Modeling and Forecasting Electricity Loads and Prices

RAFAL WERON received his M.Sc. (1995) and Ph.D. (1999) degrees in applied mathematics from the Wroclaw University of Technology (WUT), Poland. He currently holds a position of Assistant Professor at WUT. His research focuses on risk management and forecasting in the power markets and computational statistics as applied to finance and insurance.Rafal Weron is the co-author of three books and over 70 research articles, book chapters, and conference papers. His professional experience includes design of the risk management system for BOT Holding (BOT Górnictwo i Energetyka S.A.), development of insurance strategies for Polish Power Grid Co. (PSE S.A.) and Hydro-storage Power Plants Co. (ESP S.A.), as well as implementation of yield curve calibration and option pricing software for LUKAS Bank S.A. (Crédit Agricole Group). He has also been a consultant or executive teacher to a large number of banks and corporations.

• Preface ixAcknowledgments xiii1 Complex Electricity Markets 11.1 Liberalization 11.2 The Marketplace 31.2.1 Power Pools and Power Exchanges 31.2.2 Nodal and Zonal Pricing 61.2.3 Market Structure 71.2.4 Traded Products 71.3 Europe 91.3.1 The England and Wales Electricity Market 91.3.2 The Nordic Market 111.3.3 Price Setting at Nord Pool 111.3.4 Continental Europe 131.4 North America 181.4.1 PJM Interconnection 191.4.2 California and the Electricity Crisis 201.4.3 Alberta and Ontario 211.5 Australia and New Zealand 221.6 Summary 231.7 Further Reading 232 Stylized Facts of Electricity Loads and Prices 252.1 Introduction 252.2 Price Spikes 252.2.1 Case Study: The June 1998 Cinergy Price Spike 282.2.2 When Supply Meets Demand 292.2.3 What is Causing the Spikes? 322.2.4 The Definition 322.3 Seasonality 322.3.1 Measuring Serial Correlation 362.3.2 Spectral Analysis and the Periodogram 392.3.3 Case Study: Seasonal Behavior of Electricity Prices and Loads 402.4 Seasonal Decomposition 412.4.1 Differencing 422.4.2 Mean or Median Week 442.4.3 Moving Average Technique 442.4.4 Annual Seasonality and Spectral Decomposition 442.4.5 Rolling Volatility Technique 452.4.6 Case Study: Rolling Volatility in Practice 462.4.7 Wavelet Decomposition 472.4.8 Case Study: Wavelet Filtering of Nord Pool Hourly System Prices 492.5 Mean Reversion 492.5.1 R/S Analysis 502.5.2 Detrended Fluctuation Analysis 522.5.3 Periodogram Regression 532.5.4 Average Wavelet Coefficient 532.5.5 Case Study: Anti-persistence of Electricity Prices 542.6 Distributions of Electricity Prices 562.6.1 Stable Distributions 562.6.2 Hyperbolic Distributions 582.6.3 Case Study: Distribution of EEX Spot Prices 592.6.4 Further Empirical Evidence and Possible Applications 622.7 Summary 642.8 Further Reading 643 Modeling and Forecasting Electricity Loads 673.1 Introduction 673.2 Factors Affecting Load Patterns 693.2.1 Case Study: Dealing with Missing Values and Outliers 693.2.2 Time Factors 713.2.3 Weather Conditions 713.2.4 Case Study: California Weather vs Load 723.2.5 Other Factors 743.3 Overview of Artificial Intelligence-Based Methods 753.4 Statistical Methods 783.4.1 Similar-Day Method 793.4.2 Exponential Smoothing 793.4.3 Regression Methods 813.4.4 Autoregressive Model 823.4.5 Autoregressive Moving Average Model 833.4.6 ARMA Model Identification 843.4.7 Case Study: Modeling Daily Loads in California 863.4.8 Autoregressive Integrated Moving Average Model 953.4.9 Time Series Models with Exogenous Variables 973.4.10 Case Study: Modeling Daily Loads in California with Exogenous Variables 983.5 Summary 1003.6 Further Reading 1004 Modeling and Forecasting Electricity Prices 1014.1 Introduction 1014.2 Overview of Modeling Approaches 1024.3 Statistical Methods and Price Forecasting 1064.3.1 Exogenous Factors 1064.3.2 Spike Preprocessing 1074.3.3 How to Assess the Quality of Price Forecasts 1074.3.4 ARMA-type Models 1094.3.5 Time Series Models with Exogenous Variables 1114.3.6 Autoregressive GARCH Models 1134.3.7 Case Study: Forecasting Hourly CalPX Spot Prices with Linear Models 1144.3.8 Case Study: Is Spike Preprocessing Advantageous? 1254.3.9 Regime-Switching Models 1274.3.10 Calibration of Regime-Switching Models 1324.3.11 Case Study: Forecasting Hourly CalPX Spot Prices with Regime-Switching Models 1324.3.12 Interval Forecasts 1364.4 Quantitative Models and Derivatives Valuation 1364.4.1 Jump-Diffusion Models 1374.4.2 Calibration of Jump-Diffusion Models 1394.4.3 Case Study: A Mean-Reverting Jump-Diffusion Model for Nord Pool Spot Prices 1404.4.4 Hybrid Models 1434.4.5 Case Study: Regime-Switching Models for Nord Pool Spot Prices 1444.4.6 Hedging and the Use of Derivatives 1474.4.7 Derivatives Pricing and the Market Price of Risk 1484.4.8 Case Study: Asian-Style Electricity Options 1504.5 Summary 1534.6 Further Reading 154Bibliography 157Subject Index 171