Flex›Source - Konsultrapport Kostnadsincitament Regionnät Termicum (2025)

Source - Konsultrapport Kostnadsincitament Regionnät Termicum (2025)

Document: Kostnadsreducerande incitament för region- och transmissionsnät, Termicum (Kristofer Månsson och Magnus Söderberg), 2025-12-08, reviderad 2025-12-23. Uppdrag för Energimarknadsinspektionen (dnr 2025–103385).

Source type: Consultant report (Ei procurement — analytical input to RP5 regulatory methodology)

Purpose

Ei commissioned Termicum to develop a methodology for cost-reduction incentives (effektiviseringskrav) for regionnät (regional distribution networks) and transmissionsnät (Svenska kraftnät), for use in RP5 (2028–2031) and beyond.

The core problem: the standard benchmarking methods (DEA, SFA, yardstick competition) used for lokalnät require many comparable companies. Sweden has only 5 large regionnät, several small ones, and Svk as the single transmission operator — too few for robust statistical analysis.

Background: current approach

In previous regulatory periods, regionnät and transmissionsnät companies were assigned a flat 1% annual efficiency requirement (effektiviseringskrav), while lokalnät received individually calibrated requirements based on multi-company benchmarking. The flat 1% was acknowledged as a rough approximation rather than an evidence-based efficiency target.

Regulatory theory: why incentives are needed

The report covers standard regulatory theory relevant to understanding the Swedish context:

Averch-Johnson effect: Under traditional cost-plus (avkastnings-)reglering, a monopolist has an incentive to overinvest in capital because allowed return is tied to the asset base. This provides a theoretical underpinning for the CAPEX bias problem identified in the wiki — DSOs under a capital-return-based system will systematically prefer asset investment over operating solutions (including flexibility).

KPI-X regulation: Price or revenue cap set at CPI − X, where X is the expected productivity improvement. Requires calibrating X accurately — too low and the DSO earns excess profit; too high and quality or investment suffers.

Benchmarking methods (for context):

  • DEA (Data Envelopment Analysis): non-parametric frontier analysis
  • SFA (Stochastic Frontier Analysis): parametric frontier; handles data noise better
  • TFP/MTFP (Total Factor Productivity / Multilateral TFP): index-based productivity measurement

Proposed methodology: MTFP

Termicum recommends MTFP (Multilateral Total Factor Productivity) as the most suitable method for small-sample regulated entities:

Advantages:

  • Relatively robust with small sample sizes (vs. DEA/SFA which need many observations)
  • Transparent and communicable
  • Less vulnerable to legal challenge than black-box frontier models (several examples of successful regulatory challenges against DEA/SFA in Australia, Netherlands, Germany, Finland documented in appendix)

Two-variable production models work best:

  • Model 1: Abonnemang + ledningslängd (subscriptions + line length)
  • Model 2: Abonnemang + utmatad energi (subscriptions + distributed energy)

Both give similar results. Maximal effekt (peak demand) excluded — not statistically significant. Nätförluster excluded — too highly correlated with line length.

Critical finding on Opex opåverkbar: Uncontrollable opex (primarily upstream grid subscription costs, which fluctuate with Svk/regionnät pricing decisions) creates extreme volatility in MTFP outcomes when included. A single large change in upstream subscription cost can swing a company’s apparent “productivity” by 20–30 percentage points in one year. Recommendation: exclude Opex opåverkbar from MTFP inputs.

Minimum data requirement: ≥8 years for stable MTFP estimates (individual anomalous years have less weight).

Recommendations per company type

Large regionnät (5 companies):

  1. Estimate annual MTFP trend using regression (smoothes out noisy years)
  2. Set the common frontier change equal to the second-best performing company (not the single best, which may be an outlier in a small sample)
  3. Convert productivity trend to cost requirement at ~50% translation rate (some productivity improvement is due to factors outside management control)
  4. Run two parallel scenarios: with and without Opex opåverkbar

Small regionnät (wind/production-connected networks):

  • Individual MTFP too volatile due to extreme sensitivity to small changes in subscription costs or investments
  • Recommendation: same flat X-factor as large regionnät until data improves

Svk (single national TSO):

  • Fundamentally: only one entity → no cross-company comparison possible
  • Recommendation: internal benchmarking — compare Svk with itself over a sufficiently long time series; average productivity over the period becomes the norm
  • Interim measure: same X-factor as large regionnät until long enough time series exists

Relevance to CAPEX bias discussion

The report’s discussion of the Averch-Johnson effect is directly relevant to the wiki’s analysis of why DSOs prefer grid investment over flexibility procurement. The Averch-Johnson effect is the formal economic description of what the wiki calls the “CAPEX bias” — it arises from the structure of return-on-capital regulation, not from any intentional DSO choice. The TOTEX reform (ellagen 5 kap 12a§, Ei RP5) is specifically designed to counteract this by giving flexibility procurement comparable weight to investment in the revenue cap.

Relevance to wiki topics