Temperature Sensitivity of Residential Energy Demand on the Global Scale: A Bayesian Partial Pooling Model

TL;DR

This study uses a Bayesian Partial Pooling model to analyze how residential energy demand varies with temperature across 126 countries from 1978 to 2023, revealing non-linear sensitivities and differences between developed and developing nations.

Contribution

It introduces a Bayesian Partial Pooling approach to estimate country-specific temperature sensitivities of residential energy demand on a global scale.

Findings

Higher demand at temperatures below -5°C and above 30°C.

Steepening of demand-temperature relationship above 23.5°C.

Greater temperature sensitivity in developed countries.

Abstract

This paper contributes to the limited literature on the temperature sensitivity of residential energy demand on a global scale. Using a Bayesian Partial Pooling model, we estimate country-specific intercepts and slopes, focusing on non-linear temperature response functions. The results, based on data for up to 126 countries spanning from 1978 to 2023, indicate a higher demand for residential electricity and natural gas at temperatures below -5 degrees Celsius and a higher demand for electricity at temperatures above 30 degrees Celsius. For temperatures above 23.5 degrees Celsius, the relationship between power demand and temperature steepens. Demand in developed countries is more sensitive to high temperatures than in less developed countries, possibly due to an inability to meet cooling demands in the latter.