pith. machine review for the scientific record. sign in

arxiv: 2604.14257 · v1 · submitted 2026-04-15 · 💰 econ.GN · q-fin.EC· stat.AP

Recognition: unknown

Mapping the causal structure of price formation in Texas's transitioning electricity market

Shiva Madadkhani , Nils Sturma , Mathias Drton , Svetlana Ikonnikova
Authors on Pith no claims yet

Pith reviewed 2026-05-10 11:52 UTC · model grok-4.3

classification 💰 econ.GN q-fin.ECstat.AP
keywords causal discoveryelectricity marketswind generationprice formationTexas ERCOTrenewable integrationnatural gas pricesday-ahead prices
0
0 comments X

The pith

Wind generation has become the dominant causal driver of day-ahead electricity prices in Texas, with effects more than three times larger than those of natural gas prices.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper applies causal discovery techniques to Texas electricity market data to map how supply and demand factors shape wholesale prices. It finds that wind generation now exerts the strongest causal influence on day-ahead prices, exceeding the effects of natural gas prices by a factor of more than three and challenging the longstanding view of the market as gas-driven. The results also trace how wind power lowers prices in some locations while shifting congestion costs across regions in seasonal patterns. Demand and gas prices show smaller, more variable causal roles depending on time and location. The work indicates that price models must track evolving causal links as renewables expand and demand patterns change.

Core claim

Applying causal discovery to observational data from Texas's electricity market shows that wind generation has become the dominant causal driver of day-ahead prices, with effects more than three times larger than those of natural gas prices and overturning the view of the market as primarily gas-price-driven. Wind reduces prices locally but redistributes congestion costs across regions in seasonally varying patterns. Natural gas prices retain causal relevance with modest influence and a changing dominant benchmark, while demand effects vary by region and period.

What carries the argument

Causal discovery algorithms that recover the directed causal graph among wind generation, natural gas prices, electricity demand, and regional day-ahead prices from time-series market observations.

Load-bearing premise

The causal discovery algorithm accurately recovers the true causal structure from observational market data despite time-varying relationships, regional heterogeneity, potential unmeasured confounders, and possible violations of standard assumptions such as faithfulness or no hidden variables.

What would settle it

A sustained period in which day-ahead prices respond more strongly and directly to natural gas price changes than to wind generation changes, after controlling for demand and other observed factors, would falsify the claim of wind dominance.

read the original abstract

Electricity markets are changing, driven by large-scale renewable integration and rising demand from electrification and digitalisation. This raises fundamental questions about how electricity prices form as the relationships among key price determinants evolve. Here we apply causal discovery to characterise these dynamics across major supply- and demand-side drivers of wholesale electricity prices in Texas, where rapid renewable growth intersects with surging demand. We show that wind generation has become the dominant causal driver of day-ahead electricity prices with effects more than 3 times larger than those of natural gas prices, overturning the view of the Texas market as gas-price-driven. Wind reduces prices locally but redistributes congestion costs across regions in seasonally varying patterns. Natural gas prices remain causally relevant, though their influence is modest and the dominant gas benchmark changes over time. Electricity demand also shows region- and period-specific causal effects. These findings highlight the need for causal models that capture time-varying relationships across both supply and demand to guide system planners and market participants navigating the ongoing transition.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript applies causal discovery to observational time series from the Texas electricity market to characterize evolving causal relationships among wind generation, natural gas prices, electricity demand, and day-ahead wholesale prices. It concludes that wind has become the dominant causal driver with effects more than three times larger than those of natural gas prices, while documenting time-varying, region-specific effects of congestion and demand.

Significance. If the causal claims are robust, the results would be significant for energy economics by providing data-driven evidence that renewable integration has overturned the gas-price-driven paradigm in Texas, with implications for market design, congestion management, and policy under electrification. The empirical focus on time-varying causal structures is a strength.

major comments (3)
  1. [Methods] Methods section: the specific causal discovery algorithm, its assumptions (faithfulness, no hidden variables), data sources, sample periods, and procedures for handling non-stationarity or regime shifts must be explicitly stated and justified, as these directly determine whether the recovered graph supports the claim of wind dominance.
  2. [Results] Results section: the reported effect sizes (wind >3× natural gas) require the underlying causal effect metrics, confidence intervals, and sensitivity analyses to unmeasured confounders (e.g., transmission congestion, curtailment) to be shown; without these, the quantitative dominance claim cannot be evaluated.
  3. [Discussion] Discussion or robustness subsection: validation against known interventions (policy changes, weather events) or alternative specifications is needed to address potential violations of standard causal discovery assumptions in a non-stationary market setting.
minor comments (2)
  1. [Abstract] Abstract: include a brief statement on the causal discovery method and data period to allow readers to assess the findings immediately.
  2. Figures: ensure causal graphs and effect-size tables have captions that define all variables, time windows, and units so the time-varying patterns are self-contained.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments, which identify key areas where additional methodological transparency and robustness checks will strengthen the paper. We address each major comment below and will revise the manuscript to incorporate the requested details and analyses.

read point-by-point responses

  1. Referee: [Methods] Methods section: the specific causal discovery algorithm, its assumptions (faithfulness, no hidden variables), data sources, sample periods, and procedures for handling non-stationarity or regime shifts must be explicitly stated and justified, as these directly determine whether the recovered graph supports the claim of wind dominance.

    Authors: We agree that explicit documentation of these elements is necessary for readers to assess the recovered causal structure. In the revised manuscript we will add a dedicated Methods subsection that names the algorithm (PC with kernel-based conditional independence tests), states the core assumptions (faithfulness and causal sufficiency), lists the precise ERCOT and EIA data sources with sample periods (2015–2023), and describes our handling of non-stationarity via rolling-window estimation and regime-shift detection, with justification tied to the known policy and weather-driven breaks in the Texas market. revision: yes

  2. Referee: [Results] Results section: the reported effect sizes (wind >3× natural gas) require the underlying causal effect metrics, confidence intervals, and sensitivity analyses to unmeasured confounders (e.g., transmission congestion, curtailment) to be shown; without these, the quantitative dominance claim cannot be evaluated.

    Authors: The manuscript derives the >3× dominance claim from the estimated causal effects in the discovered graph; we will now report the underlying numeric effect sizes together with bootstrap confidence intervals. We will also add sensitivity analyses that bound the possible influence of unmeasured confounders such as transmission congestion and curtailment, using the methods of Cinelli & Hazlett (2020) adapted to the time-series setting, and present these results in the Results section. revision: yes

  3. Referee: [Discussion] Discussion or robustness subsection: validation against known interventions (policy changes, weather events) or alternative specifications is needed to address potential violations of standard causal discovery assumptions in a non-stationary market setting.

    Authors: We will insert a new robustness subsection that validates the main findings against documented interventions (e.g., the 2021 Winter Storm Uri and subsequent market-rule changes) and compares results under alternative specifications, including different conditional-independence tests and stationarity adjustments. This will directly address concerns about assumption violations in the non-stationary environment. revision: yes

Circularity Check

0 steps flagged

No significant circularity in empirical causal discovery application

full rationale

The paper applies standard causal discovery algorithms to observational time-series data from the Texas electricity market to recover directed graphs among variables such as wind generation, natural gas prices, demand, and day-ahead prices. The headline result (wind as dominant driver with >3× effect size) is an output of this data-driven procedure rather than a mathematical derivation or fitted parameter that reduces to the inputs by construction. No self-definitional equations, predictions of closely related quantities from the same fit, or load-bearing self-citation chains appear in the described workflow. The analysis remains self-contained against external benchmarks and does not rename known results or smuggle ansatzes via citation; therefore the circularity score is 0.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Only the abstract is available for review; full model specification, algorithm details, and data processing steps are unknown. Causal discovery methods generally rest on standard statistical assumptions about the data-generating process.

axioms (2)
  • domain assumption Causal discovery assumes causal sufficiency (no unmeasured confounders) and faithfulness of the observed distribution to the underlying causal graph.
    These are core assumptions of most causal discovery algorithms applied to observational data such as market prices.
  • domain assumption The causal relationships among price drivers are time-varying and can be recovered by the chosen method across regions and seasons.
    The abstract emphasizes time-varying and region-specific patterns as central to the findings.

pith-pipeline@v0.9.0 · 5482 in / 1410 out tokens · 41271 ms · 2026-05-10T11:52:15.511380+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

59 extracted references · 2 canonical work pages

  1. [1]

    E., Roney, C

    Bistline, J. E., Roney, C. W., McCollum, D. L. & Blanford, G. J. Deep decarbonization impacts on electric load shapes and peak demand.Environmental Research Letters16, 094054 (2021)

    2021

  2. [2]

    & Rhodes, J

    Shaffer, B., Quintero, D. & Rhodes, J. Changing sensitivity to cold weather in texas power demand. iScience25(2022)

    2022

  3. [3]

    Characteristics and risks of emerging large loads

    NERC. Characteristics and risks of emerging large loads. Tech. Rep., North American Electric Reliability Corporation (2025). URL https://www.nerc.com/globalassets/who-we-are/ standing-committees/rstc/whitepaper-characteristics-and-risks-of-emerging-large-loads.pdf

    2025

  4. [4]

    & Cowie-Haskell, A

    Norris, T., Profeta, T., Patino-Echeverri, D. & Cowie-Haskell, A. Rethinking load growth: Assessing the potential for integration of large flexible loads in US power systems. Tech. Rep., Duke University (2025). URL https://hdl.handle.net/10161/32077

    2025

  5. [5]

    R., Medlock, K

    Hartley, P. R., Medlock, K. B., III & Hung, S. Y. ERCOT and the Future of Electric Reliability in Texas. Tech. Rep., Rice University’s Baker Institute for Public Policy (2024). URL https://www. bakerinstitute.org/sites/default/files/2024-01/BIPP-CES-ERCOT-Reliability-020724.pdf

    2024

  6. [6]

    Energy Information Administration

    U.S. Energy Information Administration. Data centers and cryptocurrency mining in Texas drive strong power demand growth. U.S. Energy Information Administration,Today in Energy(2024). URL https://www.eia.gov/todayinenergy/detail.php?id=63344. Accessed: 2025-11-14

    2024

  7. [7]

    Energy Information Administration

    U.S. Energy Information Administration. Short-term energy outlook data browser (2026). URL https://www.eia.gov/outlooks/steo/data/browser/#/?v=22. Accessed: 2026-02-25

    2026

  8. [8]

    Electricity balance sheet; supply and consumption (2026)

    Statistics Netherlands (CBS). Electricity balance sheet; supply and consumption (2026). URL https://www.cbs.nl/en-gb/figures/detail/84575ENG. Data for 2024. Accessed: 2026-03-31

    2026

  9. [9]

    & Diaz Anadon, L

    Navia Simon, D. & Diaz Anadon, L. Power price stability and the insurance value of renewable technologies.Nature Energy10, 329–341 (2025)

    2025

  10. [10]

    & Pacheco, A

    Woo, C.-K., Horowitz, I., Moore, J. & Pacheco, A. The impact of wind generation on the electricity spot-market price level and variance: The texas experience.Energy Policy39, 3939–3944 (2011)

    2011

  11. [11]

    & Horowitz, I

    Woo, C.-K., Zarnikau, J., Moore, J. & Horowitz, I. Wind generation and zonal-market price divergence: Evidence from texas.Energy Policy39, 3928–3938 (2011)

    2011

  12. [12]

    Ketterer, J. C. The impact of wind power generation on the electricity price in germany.Energy economics44, 270–280 (2014). 26

    2014

  13. [13]

    & Zoppoli, P

    Clò, S., Cataldi, A. & Zoppoli, P. The merit-order effect in the italian power market: The impact of solar and wind generation on national wholesale electricity prices.Energy Policy77, 79–88 (2015)

    2015

  14. [14]

    Rintamäki, T., Siddiqui, A. S. & Salo, A. Does renewable energy generation decrease the volatility of electricity prices? An analysis of Denmark and Germany.Energy Economics62, 270–282 (2017)

    2017

  15. [15]

    Ajanaku, B. A. & Collins, A. R. Comparing merit order effects of wind penetration across wholesale electricity markets.Renewable Energy226, 120372 (2024)

    2024

  16. [16]

    & Weron, R

    Jędrzejewski, A., Lago, J., Marcjasz, G. & Weron, R. Electricity price forecasting: The dawn of machine learning.IEEE Power and Energy Magazine20, 24–31 (2022)

    2022

  17. [17]

    & Meinshausen, N

    Peters, J., Bühlmann, P. & Meinshausen, N. Causal inference by using invariant prediction: iden- tification and confidence intervals.Journal of the Royal Statistical Society Series B: Statistical Methodology78, 947–1012 (2016)

    2016

  18. [18]

    & Blaxland, L

    Cacciarelli, D., Pinson, P., Panagiotopoulos, F., Dixon, D. & Blaxland, L. Do we actually understand the impact of renewables on electricity prices? A causal inference approach.iEnergy4, 247–258 (2025)

    2025

  19. [19]

    Mjelde, J. W. & Bessler, D. A. Market integration among electricity markets and their major fuel source markets.Energy Economics31, 482–491 (2009)

    2009

  20. [20]

    & Wilhelmsen, M

    Ferkingstad, E., Løland, A. & Wilhelmsen, M. Causal modeling and inference for electricity markets. Energy Economics33, 404–412 (2011)

    2011

  21. [21]

    & Hamori, S

    Nakajima, T. & Hamori, S. Testing causal relationships between wholesale electricity prices and primary energy prices.Energy Policy62, 869–877 (2013)

    2013

  22. [22]

    Park, H., Mjelde, J. W. & Bessler, D. A. Price dynamics among US electricity spot markets.Energy Economics28, 81–101 (2006)

    2006

  23. [23]

    & Fallani, F

    Castagneto-Gissey, G., Chavez, M. & Fallani, F. D. V. Dynamic Granger-causal networks of electricity spot prices: A novel approach to market integration.Energy Economics44, 422–432 (2014)

    2014

  24. [24]

    Tiedemann, S.et al.Identifying elasticities in autocorrelated time series using causal graphs.arXiv preprint arXiv:2409.15530(2024)

    work page arXiv 2024

  25. [25]

    & Spirtes, P

    Glymour, C., Zhang, K. & Spirtes, P. Review of causal discovery methods based on graphical models. Frontiers in Genetics10(2019). 27

    2019

  26. [26]

    Models, reasoning, and inference

    Pearl, J.CausalitySecond edn (Cambridge University Press, Cambridge, 2009). Models, reasoning, and inference

    2009

  27. [27]

    & Scheines, R.Causation, prediction, and searchSecond edn

    Spirtes, P., Glymour, C. & Scheines, R.Causation, prediction, and searchSecond edn. Adaptive Computation and Machine Learning (MIT Press, Cambridge, MA, 2000)

    2000

  28. [28]

    & Schölkopf, B.Elements of Causal Inference: Foundations and Learning Algorithms, 205–208 (The MIT Press, 2017)

    Peters, J., Janzing, D. & Schölkopf, B.Elements of Causal Inference: Foundations and Learning Algorithms, 205–208 (The MIT Press, 2017)

    2017

  29. [29]

    & Fox, E

    Shojaie, A. & Fox, E. B. Granger causality: A review and recent advances.Annual Review of Statistics and Its Application9, 289–319 (2022)

    2022

  30. [30]

    & Camps-Valls, G

    Runge, J., Gerhardus, A., Varando, G., Eyring, V. & Camps-Valls, G. Causal inference for time series.Nature Reviews Earth & Environment4, 487–505 (2023)

    2023

  31. [31]

    & Sejdinovic, D

    Runge, J., Nowack, P., Kretschmer, M., Flaxman, S. & Sejdinovic, D. Detecting and quantifying causal associations in large nonlinear time series datasets.Science Advances5, eaau4996 (2019)

    2019

  32. [32]

    URL https://link

    Du,P.Renewable Energy Integration for Bulk Power Systems: ERCOT and the Texas Interconnection Power Electronics and Power Systems (Springer International Publishing, 2023). URL https://link. springer.com/10.1007/978-3-031-28639-1

    work page doi:10.1007/978-3-031-28639-1 2023

  33. [33]

    Texas counties map

    Texas Open Data Portal. Texas counties map. https://data.texas.gov/dataset/Texas-Counties-Map/ 48ag-x9aa (2023). Accessed: 2025-11-04

    2023

  34. [34]

    Energy Information Administration

    U.S. Energy Information Administration. Form EIA-860: Annual Electric Generator Report, 2019–

    2019

  35. [35]

    URL https://www.eia.gov/electricity/data/eia860/

  36. [36]

    Oil & gas data query; general production query

    Railroad Commission of Texas. Oil & gas data query; general production query. https: //webapps2.rrc.texas.gov/EWA/productionQueryAction.do (2025). Oil production by county, January–December 2024. Accessed: 2025-11-04

    2025

  37. [37]

    K., Zhu, S

    Zarnikau, J., Woo, C. K., Zhu, S. & Tsai, C.-H. Market price behavior of wholesale electricity products: Texas.Energy Policy125, 418–428 (2019)

    2019

  38. [38]

    & Ikonnikova, S

    Madadkhani, S. & Ikonnikova, S. Toward high-resolution projection of electricity prices: A machine learning approach to quantifying the effects of high fuel and CO2 prices.Energy Economics129, 107241 (2024)

    2024

  39. [39]

    & Runge, J

    Gerhardus, A. & Runge, J. inHigh-recall causal discovery for autocorrelated time series with latent confounders(eds Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M. & Lin, H.)Advances in Neural Information Processing Systems, Vol. 33 12615–12625 (Curran Associates, Inc., 2020). 28

    2020

  40. [40]

    Electricity price forecasting: A review of the state-of-the-art with a look into the future

    Weron, R. Electricity price forecasting: A review of the state-of-the-art with a look into the future. International Journal of Forecasting30, 1030–1081 (2014)

    2014

  41. [41]

    & Roh, H

    Fell, H. & Roh, H. Wind energy, transmission, and production costs: Does increased connectivity help all? (2019). URL https://cenrep.ncsu.edu/cenrep/wp-content/uploads/2020/01/Roh_JMP.pdf. Job market paper, Accessed: 2026-01-19

    2019

  42. [42]

    Energy Information Administration

    U.S. Energy Information Administration. Electricity data browser: Net generation for all sectors, Texas, annual (2024). URL https://www.eia.gov/electricity/data/browser/

    2024

  43. [43]

    2023 state of the market report for the ERCOT electricity markets

    Potomac Economics. 2023 state of the market report for the ERCOT electricity markets. Tech. Rep., Potomac Economics (2024). URL https://www.potomaceconomics.com/wp-content/uploads/2024/ 05/2023-State-of-the-Market-Report_Final_060624.pdf

    2023

  44. [44]

    Energy Information Administration

    U.S. Energy Information Administration. Difference in natural gas prices between Texas and Henry Hub narrowed in first half 2023. U.S. Energy Information Administration,Today in Energy(2023). URL https://www.eia.gov/todayinenergy/detail.php?id=60180

    2023

  45. [45]

    Losada Carreño, I.et al.Potential impacts of climate change on wind and solar electricity generation in Texas.Climatic Change163, 745–766 (2020)

    2020

  46. [46]

    L., Steininger, K

    Prol, J. L., Steininger, K. W. & Zilberman, D. The cannibalization effect of wind and solar in the california wholesale electricity market.Energy Economics85, 104552 (2020)

    2020

  47. [47]

    Energy Information Administration

    U.S. Energy Information Administration. ERCOT increasingly meets rising demand with solar, wind, and batteries. U.S. Energy Information Administration,Today in Energy(2025). URL https: //www.eia.gov/todayinenergy/detail.php?id=66464

    2025

  48. [48]

    2024 state of the market report for the ERCOT electricity markets

    Potomac Economics. 2024 state of the market report for the ERCOT electricity markets. Tech. Rep., Potomac Economics (2025). URL https://www.potomaceconomics.com/wp-content/uploads/2025/ 06/2024-State-of-the-Market-Report.pdf

    2024

  49. [49]

    Resources and Day-Ahead operations

    Electric Reliability Council of Texas (ERCOT). Resources and Day-Ahead operations. Market Education Resource, Electric Reliability Council of Texas (2023). URL https://www.ercot.com/files/ docs/2023/07/25/2024_01%20Resource_DA_Ops.pdf

    2023

  50. [50]

    Slusarewicz, J. H. & Cohan, D. S. Assessing solar and wind complementarity in Texas.Renewables: Wind, Water, and Solar5, 1–13 (2018)

    2018

  51. [51]

    Granger, C. W. J. Investigating causal relations by econometric models and cross-spectral methods. Econometrica37, 424–438 (1969). 29

    1969

  52. [52]

    Testing for causality: A personal viewpoint.Journal of Economic Dynamics and Control 2, 329–352 (1980)

    Granger, C. Testing for causality: A personal viewpoint.Journal of Economic Dynamics and Control 2, 329–352 (1980)

    1980

  53. [53]

    A., Schölkopf, B

    Mastakouri, A. A., Schölkopf, B. & Janzing, D.Necessary and sufficient conditions for causal feature selection in time series with latent common causes, Proceedings of Machine Learning Research, 7502–7511 (PMLR, 2021)

    2021

  54. [54]

    & Drton, M

    Harris, N. & Drton, M. PC algorithm for nonparanormal graphical models.Journal of Machine Learning Research14, 3365–3383 (2013)

    2013

  55. [55]

    & Wasserman, L

    Liu, H., Han, F., Yuan, M., Lafferty, J. & Wasserman, L. High-dimensional semiparametric Gaussian copula graphical models.The Annals of Statistics40, 2293–2326 (2012)

    2012

  56. [56]

    & Wasserman, L

    Liu, H., Lafferty, J. & Wasserman, L. The nonparanormal: semiparametric estimation of high dimensional undirected graphs.Journal of Machine Learning Research10, 2295–2328 (2009)

    2009

  57. [57]

    C., Guo, Z

    Constantinou, A. C., Guo, Z. & Kitson, N. K. The impact of prior knowledge on causal structure learning.Knowledge and Information Systems65, 3385–3434 (2023)

    2023

  58. [58]

    Causal reasoning with ancestral graphs.Journal of Machine Learning Research9, 1437– 1474 (2008)

    Zhang, J. Causal reasoning with ancestral graphs.Journal of Machine Learning Research9, 1437– 1474 (2008)

    2008

  59. [59]

    & Spirtes, P

    Richardson, T. & Spirtes, P. inCausal inference via ancestral graph models(eds Green, P., Hjort, N. & Richardson, S.)Highly Structured Stochastic Systems83–105 (Oxford University Press, USA, 2003). 30

    2003