A microsimulation of spatial inequality in energy access: A Bayesian multi-level modelling approach for urban India

Abstract

Access to sustained clean cooking in India is essential to addressing the health burden of indoor air pollution from biomass fuels, but spatial inequality in cities can adversely affect uptake and effectiveness of policies amongst low-income households. Limited data exists on the spatial distribution of energy use in Indian cities, particularly amongst low-income households, and most quantitative studies focus primarily on the effect of economic determinants. A microsimulation approach is proposed, using publicly available data and a Bayesian multi-level model to account for effects of current cooking practices (at a household scale), local socio-cultural context, and spatial effects (at a city ward scale). This approach offers previously unavailable insight into the spatial distribution of fuel use and residential energy transition within Indian cities. Uncertainty arising from heterogeneity in the population is factored into fuel use estimates through use of Markov Chain Monte Carlo (MCMC) sampling. The model is applied to four cities in the south Indian states of Kerala and Tamil Nadu, and comparison against ward-level survey data shows consistency with the model estimates. Ward-level effects exemplify how wards compare to the city average and to other urban area in the state, which can help stakeholders design and implement clean cooking interventions tailored to the needs of households.