Nordic Power Price Divergence: Why Price Correlation and Information Value Have Decoupled

Nordic power price correlation has split north-south, but information value is still determined by generation structure, not coupling.
After the 2021-2022 energy crisis, Nordic day-ahead prices structurally diverged: by 2024, four southern zones in Norway and Sweden were strongly correlated with DE-LU (0.54-0.71), while four northern zones were weakly correlated (0.25-0.30). Forecasting experiments show neighboring-zone previous-day prices improved accuracy by at most 2.2%, with no difference between coupled and isolated zones; removing wind/solar forecasts raised MAE by up to 6% in Norway, 8-16% in Sweden, and 31% in DE-LU. The conclusion: price correlation reflects market coupling and transmission constraints, while information value is primarily determined by national generation structure; coupling cannot judge VRE forecast value.

After the 2021-2022 energy crisis, a structural shift in Nordic day-ahead power prices is being overlooked by market participants: prices in different bidding zones no longer move in sync. A study covering 2018-2025 digs deeper—does the divergence in price correlation also change the information that prices carry? The conclusion is surprising: correlation has diverged, but information value has not followed.
Price Correlation: From Synchronized to a North-South Split
The study measures intraday price correlations between each bidding zone in Norway and Sweden and the Germany-Luxembourg (DE-LU) region. Before 2019, all regions moved in close alignment; Sweden was the first to diverge in 2020, Norway followed in 2021, and this rift has persisted since, without converging as the energy crisis eased.
By 2024, the picture was quite clear: four southern zones maintained strong correlation with DE-LU, with intraday correlations between 0.54 and 0.71; four northern zones showed weak correlation, at only 0.25-0.30. This north-south divide broadly aligns with the geography of transmission capacity, hydropower endowment, and cross-border coupling, but the study does not stop at describing correlations.
Bidding zone is the smallest geographic unit for day-ahead trading in the power market, delineated by transmission system operators based on grid congestion. The Nordic market (Nord Pool) includes Norway, Sweden, Finland, Denmark, and others, with each country further divided into multiple bidding zones. When transmission capacity between two regions is sufficient and the grid is unconstrained, prices tend to converge; when transmission is constrained, prices diverge between regions. DE-LU refers to the Germany-Luxembourg bidding zone, the core reference price for the continental European power market. Intraday price correlation typically refers to the correlation coefficient of hourly day-ahead prices between two regions, with values closer to 1 indicating more synchronized price movements. The study uses this metric to measure market coupling, but note that high correlation does not equate to identical price levels—it only reflects synchronization in the direction of fluctuations.
Information Value: Coupling Does Not Determine the Value of Forecasting
The truly valuable findings come from forecasting experiments. Researchers used the same model (selected based on 2024 data) to forecast next-day prices for each region in 2025 at 11:00 daily, testing item by item which inputs improve forecast accuracy for each region—especially neighboring-zone previous-day prices and the wind and solar forecasts published by transmission system operators.
Neighboring-zone previous-day prices brought limited improvement: the maximum reduction in mean absolute error for any region was only 2.2%, and was virtually identical for coupled and isolated zones. In other words, even if a region is deeply coupled with continental European prices, the information increment provided by neighboring-zone prices is no greater than for isolated regions.
The value of wind and solar forecasts presents a completely different picture. After removing transmission system operator forecasts, Norway's mean absolute error rose by up to 6%, Sweden by 8%-16%, and DE-LU by 31%; this ranking also holds on 2024 data. Within Norway, the value of these forecasts showed no detectable dependence on reservoir storage levels.
Mean absolute error (MAE) is one of the most commonly used metrics for evaluating forecast accuracy, calculated as the average of absolute deviations between forecast and actual values, in the same units as power prices (e.g., EUR/MWh). In this study, researchers use MAE reduction to measure the marginal contribution of an input (such as neighboring-zone prices or VRE forecasts) to forecast accuracy.
Transmission system operator (TSO) is the entity responsible for grid operation and real-time dispatch, and in the Nordic region is typically also the official publisher of wind and solar output forecasts. VRE (variable renewable energy) forecasts published by TSOs include next-day output expectations for wind and solar, and are important inputs for day-ahead price forecasting. Researchers infer the information value of an input by "how much MAE rises when that input is removed," a method known in the forecasting literature as an ablation test.
Why Information Value Follows Generation Structure
The two threads together point to one judgment: price correlation reflects market coupling and transmission constraints, while the information value in prices is primarily determined by the composition of each country's generation system. Sweden and Germany have higher shares of variable renewable energy, so wind and solar forecasts naturally make a larger marginal contribution to price formation; Norway is dominated by hydropower, where reservoir scheduling and inflow are the dominant price variables, so the incremental information from wind and solar forecasts is therefore limited.
This also explains why no detectable association was found between reservoir storage levels and forecast value within Norway—during the sample period, storage levels did not change the way wind and solar forecasts contributed to price forecasting.
Reservoir storage level is a key state variable for price formation in hydropower-dominated systems. Hydropower accounts for over 90% of Norway's installed capacity, and reservoir levels directly determine the marginal cost of hydropower: when storage is ample, hydropower can generate at low cost to meet peaks; when storage is tight, it must be dispatched at higher cost, potentially triggering thermal marginal pricing. Therefore, in a pure hydropower system, inflow conditions and reservoir scheduling are the main drivers of price fluctuations, and the marginal impact of wind and solar forecasts on price formation is naturally small. By contrast, Sweden and Germany have higher VRE shares, where the intermittency of wind and solar output directly changes the supply-demand balance in the day-ahead market, and forecast errors are transmitted to prices through marginal unit switching, giving VRE forecasts higher information value. This logic also explains why no detectable association was found between storage levels within Norway and VRE forecast value—during the sample period, storage levels did not change the way VRE forecasts contributed to price forecasting.
Implications for VRE Forecast Production and Use
The policy implications of the study are direct and practical: a region's degree of coupling with continental Europe cannot be used as a basis for judging the value of VRE forecasts in that region. For participants seeking to decide where to produce, publish, and use wind and solar forecasts, the more reliable signal is the local generation structure, not price correlation metrics.
This also means that investing in forecasting capability in high-VRE markets such as Southern Europe or Germany follows a different return logic than in hydropower-dominated markets like Norway. Price correlation tells you how markets are connected; generation structure tells you where information is valuable.
Frequently Asked Questions
Does the divergence in price correlation mean Nordic market integration is regressing?
The study shows that divergence concentrated after 2020-2021 and has persisted, but the strength of correlation itself does not directly equate to market efficiency; transmission constraints and differences in resource endowment are more direct explanatory variables.
Why is the information increment from neighboring-zone prices so limited?
In the tests, neighboring-zone previous-day prices improved forecast accuracy by at most 2.2%, with no difference between coupled and isolated zones, indicating that spatial price transmission did not translate into equivalent forecasting information.
Do these conclusions apply to other markets?
The study covers Norway, Sweden, and DE-LU; the generalizability of the conclusions requires validation in more regions, but the logic that "coupling does not equal information value" is highly transferable.
One-line Summary
Nordic power price correlation has split north-south, but the information value of prices is still determined by generation structure, not coupling.
Paragraph Summary
After the 2021-2022 energy crisis, Nordic day-ahead prices showed structural divergence: by 2024, four southern zones in Norway and Sweden were strongly correlated with DE-LU (correlation 0.54-0.71), while four northern zones were weakly correlated (0.25-0.30). However, forecasting experiments show that neighboring-zone previous-day prices improved accuracy by at most 2.2%, with no difference between coupled and isolated zones; after removing wind and solar forecasts, Norway's MAE rose by up to 6%, Sweden by 8-16%, and DE-LU by 31%. The conclusion is that price correlation reflects market coupling and transmission constraints, while the information value in prices is primarily determined by national generation structure; coupling cannot be used as a basis for judging VRE forecast value.
Related articles

Gemini 4 Neon Review: Benchmark Scores Rival Carbon, API Pricing and Quota Breakdown
Google Gemini 4 Neon is now live, benchmark scores closely tracking Carbon, API pricing 3/15 at twice Hydrogen's rate. This article compiles Bilibili creator testing: Neon can replace 90% of Carbon scenarios, AntiGravity quota consumption is only one-third of Carbon, plus subscription channel restrictions and 2API ban risks.

Gemini 4 Helium Coming Soon? Signals of Google Loosening Subscription Restrictions
Google team member hints releases will continue; community speculates Gemini 4 Helium (Flash Lite) may be next. Subscription ban controversy gets a response as product lead says "Soon," suggesting restrictions may be lifted. This article covers Gemini developments, OpenAI Coding Agent and risk management, Doubao real-time audio-video API, and compute business dynamics.

AI-Assisted React Development in Practice: Optimizing a Video App and Handling Errors
A 14-year veteran developer demonstrates AI-assisted React video app optimization: compressing video from 80MB to 5MB (~94%) via ImageKit, plus reliable error handling and retry mechanisms. Includes planning, code understanding, and verification methods for working with AI.