Sectoral energy consumption by source and output in the U.S.: New evidence from wavelet-based approach

Abstract

This paper investigates the dynamics of the lead-lag relationships between aggregate and sectoral energy consumption by source and output in the U.S. economy within a time-frequency framework. To do that, we implement three variants of the continuous wavelet methodology, namely the wavelet power spectrum, the cross wavelet, and the wavelet coherence, to aggregate and disaggregate quarterly data between 2005Q1 and 2015Q3. The wavelet analysis unveils that the relationship between the U.S. output and the aggregate energy use by source varies across frequencies and evolves over time. Our results specifically show that the consumption of all energy sources display significant interactive linkages with the U.S. output. At the disaggregate level, while we perceive the presence of co-movements and causality between pairs through frequency bands and over time in almost all sectors, the industrial sector exhibits the highest intensity of wavelet coherence with the added value. Our findings highlight the need of accounting for the various energy sources and economic sectors when studying the output-energy nexus and designing economic policies.