How does market architecture affect price dynamics ? A time series analysis of the Italian day-ahead electricity prices

How do changes in the market architecture affect the dynamics of deregulated electricity prices? We investigate this issue in the context of the Italian Power Exchange (IPEX), using data on the daily average day-ahead price (PUN) between April 2004 and December 2008. Estimates of baseline time series models (ARMAX and ARMAX-EGARCH) and their forecasting performances suggest that the trend in natural gas prices, deterministic weekly patterns, the impact of perceived temperatures, persistence in conditional volatility and the inverse leverage effect are essential features of the PUN dynamics. We then augment the best-performing models with dummies that account for changes in the market architecture, such as the introduction of contracts for differences (CfDs) to support renewables, trading of white certificates for energy efficiency, and the demandside liberalization. The findings show that changes in the market architecture have only affected the PUN volatility. Specifically, CfDs have mitigated volatility, while white certificates and demand liberalization have increased it. Moreover, after controlling for reforms the inverse leverage effect vanishes, and the persistence in volatility is lower than in the baseline estimates.