A0271
Title: A local Gaussian correlation block bootstrap test and the NordLink cables effect on electricity prices
Authors: Sondre Hoelleland - Norwegian School of Economics (Norway)
Kristian Gundersen - University of Bergen (Norway)
Jan Bulla - University of Bergen (Norway)
Baard Stoeve - University of Bergen (Norway) [presenting]
Abstract: Financial returns often exhibit strong autocorrelation and complex nonlinear dependencies, making standard bootstrap methods unsuitable due to their assumption of independence. To address this, prior studies have used local Gaussian correlation (LGC) and GARCH filtering to capture asymmetric and nonlinear dependence structures. However, GARCH filtering can be cumbersome and may distort the data. A block bootstrap approach is proposed that preserves the nonlinear and time-varying dependence in financial data without requiring filtering. A robust test for financial contagion is also introduced, designed to work with this bootstrap method. The method is illustrated, using price data from the German and Norwegian electricity markets, focusing on whether dependence structures changed after the NordLink interconnector was established in May 2021, a context known for nonlinear and contagious price dynamics.
