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

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.

• The paper reveals that West Texas wind forecast is the dominant causal driver of system lambda, outpacing natural gas prices by a mean ratio of 3.4.
• It applies the LPCMCI algorithm to build temporal causal graphs that isolate renewable, demand, and congestion effects in regional price differentials.
• It concludes that evolving market fundamentals, including shifts in gas hub relevance and dynamic load effects, necessitate adaptive forecasting models.

Mapping Causal Dynamics in ERCOT: Structural Shifts in Price Formation

Context and Motivation

The Texas electricity sector is undergoing large-scale structural transformation, marked by rapid renewable penetration and surging demand from industrial expansion, electrification of oil and gas operations, and burgeoning data center activity. These trends have reshaped both aggregate consumption and the spatial-temporal distribution of load and generation, creating complex price dynamics that are not well captured by traditional associational or static models. This paper addresses the imperative for temporally explicit causal modeling in ERCOT, analyzing the evolution of fundamental price drivers between 2019–2024 by leveraging advances in causal discovery for time series.

Figure 1: Geographic distribution of generation capacity, natural gas infrastructure, and demand centres across Texas, highlighting the concentration of wind, solar, and natural gas power plants and key demand and gas hubs.

Methodological Framework

The study applies the Latent Peter-Clark Momentary Conditional Independence (LPCMCI) algorithm, which accounts for latent confounding and autocorrelation—limitations of conventional methods such as Granger causality. LPCMCI recovers temporal causal graphs delineating actionable links between renewable generation, natural gas hub prices, regional forecast loads, and price components, with regime-specific analyses across four market regimes (peak/off-peak, warm/cool) and rolling two-year windows.

Day-ahead hub prices are decomposed into system lambda ($\lambda$) and congestion-driven hub price differentials ($\Delta_h$), facilitating isolation of system-wide versus locational effects. Domain constraints are integrated via exogeneity assumptions, blocking physically implausible causal links in the algorithm's search space.

Figure 2: Causal graph for peak hours during cooler months, showing supply-side, demand-side, natural gas, and price components with directed and undirected edges inferred by the LPCMCI algorithm.

Results: Dominance of Wind Generation

System Lambda

West Texas wind forecast emerges as the dominant causal driver of $\lambda$, with negative effects consistently larger than those of natural gas prices by a mean ratio of 3.4. The strength of wind's price-suppressing causal effect intensifies in cooler peak periods while weakening under warm peak periods despite continued capacity additions and stable curtailment rates, implicating insufficient grid and transmission expansion to match spatially heterogeneous demand growth.

Figure 3: Causal effects of renewable generation on system lambda and natural gas generation, by market regime and rolling time windows.

Lagged effects at one-day intervals suggest unit commitment constraints, where high wind output decommits gas plants not quickly restartable. Wind outside West Texas and solar generation do not demonstrate direct causal effects on $\lambda$ within prevailing market regimes, though non-West wind reduces gas generation in cooler off-peak periods.

Regional Price Differentials

Wind generation reduces local price differentials but redistributes congestion costs seasonally. West Texas wind persistently lowers $\Delta_{\text{West}}$, but during warm peak periods elevates $\Delta_{\text{North}}$—indicative of transmission bottlenecks limiting west-to-north power flows. In cool peak periods, spillovers propagate to Houston, requiring reliance on costlier local generation.

Figure 4: Causal relationships between forecast renewables and regional price differentials, revealing spatial redistribution of congestion costs across ERCOT hubs.

Results: Evolving Demand and Gas Price Effects

Regional Demand

North Texas forecast load is the most consistent demand-side causal driver, especially during warm peak hours. South Texas load becomes increasingly relevant in cool periods post-2021, driving both gas generation and system lambda, aligning with transmission-constrained flows limiting wind exports. West Texas load demonstrates non-monotonic, regime-specific influence, paralleling regional electrification and data center growth.

Houston hub’s $\Delta_{\text{Houston}}$ propagates congestion cost spillovers to North, South, and West hubs, especially in cooler periods, reinforcing its centrality in locational price signal transmission.

Figure 5: Causal effects of regional electricity load on system lambda and natural gas generation for each ERCOT region and market regime.

Figure 6: Causal relationships between regional price differentials and system lambda, showing congestion cost propagation across trading hubs.

Natural Gas Markets

Natural gas prices retain causal relevance for system lambda but with intermittent and modest effects relative to wind. The dominant hub shifts from Waha (2019–2021) to Henry Hub (post-2021), reflecting pipeline expansion and integration of Texas markets to national gas dynamics.

Gas prices exhibit minimal direct causal influence on gas-fired generation, and weak links to hub price differentials, reinforcing that congestion costs are principally driven by spatial mismatches between renewables and load rather than fuel prices.

Figure 7: Causal effects of natural gas hub prices on system lambda and natural gas generation, demonstrating the temporal shift in hub relevance.

Implications and Future Directions

The findings challenge the conventional characterization of ERCOT as gas-price-driven. Wind now functions as the primary causal driver of wholesale prices, fundamentally increasing market exposure to meteorological risk. Temporal attenuation of wind’s price-suppressing effect in stressed periods raises concerns over grid resilience and underscores the necessity for targeted transmission expansion, storage deployment, and policy alignment with spatial demand growth.

The structurally shifting causal regime—including the emergence of Henry Hub as the driver of electricity prices and South Texas load’s growing impact—demands the adoption of adaptive, causally robust forecasting and planning models. For traders and system operators, these causal maps inform risk management and operational strategies, while for planners, they flag changing regional bottlenecks and transmission needs.

Solar generation’s null result is a temporal artifact; as penetration increases, its causal role will warrant finer-resolution analysis. Future directions include causal studies of battery storage, large new loads, and evolving market rules, extending the temporal causal modeling framework in anticipation of continued structural change.

Conclusion

This paper delivers a rigorous causal mapping of price formation in Texas’s transitioning electricity market, demonstrating that wind generation’s causal impact now supersedes that of natural gas prices, with congestion costs redistributed seasonally and the causal structure dynamically evolving as market fundamentals shift. The work establishes the necessity of temporally explicit causal models for robust forecasting, planning, and market design, laying foundations for future research as decarbonization and electrification accelerate.