Demand is not just the desire of some to posses certain goods or services, rather it is that desire accompanied by a specific willingness/ability to pay. For example, a rural household might want an improved mud stove, but are not willing to pay for it because the current cooking fuels and technologies (3 stones, firewood and family members' time) are considered free and the improved mud stove is not worth the initial investment according to their cost-benefit analysis. In this case, there would be no market demand.

We can also talk about “pent-up” demand, demand that theoretically should exist, but for some reason is limited (one classic example of this is consumers delaying purchases of large durable goods they must eventually have until economic conditions become more favorable). This idea of theoretical, or latent, demand can also be applied to various energy market contexts in developing countries.

Actual demand can be described by an amount or level (a peso, rupee, or cedi figure). It can also be described by its composition, an important attribute since not all products and services are created equal. Increasing energy access through the primarily market-based mechanisms described in this toolkit works through improving the overall composition of the energy bundle. This may have a net positive, negative or neutral effect on the monetary value of demand, but in the long term demand has always risen with rising incomes.

Desire for something
Willingness to pay
Level and composition of demand

"Desire for something”

Certain types of desires are pretty universal:
  • Desire for service, cooking, lighting, other electrical --> promotes other activities of importance (eating, working, studying, entertainment, etc.)
  • Desire for quality, doesn't break easily, isn't expensive to maintain
  • Desire not to be uncomfortable, sick, or injured
  • Desire to reduce time and physical burdens

Traditional energy practices often imply significant hazards and burdens, exerting a significant toll on the people involved in producing, transporting and consuming the energy. Indoor air pollution, for example, is estimated by some to be responsible for more premature deaths every year than malaria, afftecting mostly women and children. Candles and fuel-based lamps result in an untold number of buildings burning, often with injuries and fatalities. Who knows how many collective hours are spent every year gathering firewood?

"Willingness to pay"

Everywhere in the world, the poor and the marginalized pay more for energy than do their richer and less marginalized counterparts; they spend larger percentages of their total budgets on energy and they pay more per unit of energy service delivered.

The phrase "willingness to pay for energy services" has been used in several ways. In some accounts, it is used to describe the actual amount that individuals or groups actually spend on the energy services they employ. This willingness to pay is often assumed be level of energy spending that market segment will support though, with the introduction of new products, people may be incentivized to spend more (on a levelized, life-cycle basis) in order to have a higher level of service or, conversely, they may opt to spend less for a similar level of service and pocket the savings to be used elsewhere. True willingness to pay for energy services is the amount the customer agrees to spend on a specific service, which is hard to determine in circumstances where that service doesn't yet exist and there is a limited amount of knowledge about how the technology will function in the locale.

It is important for a project developer to make a reasonable attempt to assess the customers' willingness to pay as part of the fact-finding and market reasearch phases. This might include a basic household survey which investigates 1) how much customers are already spending on energy services and, 2) after briefly introducing the new product, receiving feedback on whether customers would be interested in purchasing one and in what price range. Often, consumers are willing to pay premiums above current expenditures for higher quality energy.

The following represent a snapshot of household energy expenditures from across the world. Hopefully, the list can continue to grow and more commonalities discovered across locales, market sectors and pyramid segments.

Ahiataku-Togobo_-_Ghana_Cost_of_Cooking.JPG1. This first chart contains estimates of household cooking expenditures in Ghana, starting from the total "at-pot" energy required to cook for an average-sized household, deducting the thermal inefficiencies of the end-use appliances in each case, calculating the amount of the fuel needed, its cost at market prices, and arriving at a final estimate.

In 2005, 1 USD traded for approximately 9,000 GHC, making this last column approximately:

These results are consistent with experiences in many other countries, where, as a rule of thumb, household expenditures on cooking (where fuel and stoves are purchased) usually fall anywhere from $5 to $15 per month.

2. Instead of working backwards from cooking requirements, prices and equipment efficiency, a recent household survey in Liberia undertaken by CSET, funded by Daphne (results not yet published) examined what households were actually spending on energy consumption.

  • Average monthly rural spending on firewood and charcoal as assessed by this survey was $5 and $9 respectively.

  • Average monthly rural spending on lighting (non-electric) ranged from $4 (dry cell battery users) to $6 (candles) to $10 (kerosene).

  • Average expenditures on electricity in the rural areas (using gas-powered gensets) was $43.

The next step is to match willingness to pay, however it is determined (current spending or hypothetical spending in a new scenario), with an appropriately sized and priced product. The following tool may be useful for helping to define and envision the different market segments and test assumptions about what technology solutions may be appropriate and affordable for them, and under what conditions. This tool is presented and explained here.

"Demand, Level and Composition"

Almeida_&_De_Oliveira_-_Brazilian_Life_Style_Changes.JPG This next graph depicts primary energy demand in Brazil. It is shown how both total demand and the composition of that demand change over time. This graph is interesting for a number of reasons.

Imagine, if possible, there is another line which overlays the chart representing household expenditures. This line might start out very low, representing households that don't purchase their firewood and only acquire small amounts of electrical services using, say, dry cell batteries. From 1 to 5 - 10 minimum wage units, one can imagine that the level of spending increases pretty steeply even though primary energy demand remains relatively constant.

At some point, about 5 - 10 minimum wage units, firewood pretty much drops out of the picture and total energy demand begins to increase.

Does the shape of this graph hint at the existence of a basic, or subsistence, level of demand? The poorest households are still consuming as much primary energy as their neighbors up to 10 times as wealthy than they are. Of course, much of this energy is dangerous and wasted through various inefficiencies. It would be interesting to try and translate the results into a service-based metric, say number of person-meals cooked and kilolumen-hours of light provided, as opposed to an energy-based metric, and see how the results compare.

One of the best presentations of comparative costs and service provision was provided by Evan Mills, "The Specter of Fuel-based Lighting," document re-posted here: Mills_Science_FBL_Full.pdf


Household energy expenditures, matching purchasing power with affordable technologies over time

Once household energy expenditures have been ascertained, the next step is to determine what else those same expenditures can purchase.

This spreadsheet, Purchasing Power of Different Market Segments (over time).xls aims to create a framework for thinking about these issues.

1. Enter the type of service being considered, ie cooking, lighting, motive transport, agricultural processing, etc.

Type of energy fuel or service

2. Choose how to segment the market based on energy expenditures on that service.


Enter upper limit of HH EE per mo. for each segment

Segment 1
Segment 2
Segment 3
Segment 4
In this case, the poorest-of-the-poor or bottom-of-the-pyramid households have been defined as households spending less than $6 per month on lighting. In this example, the definition is completely arbitrary. In practice, local circumstances should be used as guidelines to define the different market segments. For instance, in one setting, families living in mud houses with thatched roofs may be considered the poorest segment. How much do they spend on lighting?

3. Enter the discount rate for each segment.

The discount rate entered represents a monthly discount rate. In this case, .02 has been chosen as a rough estimate for all segments, but in reality, it is expected that the poorer the household, the higher the discount rate might be. For a review of NPV and discount rates, click here.

Enter Discount Rate for each segment

Segment 1
Segment 2
Segment 3
Segment 4

4. Time horizon. How long of a time horizon is being considered?

Factors influencing the choice of time horizon might include the longevity of the product and the maximum possible financing term lengths available in the market. Enter this number in years.

Time horizon (yrs)

5. Choose a product to test and input its cost structure.

First is the level of service equal to of greater than what the household is currently using? How much does it cost up front? To operate? What are the capital and operating subsidies available to help defray these costs.

Level of service
Initial cost
Operating/ maintenance per mo.
Capital subsidy
Operating subsidy

6. Based on this information, the tool will calculate the anticipated levelized cost over the time period.

Final Levelized Cost over the time period

This is equal to the capital cost minus the capital subsidies plus difference between the operating costs and subsidies over the time horizon.

7. The output of the tool calculates the NPV of the segments' purchasing power over the time horizon.

Segment 1
Segment 2
Segment 3
Segment 4
Here, for example, the range of segment 1's purchasing power (from $0 - $6 per month) is $0 to $208. This represents what a segment 1 household would have available to spend if it were able to convert all future energy cash flows over a period of five years into today's net present value (using a 2% monthly discount rate).

8. This information is also represented graphically.

9. The color-coding on the right-hand side of the spreadsheet indicates whether the technology chosen will be affordable for all households in the segment (green), some households (yellow), or none (red).

In this example (note, this example is totally arbitrary and all numbers made up), no households in segment 1 or 2 can afford the proposed technology. Some in segment 3 can and every household in segment 4 can.

Discussion: Notice that if the time horizon was changed from 5 years to 1 month, no household would be able to afford a system. This is obviously not what happens in reality, but it does demonstrate the power of credit to make energy products more available for poor households. In reality, some people can purchase improved products out of their savings, but this is severely limited.

The model could be improved by adding including a space for a one-time payment from household savings (say, a down payment) and could be expanded to more precisely try and model household cash flow to repayment terms (perhaps even making adjustments for seasonality of income), but that is beyond the scope of this tool. The take away messages are two-fold:
  1. Even the poorest segments of the market have relatively large amounts of purchasing power when allowed to aggregate that power over time.
  2. Some segments may never be able to afford what is deemed to be a basic and sufficient energy consumption bundle. But, before simply announcing that such a segment is simply "too poor to pay," that basic energy entitlement package must be defined and some simple analysis performed to show that in fact, even with microfinancing, the purchase is out of reach. Last, what level of additional subsidy is actually needed in order to bring that purchase within reach?