Billing Education and Electricity Consumption under Increasing Block Tariffs: Experimental and Machine-Learning Evidence

Abstract

Many electricity utilities in developing countries use increasing block tariffs (IBTs) to pursue affordability, conservation, and cost recovery. However, limited consumer knowledge of marginal prices and how bills are calculated under an IBT can undermine policy goals and lead to suboptimal consumption decisions. We conduct a randomized field experiment in Addis Ababa, Ethiopia, to test whether billing education and price information affect residential electricity use under an IBT in a postpaid billing system. Linking household survey data to 48 months of administrative billing data from the state-owned utility, we find that the intervention led to a modest increase in electricity consumption over approximately two and a half years in a setting where electricity prices are low and heavily subsidized. Machine-learning models trained on pretreatment billing data and baseline household characteristics yield estimates that closely match the experimental estimates, even when using monthly rather than high-frequency smart-meter data. Estimates from a causal forest further suggest larger treatment effects among households with higher baseline consumption, greater appliance ownership, prior awareness of the tariff structure, and reported checking of bill details. The findings indicate that light-touch billing education and price information can influence electricity use even with delayed consumption feedback and billing interval uncertainty.