Analyzing Renewable Integration and Market Design: Distributional and Dynamic Perspectives in Electricity Markets
研究了2021至2023年加州和PJM电力市场在可再生能源渗透率差异下的价格形成机制，发现加州市场对负荷和温度变化更敏感、均值回复更快，而PJM更稳定，为市场运营者提供可再生能源整合的针对性建议。
This study investigates price formation mechanisms in electricity markets with contrasting renewable penetration levels by examining the California Independent System Operator (CAISO) and Pennsylvania-New Jersey-Maryland Interconnection (PJM) markets from 2021 to 2023. We employ a novel methodological approach to address a critical gap: while existing literature examines individual markets, the joint distributional and dynamic aspects of price formation, particularly during extreme events, remain poorly understood. Our innovative combination of quantile regression and modified Ornstein-Uhlenbeck process captures both how market drivers influence electricity prices across different market conditions and their subsequent adjustment patterns. Our findings reveal distinct patterns in how these markets respond to load variations, renewable generation, and weather sensitivity. CAISO demonstrates consistently stronger price elasticities to load changes and temperature fluctuations. It also shows faster mean reversion rates, reflecting a market design optimized for renewable integration but with higher volatility. PJM exhibits more moderate responses and slower mean reversion, indicating greater price stability through resource diversity. During extreme events, these differences intensify significantly – CAISO exhibits enhanced price-suppressing effects from renewables during off-peak periods and intensified gas generation impacts during stress periods, while PJM maintains more stable price dynamics. These results suggest that successful renewable integration depends heavily on market structure, requiring regionally tailored approaches that balance short-term flexibility with long-term stability. The findings provide actionable insights for market operators and participants navigating the transition to higher renewable penetration. The robustness of our results is established through bootstrapping, threshold sensitivity analysis, and residual diagnostics to ensure reliability.