Modeling rural Vietnamese households' use of cooking fuels
A majority of rural households in the developing world still use solid biomass fuels for cooking, which has severe negative health effects, may be expensive, time consuming, and contributes to global warming. Options for interventions aimed at improving the energy situation for households include the dissemination of improved cook stoves (ICSs), increase of households possibilities to switch to more modern fuels, for example through subsidies for LPG, as well as increase of the rate of electrification. However, few stove programs have been successful and the rate of fuel switching in rural areas remains low.
The aim of this thesis is to add to the existing literature of rural household’s fuel usage and furthermore to examine implications of fuel use variations on an ICS program. This is done through two papers concerned with current fuel use and one paper which present a modeling of stove interventions in six different villages.
Methods used to model current fuel usage are regression and a machine learning algorithm called Random Forest. Furthermore, a simplistic household model is developed and used for examining variations in effects from a possible future ICS dissemination in different areas. Both primary data from a survey carried out in the Vĩnh Phúc province of northern Vietnam and as well as secondary data from another survey in Vietnam are used.
Results of the regressions include that current fuel usage can to a high degree be modeled by the distance to town or household density together with the household income. The outcome of the Random Forest method confirms these results but also emphasizes a high dependence on a stable household economy over a longer period of time in order for fuel switching to occur. Results from the modeling of stove interventions revealed large differences between the respective villages and possible non-linear relationship between stove efficiency and benefits.
The possibility of modeling household usage by household density or distances to urban centers, casts doubt on a casual interpretation of, in the literature, previously described relations between various development factors and fuel switching, but also towards a possibility of predicting policy effects in different regions without detailed household surveys. Combining these results with a further development of the household modeling and field measurements may also provide possibilities to predict the effects of ICSs dissemination programs.