A0458
Title: Functional modeling of electricity returns with dynamic volatility
Authors: Eric Costa Andreu - Universidad de Alicante (Spain) [presenting]
Angeles Carnero - Universidad de Alicante (Spain)
Pedro Galeano - Universidad Carlos III de Madrid (Spain)
Abstract: A functional time series model is used to capture the dynamics of daily return curves of electricity prices. While electricity prices are typically recorded hourly as discrete values, representing them as continuous functions could better capture the underlying structure of price fluctuations. Using data from the Spanish electricity market, functional representations of daily returns are constructed, their temporal stability is assessed, and functional principal component analysis (FPCA) is applied. Components accounting for a large amount of the total variability are retained, and their scores are individually modeled using a seasonal ARMA-GJR-GARCH process estimated through maximum likelihood. Predicted scores enable the reconstruction of the return curves. Moreover, the proposed functional model facilitates the estimation and forecasting of return volatility, providing a natural measure of market risk. Lastly, the statistical properties of the model's parameter estimators are examined by analyzing their empirical distributions using both parametric and residual-based bootstrap techniques.
