over-long reduces the BRP’s imbalance problem. The aggregator cannot observe the BRP’s real-time portfolio position, so it cannot know which case applies.

Over many activations, the net effect on any given BRP may be close to zero — some activations help, some hurt, and they roughly cancel out. A symmetric compensation formula would allow this to be netted: when an activation causes a cost, the aggregator pays; when it creates a benefit (reduces BRP imbalance), the BRP pays the aggregator. The expected annual net payment would be small in either direction.

The asymmetry problem — why Ei’s proposed formula disadvantages aggregators

Ei’s 2021 proposal did not adopt a symmetric design. It specified that compensation flows aggregator → BRP only, with benefits allowed to reduce what the aggregator owes but floored at zero. The formula:

Compensation = max(0, Direct costs to BRP − Benefits to BRP)

Benefits can cancel the aggregator’s liability but cannot reverse it. Even when the aggregator’s activation was net-beneficial to the BRP — saving the BRP more in imbalance costs than it imposed — the BRP keeps the windfall and the aggregator receives nothing.

Konkurrensverket (KKV) identified this as a structural entry barrier in its June 2021 consultation response (Source - Konkurrensverket Yttrande Ei R2021-03):

“På så vis skapas ett inträdeshinder vars signifikans ökar med nyttan som den potentiella aggregatorn skulle tillföra nätet.”

The logic: because compensation only flows one way, the aggregator’s expected compensation liability is always ≥ 0, even for aggregators whose activations are, on balance, beneficial to BRP portfolios. A symmetric formula would give expected net close to zero; the asymmetric floor pushes it positive.

More consequentially, the expected liability scales with activity level. A large, frequent aggregator making 400 activations per year generates 400 gross compensation events — each potentially resulting in a payment to the BRP. A small aggregator making 20 activations generates 20. The large, useful aggregator faces proportionally larger gross exposure under the one-directional formula, even if its net system contribution is positive. This is the “entry barrier grows with utility” dynamic: the formula’s cost burden is not symmetric to market value created; it is symmetric to scale of activity regardless of net direction.

The reference offer problem compounds this. KKV noted that the Ei formula functions as the starting point for bilateral negotiation between aggregators and BRPs. If the reference offer already puts aggregators in a structurally weaker position — they always potentially owe, BRPs never do — then all bilateral negotiations start from that disadvantaged baseline. A BRP with a favorable reference offer has less incentive to grant concessions; the aggregator’s outside option (the reference offer) is already worse than a symmetric design would provide.

Current status

Ellagen 8 kap. 16 § (in the forthcoming elmarknadslag, Prop. 2025-26-240) does require that BRP compensation terms take into account benefits to other BRPs — the legislative hook for a more symmetric design. But the actual methodology has not been designed, Ei has not approved any formula, and the infrastructure (centralt datahanteringsverktyg) to administer compensation flows does not yet exist. The compensation mechanism remains, as of 2026, a named requirement without operational substance — producing the friction documented in FlexAbility 5: BRPs resist aggregation of their customers and aggregators face uncompensated risk when proceeding anyway.

What the market data shows about supply

SWITCH’s empirical record reveals the supply problem directly. In the SWITCH market structure, DSOs place availability orders expressing how much capacity they want to buy, at stated prices. FSPs respond by offering capacity; the DSO awards what it needs. Zero clearing means the DSO placed orders — at 3,000–3,500 SEK/MWh, remuneration that is competitive with many TSO products — and no FSPs responded with supply.

The pattern across three seasons (Source - SWITCH Marknadsdata (info.switchmarket.se, 2026)):

  • E.ON deliberately over-orders in most markets to signal credible demand and attract FSPs — the demand side is not the constraint
  • 6 of 11 markets in V2025/26 had zero clearing: DSO orders placed and priced, no FSPs available to supply
  • The constraint is supply-side: few or no FSPs participate in most market geographies

Södra Skåne is the only market consistently meeting its DSO orders (13–59% fill rates across three seasons). Its supply advantage comes from several reinforcing conditions: (1) active since 2019, giving FSPs time to build participation routines and DSO-FSP relationships; (2) dense urban area with high concentration of home BESS; (3) large geographic uptake area covering most of southern Skåne, maximising the eligible resource pool; (4) large CHP resources providing a significant supply anchor (~20 MW); (5) E.ON’s deliberate market stimulation — over-ordering, over-activation even at low congestion risk, and above-standard remuneration — to maintain FSP engagement and market liquidity. Södra Skåne’s higher reported price (~5,800 SEK/MWh vs 3,000–3,500 elsewhere) is largely a mechanical effect of the impact factors (påverkansfaktorer ~0.6–0.68) applied in that market, not an indication that FSPs charge more in absolute terms.

Most other markets share none of these supply-side characteristics. Newer, smaller market areas with sparse FSP bases, no established participation routines, and no large anchor resources. The DSO has the budget to pay; there are simply no sellers.

The implication for policy is specific: the fix for thin SWITCH markets is not better-designed DSO demand signals. Those already exist and are well-priced. The fix is FSP recruitment and aggregator development specific to each market geography — exactly what is hard to do at scale.

What supply needs to scale

Seven conditions recur across the research as prerequisites for a larger FSP base:

  1. BSP role (full implementation, 2028): the single most damaging unresolved structural issue; enables cross-BRP aggregation without administrative overhead
  2. BRP compensation mechanism: removes retailer resistance to customer aggregation; requires datahanteringsverktyg infrastructure
  3. Longer contracts: multi-year agreements that justify enabling technology investment at the FSP site
  4. Prequalification portability: one prequalification valid across multiple flex markets and TSO products; NC DR is supposed to deliver this through the FIS and standardized CU/SP/SPU/SPG pathway
  5. Automation and standardized interfaces: OpenADR for dispatch, LFM-h/p/e for product standardization; manual bidding and tracking is not viable for small or residential resources
  6. Reliable activation signals: grid state data from DSOs in real time, so FSPs can manage their resources rationally around expected activation events; SWITCH’s onboarding model (DSOs push actuals, forecasts, and limits via API) is the right architecture
  7. Aggregator infrastructure: legal clarity on the aggregator role, compensation model, and NC DR pathway; without professional aggregators, small resources can’t participate

The champion problem at scale

The champion finding is the most troubling result from a policy perspective. If organizational participation depends on one engaged individual inside each organization, flex supply cannot scale through market mechanisms alone. The champion is not replicable by price signals. When the champion leaves or loses organizational support, participation lapses. When a PFSP’s champion can’t secure internal buy-in, the organization stays out indefinitely — even with technically suitable assets.

Policy typically assumes that correct price signals and reduced barriers automatically generate supply. The champion finding suggests this is incomplete. Scaling FSP supply likely requires structured flexibility audits that identify assets and convert PFSPs into FSPs; demonstration programs that create champions in new organizations by giving them a concrete experience of participation; and aggregators who absorb the organizational education burden as part of their service model. The NC DR FIS and standardized products address the technical and administrative barriers. They do not address the organizational and informational ones.

Data gaps

  • CheckWatt’s Sweden-specific installation count (15,000+ sites is Nordic-wide; Swedish share unknown)
  • Revenue share between TSO ancillary, DSO flex, and wholesale market in CheckWatt’s value stack (which layer contributes most to the 2.5×/4.0× SE3/Finland multiplier)