How good are weather shocks for identifying energy elasticities? A LASSO-IV approach to European natural gas demand

Abstract

We estimate the price elasticity of residential natural gas demand for 23 European Union (EU) countries using monthly data from 2011 to 2022. While neighboring countries’ weather shocks can in theory act as a supply shifter to identify demand, it is unclear which subset of neighbors in practice should be used, if any. To address this issue, we compare four traditional instrumental variables (IV) models to several post-LASSO approaches: post-LASSO Ordinary Least Squares (OLS), post-LASSO IV, and two-stage post-LASSO IV. We compare these models on a country-by-country basis and for the full panel. We find that the third traditional IV model that we examined performs best in most cases for individual countries. In addition, we find that the first traditional IV model has the most reliable results at the panel-level. Our preferred estimates suggest that country-level price elasticities range from $-0.61$ to $-0.01$, with a median of $-0.13$, in line with estimates from the previous literature. We find that residential natural gas price elasticities vary widely across Europe, with Hungary, Germany, France, and Lithuania being the most elastic, and Estonia and Portugal the least. However, Gross Domestic Product (GDP) trends during the energy crisis do not align perfectly with elasticity patterns, highlighting the need for caution in linking demand elasticities directly to economic outcomes. Broader macroeconomic factors also play a significant role in shaping national responses to the crisis.