A field experiment conducted in rural Rwanda evaluates the revealed willingness to pay for different off-grid solar technologies
Imagine the following: it is 6 a.m. and you wake up just as the first rays of sunshine stream in. You light your kerosene lamp and a pungent smell starts to spread. Yesterday, you managed to charge your cellphone with the barber’s car battery. It took you 50 minutes by foot and ten cents, even though you only had a dollar to spend for the day. If that old cellphone still has charge remaining, you may use it to tune in to the radio. There is no refrigerator, TV, computer or electric cooker you may avail to start your day. Indeed, you will most likely not use any electronic appliance today, either at home or at work. This is the daily routine for 1.1 billion people in developing countries who lack of access to modern electricity.
Change of scenery: It is 9 p.m., and you—a hands-on high-level policy advisor in an African capital— have just left your Ministry office behind and are in your car on your way home. Your thoughts wander. You are convinced that electrification was a necessary condition for the economic transition of nowadays-industrialized countries. However, you are reluctant to believe that it is a sufficient one for today’s developing countries. Astronomical levels of investment appear to be needed to ensure universal electrification, amounting to an additional 19 billion USD annually until 2030 for Africa alone. At the same time, a myriad other unfulfilled basic needs vie for your government’s limited financial resources. On top of that, researchers have struggled to provide you with a consistent picture of the short- and medium-term benefits of electrification. As you pull into your driveway, you wish somebody would just tell you how highly electricity access is actually valued (or not) by the population itself, and what might be appropriate electricity technologies to consider. How cost-effective re these different technologies? Is the grid necessary? Or could the market just bring off-grid solar electricity to all rural homes by 2030? Maybe if paired with microcredit access? Providing subsidies always feels uncomfortable. At least this is what the Wall Street Journal conveys.
Change scenes, yet again: In autumn 2015, we embarked on a field study in Rwanda to answer these questions. We conducted an experiment with 325 households living in fifteen remote villages without access to modern energy across the country. We offered them three types of solar kits: (1) a 0.5-watt, (2) a 3.3-watt, and (3) a 20-watt device. These devices, broadly speaking, are the lowest tier of modern electricity—or, in other words, the “lower bound of the electricity technological spectrum”. The first is essentially a solar lamp that provides clean lighting, while the second also enables cellphone charging or use of a radio. The third can light several rooms and power several low-wattage gadgets, such as radios, simultaneously. What they cannot do is run productive machinery or high-wattage appliances such as TVs or refrigerators. If you demand these services, access to the main electric grid (the “upper bound”) is required. The Rwandan government has actually been quite effective in expanding its national grid to rural areas through its Electricity Access Roll-Out Programme (EARP). We evaluated these efforts and found that in spite of high power supply demand for electricity is generally low, amounting to a median monthly consumption of no more than six kWh. This reflects use of a few light bulbs and a radio, whereas machinery usage is rare. You may now argue that it takes time for people to purchase devices, to adapt their lifestyles. This may be true, but we observe no such change over a four-year period. While access to electricity certainly does no harm—indeed, people are generally keen to connect, and report being satisfied—this is where cost-effectiveness considerations come into play in a big way. Our simple back-of-the-envelope calculation suggests that connecting one household via EARP in Rwanda costs around 1,500 USD. A recent study in Kenya by Lee, Miguel, and Wolfram takes a more detailed look at demand for and costs of grid electricity, and observes that households’ willingness to pay (WTP) for grid connections falls considerably short of their cost—by around 500-1,000 USD per household. This is the non-trivial potential “social cost” of grid expansion. Needless to say, the upper bound may not be feasible in many settings.
So, back to the lower bound: the three solar kits we bring to our intervention cost 13, 37, and 180 USD (depending on capacity), and just like our colleagues in Kenya, we measure households’ WTP and compare it to costs. However, measuring people’s valuations for unknown products is not easy: respondents may be undecided, try to get a good deal by being strategic, or just tell you whatever they think you want to hear. To solve these issues, we make a real purchase offer, and apply what is known as the Becker-deGroot-Marschack bidding mechanism. Behavioral economists like this approach because it is “incentive-compatible”, which means participants fare best if they reveal their true WTP. It works just like eBay—you place a bid for some device (in our case, solar kits) and if your bid is higher than one randomly picked from an urn (which plays the role of other eBayers!) you get to purchase the device for this lower price. If your bid is less than the randomly drawn price, you walk away empty-handed.
We also took the opportunity to randomize a policy intervention into our field experiment. As many, we believe that liquidity constraints explain much of the poors’ inability to purchase potentially beneficial technologies. To test this, we randomly assigned three different payment periods to the households: paying in full with seven days, six weeks, or five months for their solar kits.
So, how much are the poor willing to pay for solar energy? And did the randomized credit schemes make a differences? Here are the answers: rural Rwandan households are willing or able to pay only between 38 and 55 percent of solar kits’ current market prices. Only? Well, yes, if you want prices paid by them to cover costs. But this valuation is as high as twenty percent of total monthly household expenditures for the small solar lamp and up to 340 percent for the large 20-watt device, suggesting high latent demand for technologies that provide modern energy services. Relaxing households’ credit constraints does not seem to make much of a difference: credit schemes of up to five months do not significantly increase WTP. This implies that market-based mechanisms on their own may not be sufficient to meet the universal energy access targets now enshrined in the UN Sustainable Development Goals.
Comparing these valuations for solar kits with their costs, we find that off-grid electrification still entails a funding gap (or “social cost”) of between 8 and 83 USD per connection, depending on the type of solar kit. This gap, however, is much lower than the shortfall estimated by Lee, Miguel and Wolfram for on-grid electrification. In addition, households may not be factoring in beneficial external effects on neighbors, the environment, and public health, or long-term private gains, such as better education. While these additional benefits are surely higher for on-grid electrification than for off-grid electrification, we expect such considerations to bridge only some of the large social cost gap between on-grid and off-grid approaches. (This expectation, by the way, is based on a prior study we conducted in Rwanda.)
So, what would we tell the concerned policymaker? We suggest the following: policy should highly subsidize medium-size solar technologies to make sure they reach the poorest of the poor, and concurrently pursue an on-grid strategy in areas with business potential, where electricity demand is high and impacts can be expected to be, too. Looking ahead, we should think about ways to develop smart subsidies for solar electricity, accompanied by careful cost-benefit mechanisms to plan how and where to extend the grid. Needles to say, this story is to be continued…
This blogpost was first published at Duke Household Energy and Health Initiative Bolg