Efficient energy usage typologies: Adoption and determinants of energy-saving appliances among Ugandan households

Abstract

This study investigates household adoption of energy-efficient technologies in Uganda using nationally representative data from the Uganda National Household Survey 2020/21. We develop an integrated econometric framework that combines Latent Class Analysis (LCA), Multinomial Logit (MNL) models, and Logistic and Poisson regressions to capture adoption heterogeneity, identify structural determinants, and differentiate between adoption incidence and intensity. LCA probabilistically classifies households into four adoption typologies: nonadopters, lighting-only adopters, heating-only adopters, and broad adopters. MNL models quantify the demographic, geographic, and infrastructural factors that shape class membership, whereas logistic and Poisson specifications reveal who adopts at least one technology and how deeply it extends across multiple domains. The results showed that over 90 % of the households remained non-adopters, with efficient lighting as the primary entry point, minimal cooking, and virtually no heating. Adoption is consistently constrained by electricity access, affordability, and household asset levels, whereas urban residences, education, and younger household heads significantly increase the likelihood of adoption. By integrating latent adoption typologies with incidence–intensity modeling using nationally representative household data, this study advances beyond conventional binary adoption analyses and provides robust, policy-relevant evidence on the structural constraints shaping Uganda’s household energy transition.