In recent years, the penetration of renewable energy sources such as wind and solar power has transformed electricity markets worldwide. Understanding how these fluctuating renewables influence electricity prices has become a critical question for policymakers, system operators, and market participants. Researchers from Imperial College London, in collaboration with the UK National Energy System Operator, have released a groundbreaking study that deploys a rigorous causal machine learning framework to untangle the complex relationship between renewable generation and UK electricity prices. Their research offers novel insights that challenge simplistic assumptions and open new avenues for adaptive market design and policy.
Unlike traditional analyses that rely heavily on correlations, this study meticulously distinguishes correlation from causation, a crucial difference given the myriad confounding factors affecting market prices. By leveraging UK market data spanning from 2018 to 2024, the team applies advanced causal inference techniques to predict how incremental changes in wind and solar output impact day-ahead and intraday electricity prices. This approach controls for external drivers such as electricity demand fluctuations, fossil fuel price volatility, seasonal cycles, and diurnal patterns, enabling nuanced context-specific estimates that reflect the actual market dynamics.
Central to their methodology is the use of a local, partially linear Double Machine Learning (DML) framework, which orthogonalizes both electricity prices and renewable generation against a rich set of covariates. These covariates include gas prices, carbon permit costs, hourly timestamps, monthly seasonal variations, installed renewable capacity, as well as daylight duration. The DML approach employs cross-fitting and bootstrap procedures to mitigate overfitting risks often encountered in observational data analysis and provides robust uncertainty quantification through confidence intervals. This level of statistical rigor ensures that the results reliably approximate plausible causal effects rather than simple associative trends.
The empirical findings reveal striking non-linearities in the price effects of renewables, debunking the common assumption of constant marginal impacts. For wind power, the causal price effect forms a distinct U-shaped curve relative to penetration levels. At both low and high levels of wind penetration, an additional gigawatt-hour of predicted wind generation corresponds to substantial day-ahead price reductions. However, at moderate penetration levels, the downward price pressure is less pronounced. This nuanced pattern suggests that system conditions and market response mechanisms vary significantly across penetration regimes.
Solar power exhibits a somewhat different but related dynamic. The study finds that at lower penetration levels, predicted solar output consistently suppresses electricity prices, though the magnitude of this effect diminishes as penetration rises. Anticipating solar capacity growth trajectories, the researchers project that solar price impacts may increasingly resemble the U-shaped pattern observed with wind as the market enters more saturated capacity states. Importantly, the general trends discerned for intraday market prices align closely with those observed in the day-ahead timeframe, supporting the robustness of these causal inferences across trading horizons.
A critical component of this analysis is the rigorous control of confounding factors. The initially observed anomalous “mid-range price bumps” detected in descriptive statistics are largely neutralized when adjusting for variables such as demand shifts and fuel cost fluctuations. This realignment with classical merit-order theory underscores the indispensability of causal frameworks in distinguishing spurious correlations from genuine system responses. It signals that interventions and policy formulations predicated on raw correlation analyses risk misestimating the impact of renewables on price formation processes.
The researchers emphasize that the increasing magnitude of renewable-induced price effects over time is consistent with growing penetration and evolving fundamental market drivers. This temporal evolution highlights practical implications for market architecture and regulatory design. Support mechanisms, capacity adequacy assessments, and pricing regulations may achieve greater efficacy if dynamically adapted to prevailing penetration and seasonal conditions rather than relying on static, average effect assumptions. Such time-sensitive responsiveness could better balance incentives and system stability in the ongoing energy transition.
Nonetheless, the authors prudently acknowledge the limitations inherent in their observational study framework. Residual unobserved confounding cannot be completely ruled out, and the granularity and quality of available data, particularly at very high penetration levels, constrain the precision of causal estimates in these regimes. Hence, the researchers position their work as a foundational, transparent contribution that invites scrutiny, replication, and iterative refinement as data quality improves and additional contextual factors are incorporated.
This study’s novel causal machine learning pipeline and its documented open-source implementation provide a valuable resource for analysts and regulators. They enable extension to other electricity price signals, such as balancing market prices, and adaptation to different energy markets globally where comparable data exist. By furnishing a replicable framework for context-aware causal inference, the research empowers stakeholders to make informed decisions based on robust evidence rather than simplistic assumptions.
Dr. Davide Cacciarelli, lead author of the study, emphasizes the study’s practical orientation: “Our goal is to furnish transparent, actionable estimates that system operators, government regulators, and researchers can validate, customize, and enhance to meet the rapidly accelerating demands of the energy transition.” Professor Pierre Pinson further notes the importance of the methodological advances in separating correlation from credible causal signals in complex market settings.
The research was published in the November 6, 2025, issue of iEnergy, a peer-reviewed, fully open access journal hosted by the prestigious Tsinghua University Press. iEnergy is dedicated to disseminating high-impact research that propels the development of future power and energy systems. Its wide international authorship and comprehensive indexing attest to its growing influence in the scientific community. The journal’s commitment to open access until 2026 facilitates broad dissemination and uptake of cutting-edge findings such as those reported in this causal analysis of renewables and electricity price dynamics.
In sum, this pioneering investigation reframes our understanding of renewable energy’s influence on electricity markets with robust causal inference techniques and detailed data-driven insights. By uncovering the non-linear, penetration-dependent price effects of wind and solar generation, it equips market designers and policymakers with the knowledge needed to optimize the energy transition’s economic and operational outcomes in the UK and beyond.
Subject of Research: The causal impact of wind and solar renewable energy penetration on UK electricity prices, analyzed via advanced machine learning techniques.
Article Title: Do we actually understand the impact of renewables on electricity prices? A causal inference approach
News Publication Date: 6-Nov-2025
Web References:
- Article DOI: 10.23919/IEN.2025.0027
- Journal iEnergy: https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=9732629
Image Credits: iEnergy
Keywords: causal machine learning, renewable energy, electricity prices, wind power, solar power, UK electricity market, penetration effects, Double Machine Learning, price dynamics, energy transition, market design, intraday electricity prices
