Evaluations of distribution infrastructure investments may need to be focused on a wider geographic scale to measure and detect reliability impacts. Although the evaluation design contained valid treatment and control groups within the districts targeted for line bifurcation investments, these interventions comprised only a portion of the overall suite of infrastructure upgrades completed under the Ghana Power Compact. While the sampled districts received transformer injections into the low-voltage network, the broader Electricity Company of Ghana Financial and Operational Turnaround Project also supported sizable investments into the medium- and high-voltage network, including construction of new primary substations and two major bulk supply points at the interface to the transmission system leading into Greater Accra. Moreover, the evaluation showed that a significant share of outages tended to occur at the medium- and high-voltage levels within the Accra grid, typically exceeding the extent of outages observed at the low-voltage level, which was the focus for detecting effects via the collection of both survey and GridWatch data for this evaluation. This suggests that to increase the likelihood of detecting impacts on reliability resulting from power infrastructure projects, future evaluations should consider a measurement strategy that captures the totality of MCC investments within the relevant geographic area, treating the low-voltage, medium-voltage, and high-voltage segments of the local grid holistically as a ‘supply chain’ where aggregate outages may arise from faults along multiple sections of the network. Correspondingly, any improvements in reliability achieved by MCC investments would reflect the cumulative new, upgraded, or rehabilitated constituents of the power grid resulting from the program, including all new transformers, substations, and power lines. Evaluation of future power infrastructure projects should therefore explore a design approach that spans an appropriately inclusive geographic (and geometric) scope of the electric grid, in order to facilitate detection of outage improvements at a greater magnitude, while reflecting a fuller extent of outage reductions resulting from project investments. This may require greater reliance on interrupted time series designs in lieu of difference-in-difference or regression discontinuity designs, which are more suited to sampling at a more localized level. However, the high-frequency nature of outage monitoring technology such as that used for the line bifurcation evaluation would enable clearer strategies for identifying reliability improvements over extended time periods, and linking changes to the timing of these changes to the implementation timeline of larger infrastructure assets coming online.
Lesson Learned