
- Date
- 14th August 2026
- Categories
- eCooking
By Dr. Simon Batchelor OBE (Gamos Ltd. / Loughborough University).
Electricity access statistics tell us whether a household is connected. They tell us much less about what happens after connection: how much electricity households actually use, how that use is distributed across the population, and how electricity fits within the wider household energy economy.
We have a growing body of evidence, that eCooking stimulates use of electricity and therefore has the potential to improve the economics for the investors. COSMO shows this quite clearly in mini grids, and the modelling behind the M300 Burundi action by Weza Power and Article 6 Group illustrates it. (The BAKU Cop29 NDC talk on the plans for Burundi are available here (1hour) – but I am also preparing a blog for next month on it). They make assumptions on the consumption by households and how that might change if eCooking was introduced as part of the connection planning. After the ESMAP webinar on the Multi-Tier Framework the other day, I was reminded what a wealth of data it is. Dr Scott, T Jones and I analysed some of it some years ago, and so I wanted to see if the MTF data had some insights into consumption and spend….total household energy expenditure including cooking, which some or all could pivot into electricity consumption.
The starting point was the World Bank’s Multi-Tier Framework household survey data. While they have undertaken 29 MTF surveys (in 27 countries), many of the data files are not freely available (and I confess I didn’t spend a lot of time asking for them!). So I looked at 7 sets, and 3 had the granularity of data I was looking for (although data was collected back in 2018 and 2019). Nevertheless as a first step this looked promising – since I was looking for the relationship between electricity consumption and spend on cooking fuel, it was good enough.
Rather than using country averages for electricity consumption, I took the underlying household-level data for Kenya, Rwanda and Ethiopia and grouped connected households into monthly electricity-consumption bands: less than 5 kWh, 5–10 kWh, 10–20 kWh, 20–50 kWh, 50–100 kWh and more than 100 kWh.
A national average hides very different kinds of electricity customer. In Rwanda, around a quarter of the connected population represented in the data lived in households using less than 5 kWh a month, and just over half were below 10 kWh. Ethiopia also contained a substantial population of extremely low-consuming connected households. Kenya was shifted further up the consumption ladder, although there were still significant numbers at very low levels of monthly consumption.
Using the household identifiers in the MTF datasets I connected those electricity records with the same households’ cooking information.
What does the household energy economy look like if we put cooking back alongside electricity?
Matching electricity and cooking expenditure
The surveys record considerably more about cooking than is sometimes apparent from the headline MTF indicators. They identify fuels used for cooking and contain information on expenditure associated with the main cooking stove and its principal fuels. So for each electricity-consumption band, I calculated the associated cooking-fuel expenditure.
For comparability across the three countries, I converted the reported expenditures into approximate US-dollar values using exchange rates appropriate to the survey period. These are therefore historical, nominal comparisons rather than estimates of what households would spend today.
The output is in Figure 1 above.
Among Kenyan households using less than 5 kWh of electricity a month, the median electricity bill was around US$2 per month, while reported median cooking-fuel spending was close to US$7 per month.
In Rwanda, households consuming 5–10 kWh a month were spending roughly US$2–3 on electricity, but around US$9 on cooking fuel.
In Ethiopia the historical electricity tariff was exceptionally low! Households in the lowest consumption bands could be spending only a few cents or a fraction of a dollar on electricity while spending several dollars a month on cooking fuels.
The important point is not the precise ratio in any single country. It is that households which appear to the electricity system to be very small energy consumers may in fact be spending several times their electricity bill elsewhere in the household energy economy. We have always known this but the MTF lets us see it in the data. Cooking is simply outside the meter.
A different way of seeing a low-consumption customer
This matters because electricity-sector analysis often starts from the electricity bill.
A household consuming 5 kWh a month can look commercially unattractive. The connection has been built, the meter installed and the network extended, yet monthly consumption is extremely low. Often as a rule of thumb, a rural connection is costed at about $1000 per household by international agencies – and access expansion has only succeeded because of ‘blended finance’ with concessional finance. From a utility perspective, revenues are often disappointingly small…but as the world agreed in SDG7, equitable energy access is a worthy target.
However, a research hypothesis has always been – that household may at the same time be spending substantially more each month on charcoal, LPG, kerosene or purchased firewood. If cooking moves onto electricity, some or all of that existing household expenditure can move with it.
Importantly, that does not necessarily imply that total household energy expenditure rises. Electric cooking is generally much more energy efficient than combustion-based cooking. In a number of markets, households that make substantial use of efficient electric appliances find that cooking with electricity costs less than the fuels it replaces. So the opportunity is not simply to capture every dollar currently spent on cooking fuel but to enable the household to find savings in their spend on cooking fuels. (And of course it triggers all the other benefits we here at MECS talk endlessly about – the reduction of household air pollution and its consequent effects on health, the potential reduction of global emissions, time saves, gender equity, etc).
But this blog is about pivoting cooking spend into the utility whether that be national grid or minigrid. The important observation confirmed by the data is that there is already an energy expenditure stream in the household which is largely invisible to the electricity system.
A transition to eCooking can convert part of that expenditure into electricity demand and utility revenue, while potentially reducing the total cost of delivering the cooking service.
The scale of the low-use population
Across the three survey populations, more than 8 million people were living in connected households consuming less than 10 kWh of electricity a month, and more than 11 million were living in households below 20 kWh.
For these households, cooking would not be a marginal electrical load. Adding perhaps 20 or 30 kWh a month for cooking to a household currently consuming 5 kWh is a very different proposition from adding the same amount to a household already consuming 100 kWh. For the former, cooking could become the largest electricity service in the home.
But what I don’t know from the data is whether that creates challenges around wiring, local transformer loads, network capacity and peak demand.
Nevertheless pivoting existing expenditure into the electricity system would create new load, new demand, on networks that have already been built, and assuming there is full cost recovery built into the overall tariff system it would improve utility revenue. (I note that Ethiopia at that time was not full cost recovery, and had a very low tariff…I know they are trying to sort that out while keeping the well being of the population in mind, but I don’t know the details yet). The point remains though that higher utilisation can improve the economics of existing grid infrastructure where the additional revenue exceeds the incremental costs of serving the load.
What I have deliberately left to one side
There is an important part of the household cooking economy that this cash-expenditure comparison does not capture: freely collected biomass.
Many households, particularly in rural areas, collect their own firewood rather than purchasing it. The MTF survey may therefore record very little or no cash expenditure even though the household is consuming substantial quantities of cooking energy. I have deliberately not tried to turn that collected wood into a monetary value in this exercise. Doing so would require assumptions about collection time, labour value, local wood markets and opportunity cost. Those are important questions, but they would change the nature of this particular comparison.
Here I wanted to answer a narrower question: how much cash expenditure can actually be seen moving through the household energy economy? The way I have handled it, households reporting zero cash expenditure on cooking fuel were left in the analysis rather than excluded. Many of these are likely to be households relying partly or wholly on collected biomass. Some households may also collect wood alongside purchasing other cooking fuels. So the figures are intended to represent the full range of households within each electricity-consumption group, rather than only those that purchase cooking fuel.
This treatment means that zero-cash households pull the median down relative to what we would see if we looked only at households purchasing cooking fuel. The graph is therefore conservative when compared with the typical cash spend of fuel purchasers: it shows median reported cooking-fuel expenditure across the whole analysed group, including households reporting zero.
I took this approach deliberately because it prevents exaggerating how much cash is already being spent on cooking fuel. It also means that the graph does not attempt to value the potentially substantial time, labour or economic cost associated with freely collected biomass.
In Figure 2 below, I reran the exercise with households reporting cooking-fuel cash spend of zero excluded. This makes quite a difference in the low-use groups. For example, Rwanda <5 kWh rises from $2.3 → $7.5/month, Kenya 5–10 kWh from $3.9 → $7.1, and Ethiopia <5 kWh from $4.2 → $6.2.

There is another equally important group that this analysis does not capture: households that are not yet connected to electricity. The analysis starts with connected households because I wanted to understand what happens to the economics of an electricity connection once cooking is included. It should not be read as an assessment of the whole population.
Why this matters for electricity investment (Hello M300?)
This exercise points towards a wider planning question. Electricity investment decisions are often made using assumptions about relatively modest residential consumption: lighting, televisions, phone charging, refrigeration and perhaps a few other appliances. Cooking is frequently treated separately, as part of a clean cooking strategy rather than an electricity-demand strategy.
But if cooking migrates onto electricity, the economics of the electricity system can change. A household consuming 5 or 10 kWh a month today may become a 30 or 40 kWh customer. A rural feeder populated by apparently very small consumers may have substantially greater underlying demand potential than its existing meter data suggest.
That does not automatically make every grid extension commercially attractive. Nor does it remove the need for investment in generation, networks, reliability or household appliances. But it possibly changes the denominator.
If expected electricity demand excludes one of the household’s largest energy services, then the expected return on electricity infrastructure may also be understated. This is particularly relevant where governments, utilities and development banks are considering the economics of extending or strengthening networks. The investment case is often assessed against the electricity demand that households exhibit shortly after connection.
The MTF data suggest that this may be a poor guide to the amount of energy service demand those households actually have. Cooking expenditure shows that some of the demand already exists. It is simply being met by another fuel. This is the approach of the Burundi programme, build in demand for eCooking and it improves Average Revenue Per User, enabling either improved IRR and NPV, or enables a lower tariff.
Putting cooking back into electricity planning
The exercise began as a relatively simple manipulation of the MTF data: take household electricity consumption, put customers into bands, and see how many households and people sit at different levels of use. Linking those records to cooking expenditure confirmed my hypothesis. Very low electricity consumers are not necessarily households with low energy-service demand or low energy expenditure.
In some groups, households spending almost nothing on electricity are spending two, three or more times as much on cooking fuels. If even part of that cooking service shifts to efficient electric appliances, electricity demand and utility revenues could rise significantly while household cooking costs may fall.
Rather than asking only: “How much electricity do newly connected households currently consume?”
we should also be asking: “How much energy are those households already consuming and paying for outside the electricity system — and what happens if cooking becomes part of the load?”
For electricity planners, utilities and investors, that may be an important part of the demand story.
…………………………………………….
AI disclaimer: AI assisted with data analysis, and improved paragraph constructs.
Figure 1, top: MTF energy consumption data from three countries organized by electricity consumption ‘buckets’. Assisted by AI. (Credit: S. Batchelor).