[Paper Review] Changes in electricity demand pattern in Europe due to COVID-19 shutdowns

Michał Narajewski, Florian Ziel|arXiv (Cornell University)|Apr 29, 2020

Energy Load and Power Forecasting4 references40 citations

TL;DR

The paper analyzes how COVID-19 shutdowns shifted electricity demand patterns in five EU countries using high-dimensional change-point models to separate shutdown effects from seasonal and weather factors.

ABSTRACT

The article covers electricity demand shift effects due to COVID-19 shutdowns in various European countries. We utilize high-dimensional regression techniques to exploit the structural breaks in demand profiles due to the shutdowns. We discuss the findings with respect to coronavirus pandemic progress and regulatory measures of the considered countries.

Motivation & Objective

• Investigate how COVID-19 shutdowns changed electricity demand patterns in selected European countries.
• Distinguish shutdown-induced changes from seasonal, calendar, and weather effects.
• Quantify the impact of different lockdown phases on hourly and weekly demand profiles.

Proposed method

• Use high-dimensional time series change-point models to detect structural breaks in electricity load.
• Augment a baseline load model with permanent changes in level for daily and weekly profiles via change-point dummies.
• Estimate via lasso regularization tuned by BIC and simulate 10,000 trajectories to compare shutdown vs. no-shutdown scenarios.

Experimental results

Research questions

• RQ1 What is the impact of COVID-19 shutdowns on the level and pattern of electricity demand in Italy and other European countries?
• RQ2 How do daily and weekly demand profiles shift during lockdowns, beyond seasonal and weather effects?
• RQ3 Do the timing and intensity of lockdown measures correspond to observable structural breaks in load data across countries?

Key findings

• Shutdowns significantly altered both the level and weekly pattern of electricity demand across Italy and other European countries.
• The morning peak is flattened while the evening peak remains, with a reduced difference between weekends and weekdays.
• A late-morning shift and “getting up late” pattern are observed in several countries, reflecting changes in daily routines.
• France shows an early-onset impact possibly due to announcements and anticipatory behavior before full lockdowns.