---
id: VOLT-HOME-WP-070
title: "Did 15-minute market coupling change cross-border price convergence?"
slug: did-15-minute-market-coupling-change-cross-border-price-convergence
description: "A descriptive before-and-after daily panel finds wider neighboring-zone spreads after the 15-minute transition date, but cannot attribute that difference causally to market coupling."
published: 2026-08-30
cluster: "Grid coupling, flows, and outages"
status: measured
evidence: "/research-data/home-papers/did-15-minute-market-coupling-change-cross-border-price-convergence.json"
figure: "/research-media/home-papers/did-15-minute-market-coupling-change-cross-border-price-convergence.webp"
figure_alt: "Chart for Did 15-minute market coupling change cross-border price convergence?: post-minus-pre mean neighboring-zone daily spread, shown as pre, post."
source_ids:
  - entsoe-sdac
  - ec-sdac-15m
  - iea-electricity-2026
  - acer-retail-2025
  - volt-architecture
peer_reviewed: false
---

## Abstract

European day-ahead markets moved to finer market time units around the end of September 2025. This paper compares neighboring-zone daily mean-price spreads before and after 2025-10-01. For every matched border-link-day, it computes the absolute difference between the endpoint zones’ daily mean prices. It then subtracts the pre-period mean spread from the post-period mean spread.

The frozen evidence reports 47,461 post-period border-link-days and a post-minus-pre difference of 53.8148442615 EUR/MWh. Neighboring-zone daily mean spreads are wider, not narrower, in the post-period under this pooled descriptive comparison. The result does not show that 15-minute coupling caused widening. Season, market regime, energy fundamentals, border composition, data coverage, and the increased ability to observe intrahour variation coincide with the timing change.

This is not a causal policy evaluation. There is no untreated control group, parallel-trends demonstration, event-time model, or adjustment for zone and date conditions. Daily averaging also means the paper does not directly measure quarter-hour convergence. The valid conclusion is that observed average daily spreads differ across the chosen timing split. The result is relevant to household tools that must preserve native market intervals, but it is not a claim about retail bills, welfare, or trading performance.

## Plain-language answer

After the 2025-10-01 split used in the analysis, the average absolute daily price difference between neighboring zones is 53.815 EUR/MWh higher than in the earlier observations. That points to less convergence in the simple before-and-after data.

It does not tell us that 15-minute market coupling made prices diverge. Many other things changed between the periods, and the post period is much shorter and occurs in a different seasonal and system context. Finer prices can also reveal variation that hourly aggregation previously hid.

For households, the practical message is about data handling rather than a causal verdict. Controllers and tariff studies should use native 15-minute prices where they apply and should not assume that a market-design date alone explains changes in local-versus-neighbor prices.

## Research question

The registered question asks whether 15-minute market coupling changed cross-border price convergence. The implemented estimand is a timing contrast:

`mean absolute neighboring-zone daily spread after or on 2025-10-01 - mean absolute spread before 2025-10-01`.

This measures a change in observed average spread across two periods. It does not isolate “market coupling” as a treatment. SDAC existed before the move to the finer market time unit; the relevant change is temporal granularity within the coupled day-ahead market. The title is retained from the registry, while the paper uses precise language for the actual comparison.

“Convergence” means a smaller absolute difference in endpoint daily mean prices. A positive post-minus-pre value therefore indicates wider spreads. The question concerns wholesale bidding-zone prices, not the all-in retail price paid by households.

## Data and provenance

The evidence file is `/research-data/home-papers/did-15-minute-market-coupling-change-cross-border-price-convergence.json`. It is marked `measured` and frozen at publication cutoff `2026-08-30T00:00:00Z`. The corpus lists daily prices from 2021-01-01 through 2026-08-29, detailed intervals beginning 2025-10-01, and long history beginning 2015-01-01. The before-and-after metric uses matched daily border-link spreads.

The registered source contracts are `border_flows`, `day_ahead_prices`, `outage_events`, `generation_mix`, and `zone_load`. Flow-link identity and endpoint prices enter the comparison. Outage, mix, and load variables are not controls in the reported value.

The analysis used a read-only transaction and a 180-second timeout. Snapshot SHA-256 is `7e97489fc8528c8cc8c38830e05b48d949ce1f67b98575dff26f5d7c321e4c67`. Analysis-code SHA-256 is `57c57de79cdab2b5b6d6c54c485cb5162598c5ba0b0bfe995da40d75e6c52ba9`; protocol SHA-256 is `adb36bf6b447af9f96339249b8becaefc20422499cca1977242866347a97bd4b`; paper-registry SHA-256 is `7bcb91d7476d0a69fe9fa75a5c7782f8117e0153f82f9112b7e1d307d3943717`; and source-registry SHA-256 is `07949550ac443ff673fda5c0209b99f137544f3ffecf6775f109bb9d09663bd6`.

ENTSO-E’s SDAC page provides the official coupled-market context and records the transition to 15-minute market time units. The European Commission source dates and describes the day-ahead transition. The IEA provides broader system and flexibility context. ACER and CEER explain why consumer exposure depends on retail contract design. Voltcast’s architecture documents the 2025-10-01 native-resolution seam used internally. The measured 53.8148442615 EUR/MWh contrast comes from the frozen evidence.

## Method

The common grid routine maps daily flow observations by source zone, destination zone, and UTC date, then matches a daily mean price for both endpoint zones. Each complete row contains the absolute difference in endpoint mean prices.

Rows dated before 2025-10-01 form the pre group. Rows on or after 2025-10-01 form the post group. The primary result subtracts the pre-group arithmetic mean from the post-group arithmetic mean. The evidence sample size of 47,461 is the size of the post group passed into the outcome, not the total number of pre and post rows. The pre count is not reported and is not invented.

The design uses no matching, regression adjustment, border balancing, or weighting. It does not ensure that the same links contribute equally to both periods. It analyzes daily mean spreads even though the institutional change concerns market time units. That choice gives a broad temporal comparison but limits mechanism.

The regenerated evidence reports no bootstrap interval and names the interval method “not reported for this estimand.” It therefore provides no uncertainty bound around the 53.8148442615 EUR/MWh timing contrast.

## Results

The reported post-period sample contains 47,461 matched border-link-days. The mean absolute neighboring-zone daily spread after minus before 2025-10-01 is 53.81484426151043 EUR/MWh. Rounded, average spreads are 53.815 EUR/MWh wider in the post group.

Under the study’s definition, wider spreads mean less observed convergence. The point estimate therefore does not show an improvement in daily cross-border convergence after the split.

The figure labels are “pre” and “post,” with mean spread values 15.422906160941695 and 69.23775042245212 EUR/MWh, rendered as 15.4 and 69.2. Post minus pre reproduces the primary 53.81484426151043 EUR/MWh contrast. No interval is reported and no statistical-significance claim is made.

There are no secondary results. Pre and post means, pre count, border-specific contrasts, quarter-hour spread distributions, seasonal controls, or household bill effects are not in the JSON. The paper does not infer them from the figure.

## Robustness and placebo checks

The date rule is explicit and reproducible, and absolute spreads make convergence symmetric with respect to endpoint order. Matching both endpoint prices prevents one-sided rows. These checks ensure a clear descriptive calculation.

They do not identify treatment effects. A credible robustness analysis would balance links across periods, compare the same seasons, include zone-pair and date controls, inspect pre-trends, and use event-time coefficients. A difference-in-differences design would require a defensible comparison group not exposed to the same transition. The evidence contains none.

Placebo dates before the actual seam could show whether similar jumps occur under arbitrary splits. Native-interval analysis could distinguish newly visible intrahour dispersion from daily price-level separation. Alternative windows could test sensitivity without turning window selection into outcome-driven search. Again, no such output is reported.

The preregistered family requires association language and names Holm control for inferential claims. This paper makes no causal or significance claim. Its strongest robust statement is that the frozen pooled date split produces a positive post-minus-pre spread difference.

## Limitations

The before-and-after periods are not exchangeable. The post sample is bounded by 2026-08-29, while the pre period begins in 2021. They differ in length, season composition, fuel and weather regimes, capacity, demand, outages, and market conditions.

The composition of represented borders can change with data availability. A pooled mean can move if high-spread links contribute more rows after the split. The method does not report a balanced panel or link fixed effects.

Daily mean prices are a coarse outcome for a 15-minute design change. Finer market units can change dispatch and bidding, but they can also reveal intrahour values that hourly series averaged. Using daily means suppresses the shape that would help separate these channels.

The study has no causal control. Market redesign, season, system stress, renewable conditions, network constraints, and other developments coincide. Scheduled flow and prices are jointly determined in the coupled market. The result cannot attribute 53.815 EUR/MWh to the time-unit change.

Wholesale spread is not retail bill spread. Dynamic contracts vary in pass-through; fixed contracts, taxes, supplier margin, network tariffs, and metering rules alter household exposure. No customer savings or welfare outcome is measured.

## Practical implication

Household-energy systems should preserve market-time-unit metadata and avoid comparing pre- and post-seam series as though their measurement were identical. Native 15-minute prices should remain native through scheduling and aggregation, with DST and completeness handled explicitly.

For public claims, say that the frozen daily panel shows wider average neighboring-zone spreads after the split. Do not say the market change caused divergence or raised household bills. A causal answer requires a new design with balanced borders, event-time controls, and native intervals.

## Reproducibility

Build the matched directional border-day panel from flow-link identities and endpoint daily mean prices. Calculate the absolute endpoint spread. Partition rows at 2025-10-01, with earlier dates in pre and dates on or after the cutoff in post. Subtract the pre arithmetic mean from the post arithmetic mean. Expected published outputs are post sample size 47,461 and difference 53.81484426151043 EUR/MWh.

Record UTC-date semantics, endpoint mapping, missing matches, period counts, and link composition. A local-date cutoff, balanced panel, seasonal window, interval-level outcome, or regression-adjusted design is a different analysis. Preserve the current null interval rather than inventing uncertainty for the difference. Licensing appears at `/legal/data-licensing` and the canonical licensing document.

## Disclosure

Analysis and drafting were model-assisted. This public working paper is not peer reviewed. It explicitly makes no causal claim for 15-minute coupling and exposes daily-panel, period-composition, and confounding limits. No household savings or trading performance are measured. Volt has no live traders or live capital. This paper is not trading advice or financial advice.

## References

- [ENTSO-E — Single Day-ahead Coupling (SDAC)](https://www.entsoe.eu/network_codes/cacm/implementation/sdac/)
- [European Commission — EU electricity trading in the day-ahead markets becomes more dynamic](https://energy.ec.europa.eu/news/eu-electricity-trading-day-ahead-markets-becomes-more-dynamic-2025-10-01_en)
- [International Energy Agency — Electricity 2026](https://www.iea.org/reports/electricity-2026)
- [ACER and CEER — Rewarding flexibility: How retail contract choice can help unlock consumer flexibility](https://www.ceer.eu/wp-content/uploads/2025/11/ACER-CEER-2025-Retail-monitoring.pdf)
- [Voltcast — Voltcast Architecture](https://github.com/ossedk/voltcast/blob/main/docs/voltcast/ARCHITECTURE.md)