A Social Logic of Energy: A data science approach to understanding and modelling energy transitions of India’s urban poor
Continued use of traditional solid biomass fuels for cooking in Indian households poses a serious public health risk. Particulate emissions in the form of soot contributed to approximately 600,000 deaths in 2019, a burden that falls disproportionately on women, children, and vulnerable populations. Despite over 95% of the population having access to clean cooking fuel distribution, following recent government initiatives to promote liquefied petroleum gas, biomass cooking fuel use is still widespread. This is the case even in cities, where low-income households have low levels of sustained clean cooking fuel use.
Interventions to promote transition to clean cooking often focus on cost and technology, informed by an economic-technical view of energy transition, but not all households benefit as expected from these interventions. Previous studies on socio-economic determinants of transition offer limited insight into the reasons for why some households can slip through the net of such interventions. The explanation lies in the socio-cultural and economic heterogeneity across households and the inherent spatial inequalities in urban India.
This thesis explores the influence of local socio-economic and cultural factors, and household practices and habits, on clean cooking transition with a view to understanding how the associated heterogeneity can be characterised, and integrated into quantitative energy models and methods. Public national survey and census data is supplemented with primary data collection, which provides valuable quantitative and qualitative data on low-income urban households.
Tree-based regression is used to investigate the influence of socio-economic and cultural factors within quantitative models. Determinants are found to exhibit non-linear trends, with thresholds for change in influence on transition. A statistical clustering reveals different typologies of household amongst clean cooking adopters, indicative of different enabling circumstances and pathways to transition. Continued use of biomass is found to be common across recently transitioned households.
The heterogeneity amongst low-income households, and the emergent transition pathways, are further investigated through data collected on low-income households in Bangalore. A novel method is used which combines mixed data in a two-stage clustering analysis, offering a means to characterise heterogeneity across households, identifying distinct transition pathways and associated barriers. The findings illustrate how wider socio-economic inequality is intertwined with access to sustained clean cooking.
A Bayesian multilevel microsimulation approach is proposed to model the spatial heterogeneity in clean cooking at a city scale. This approach combines publicly available data to generate a synthetic population, and estimates cooking fuel use and fuel stacking using a Bayesian multilevel model. The model takes into account household cooking practices, local spatial effects, and city level economic and policy context. The model reveals how low uptake of clean cooking fuel, and continued biomass use, is related to underlying spatial socio-economic inequalities in cities.