Ghana - Regulatory Strengthening

The evaluation will use a variety of data source for the regulatory project performance evaluation, including a combination of implementation and project records, administrative data, external surveys, and KIIs.
We will conduct a longitudinal household and enterprise survey that covers the Greater Accra area as well as the next largest 7 cities in the ECG catchment area. We will work closely with GridWatch to ensure that our sample targets locations where they are collecting outage data, when possible, and so that we can avoid surveying the same households and enterprises covered in their baseline survey and so that we can take full advantage of the data they are collecting on outages and voltage fluctuations. We will work with Ghana Statistical Service (GSS) and the World Bank to develop two sampling frames-one for households and one for businesses. We will use a multi-stage sampling plan. First we will sample enumeration areas. Second, we will sample electric poles within enumeration areas. Third, we will sample households and enterprises served by those poles. We will oversample larger enterprises to improve the precision of our data for addressing the Economic Rate of Return calculations. We will conduct Key Informant Interviews and Focus Group Discussion to help enrich our understanding of key issues. The evaluation will have three rounds of data collection, baseline in 2019, a qualitative midline in 2021 near the end of the compact, and an endline starting in 2023.

We plan to sample 2,394 enterprises in the Greater Accra area and estimate a minimum detectable effect (MDE) of 0.12 standard deviations for enterprises in that area. The minimal detectable differences (MDD) are about half as large as the MDEs for binary outcomes such as, whether the entity has a legal connection. This outcome had a mean of around 0.5 in the baseline data for the Access project. This means that we should be able to detect changes in the fraction of enterprises with a legal connection as small as 0.06 between the baseline and follow-up surveys. Given MCC's greater interest in enterprises compared to households we are aiming for a much larger MDE for households, at around 0.21 in the Greater Accra area. We are expecting to have slightly larger MDEs, of about 0.24 standard deviations, when considering samples of enterprises from smaller geographic areas-in particular when looking only at the geographic areas targeted by one of the following interventions: new substations, bulk supply points, or line bifurcation. We will be able to address the possibility of spillover within the Greater Accra area somewhat more precisely as we expect an MDE of 0.15 for enterprises there. We will have just a bit less precision when looking at enterprises in the 7 largest cities in the ECG catchment area outside of Greater Accra, where we expect an MDE of around 0.17. This will enable us to say something about the degree to which PDS has impacts in areas where the MCC infrastructure is less likely to have made a difference. As noted above, we plan to coordinate with GridWatch so that we can obtain outage and voltage fluctuation data based on the PowerWatch devices for our survey samples in the Greater Accra area. We expect that this will require approximately 798 PowerWatch devices for the enterprise surveys and another 150 for the household surveys for a total of 948 PowerWatch devices. This is based on an assumption that we will have one transformer per EA and three PowerWatch devices per transformer. In doing these calculations we have assumed no overlap between the transformers used for the household and enterprise surveys. In reality, we expect that there will be some overlap which may enable us to add additional EAs and thereby further reduce our MDIs; however, if the number of devices available are near the estimated required number of 948, we recommend deploying them all appropriately between households and enterprises to help improve our ability to capture outages and voltage fluctuations. See Table X.2. in the EDR for more details.


Not applicable for evaluation design report.


Not applicable for evaluation design report.


Not applicable for evaluation design report.

Not applicable for evaluation design report.