Household electricity demand and structural drivers: A two-stage least squares instrumental variable approach

This dataset was prepared to test the hypothesis that modern energy adoption influences household electricity consumption in Uganda, but that this relationship is shaped by structural conditions such as electricity reliability, housing quality, household size, urban/rural residence, regional location, and fuel market exposure.

The data were derived from the Uganda National Household Survey 2019/2020 conducted by the Uganda Bureau of Statistics. The dataset contains cleaned household-level variables on electricity consumption, electricity expenditure, grid access, daily hours of power availability, household demographics, housing materials, regional location, urban/rural status, asset ownership, cooking fuels, stove types, and fuel sourcing behaviour. Several derived variables were created, including monthly electricity consumption in kilowatt-hours, a Modern Energy Adoption Index based on efficient lighting, efficient cooking and electric heating indicators, a Housing Quality Index based on roof, wall and floor materials, and an asset ownership index. Survey weights are retained to support nationally representative analysis.

The data show that household electricity consumption in Uganda is highly right-skewed, with most households consuming relatively low quantities of electricity and only a small proportion recording high monthly consumption. Modern energy adoption also varies strongly across regions and settlement types, with urban households generally showing higher electricity use and stronger adoption patterns than rural households.

The dataset supports descriptive analysis, principal component analysis, ordinary least squares regression, log-linear models, Gamma-log generalized linear models, quantile regression, instrumental-variable estimation, two-stage least squares, limited-information maximum likelihood, generalized method of moments, and urban–rural heterogeneity analysis. The results indicate that electricity reliability, urban residence, household size, and regional location are important structural determinants of electricity consumption. Modern energy adoption is strongly associated with dwelling characteristics, stove type, cooking fuel, and market exposure, but its causal effect on electricity consumption becomes weaker once endogeneity is addressed.

The dataset can be used by researchers studying household electricity demand, energy poverty, modern energy adoption, electrification policy, and structural constraints in low-income energy systems. It should be interpreted as household survey-based evidence rather than metered engineering consumption data. Survey weights should be applied when making population-level inferences.