This paper presents the performance analysis of mini-grids using DEA. It has used mini-grid data from a published source and determined the relative production efficiencies of mini-grids in developing countries across the world. Data for 81 minigrids was extracted, and two separate data envelopment analysis models were

used to analyse the dataset. The Charnes, Cooper & Rhodes (CCR) model offered an analysis under a constant returns-to-scale assumption, while the Banker, Charnes and Cooper (BCC) model offered an analysis with a variable returns-to-scale assumption. This allowed for the determination of the most-productive-scale-size for

the mini-grids analysed, by comparing the pure technical efficiency given by the BCC model with the technical and scale efficiency of the CCR model. Four inputs and one output variable were considered for this analysis, with the capacity per person served by the mini-grid, the renewable energy share of the mini-grid and the

age of the mini-grid being used alongside a code for the ownership model of the mini-grid as inputs, and the cost per watt delivered as the single output.

This analysis has shown that there are significant issues present with the productive efficiency of mini-grids in developing countries. Compared to their peers in the dataset, the majority of mini-grids are inefficient, with a large number of mini-grids being very inefficient compared to their peers. The input-oriented models in

particular highlight this inefficiency: input-oriented scale efficiency is low across the corpus, indicating that the majority of mini-grids are operating far away from their most productive scale size. Moving forward, the mini-grid sector in developing countries needs to assess the scale of their operations, and determine whether

increasing or decreasing returns to scale are present to expand or diversify their operations as appropriate.