Update: This blog was updated on 10/16/2014 from the original version.

What is the best way to promote access to reliable and affordable electricity for the estimated 600 million Africans that currently live without it?  Should efforts focus on extending centralized grids? Should governments emphasize off-grid solutions? Or, is the magical formula a mix of the two? These are big questions for African policymakers and their global partners. It’s also at the heart of a raging debate concerning President Obama’s Power Africa initiative. Some have categorically argued that focusing on grids is the wrong way to go. But what if tens of millions of people actually live in areas where the grid already exists, yet aren’t connected to it for some reason?

Our forthcoming CGD research on infrastructure service availability, which utilizes new Afrobarometer survey data, may shed some light on this debate. Whenever Afrobarometer enumerators survey individuals in the field, they track whether there is an “electricity grid that most houses could access” in the immediate area. (We’ve included more detail on their sampling methodology below.) In their last survey round, they collected this information for 33 countries across North Africa and Sub-Saharan Africa. This includes five of the Power Africa countries (Ghana, Kenya, Liberia, Nigeria, and Tanzania). Unfortunately, they don’t ask whether the survey respondent actually has access to electricity as well. As such, we have to rely on other data sources, such as Demographic and Health Surveys, for national and sub-national access estimates. 

Comparing grid coverage and household access rates, even in an apples-to-oranges way, may inform decisions about what types of policies to promote. For instance, should policymakers tackle regulatory or commercial challenges that hold back new electricity connections, such as high hook up costs or unsustainable tariff structures? Or should they expand household access through grid extension and/or off-grid solutions for under-served populations?

Here are some of our preliminary findings for the Power Africa countries: 

Shedding New Light on the Off-Grid Debate in Power Africa Countries
[Tweet this]

Shedding New Light on the Off-Grid Debate in Power Africa Countries
[Tweet this]

  1. Electrical grid coverage is surprisingly high in several countries. For instance, over 90 percent of surveyed Nigerians live in an area where the grid is present. In Ghana, that figure is over 80 percent. And it’s roughly 70 percent in Kenya. If extrapolated to the broader population, up to 230 million people could live near the electrical grid in these five Power Africa countries (out of roughly 300 million people total).
  2. Yet household access rates remain very low. According to recent DHS data, only 56 percent of Nigerian households have access to electricity. It’s a little higher in Ghana (61 percent), much lower in Kenya (23 percent), and even lower yet in Liberia and Tanzania. Collectively, an estimated 165 million people live without electricity in these five countries.
  3. This dynamic applies to both urban and rural populations alike. While grid coverage rates are lower in rural areas, they are still quite high in several countries. For example, nearly 90 percent of surveyed rural Nigerians live in areas where the grid is present. Yet, DHS data puts rural household access at a mere 34 percent. Liberia and Tanzania, which have both extremely low grid coverage and household access rates in rural areas, are the two exceptions.
  4. Focusing on grid connections might be lower hanging fruit than pursuing capital-intensive investments in grid extension or off-grid solutions. It’s not perfectly clean to compare DHS and Afrobarometer survey data on household access and electrical grid coverage. Even so, the huge differential between the two suggests that hooking up households to the grid in their immediate area could put a big dent into African energy poverty. Based on our back-of-the-envelope estimates, there may be up to 95 million people living ‘under the grid’ and without access in five of the Power Africa focus countries. A great paper on Kenya by researchers at Berkeley comes to a similar conclusion.

Shedding New Light on the Off-Grid Debate in Power Africa Countries
[Tweet this]

None of this suggests that grid extension or off-grid approaches should be ignored. But, it does argue that any effort to combat African energy poverty should look hard at connecting people to the grid that is already in their proverbial backyard. And if our rough estimates are even remotely accurate, then the much argued need for leapfrogging over traditional grid structures may reflect ideological ambitions more than real world opportunities for promoting access at a grand scale. 

**Methodological Note**

Afrobarometer survey samples are designed to produce a representative cross-section of all voting age citizens within a given country. The sampling frame attempts to ensure that every adult citizen has an equal and known chance of being selected for an in-person interview. Afrobarometer samples typically include either 1200 or 2400 cases. A randomly selected sample of 1200 interviews allows national adult population inferences with a margin of sampling error of +/- 2.8 percent at the 95 percent confidence level. With a sample size of 2400, the margin of error is +/- 2.0 percent at the 95 percent confidence level.

Afrobarometer stratifies the sample by the main sub-national unit of government (e.g., state, province, or region) and by urban or rural location. This reduces the likelihood that distinctive ethnic or language groups are omitted from the sample. Afrobarometer occasionally oversamples certain politically significant populations within a country to ensure that the size of the sub-sample is large enough for rigorous analysis. Data sets include weighting factors at the primary sampling unit (PSU) level to account for individual selection probability. These sampling units typically correspond to national census units.

Afrobarometer enumerators identify the availability of five types of infrastructure in the respondents’ enumeration area: electricity grid, piped water, sewage system, mobile phone service, and surfaced roads. Afrobarometer protocols require that both enumerators and field supervisors jointly assess the presence of infrastructure services in the enumeration areas.

Afrobarometer data lack specificity on the electrical grid available in an area. We are unsure whether positively identified areas have high voltage lines and step down transformers necessary to facilitate household connections. If areas are lacking step down transformers, the cost of connecting houses would significantly increase.  This is an issue that we plan to explore further with Afrobarometer project staff.     

Demographic and Health Surveys also produce nationally and sub-nationally representative results based upon statistically rigorous sampling frames. When determining household electricity access, DHS enumerators rely upon individual responses and then apply those to all documented members of the household. In contrast, Afrobarometer only collects information on the primary survey respondent. It does not collect information on other members within the related household. However, since both Afrobarometer and DHS data is nationally representative, including for urban and rural cross-tabulations, we are broadly comfortable making rough comparisons across the two datasets. Nonetheless, there are inter-temporal differences in the survey coverage that merits consideration when interpreting our back-of-the-envelope results. Importantly, we do not attempt to extrapolate mid-point estimates of either grid coverage or household access rates in between survey years.