For many rural households beyond the grid, the cheapest source of electricity is a solar home system: a solar panel and battery sold with lights and small appliances. Under a pay-as-you-go (PAYGo) contract, a household pays off the system by buying access time with mobile money, and the company switches the system off remotely whenever that time runs out. In rural Kenya, a week of access costs about KES 140, against about KES 110 a week spent on the traditional energy a system replaces.
Lang (2026) studies how existing PAYGo customers in Kenya respond to the price of access time, using an experiment run with a solar company from August 2018 to February 2019. About 400 randomly selected customers were offered one or two free days of access, also randomly assigned, in any month in which they bought at least four weeks. About 400 others received only an informational phone call. On average the reward did not significantly change purchases, but under the two-day reward the customers with the highest prior demand bought about two more days a month, nearly 8 percent more. Using these responses and a model of linear demand, the paper bounds each group’s price elasticity and consumer surplus: at least US$19.46 a month for the highest-demand customers, US$0.87 for the lowest, and US$10.65 on average.
The paper then calculates the MVPF of a hypothetical government subsidy to the price of a PAYGo solar home system, using the framework of Hahn et al. (2024) for goods with environmental externalities. The paper separately estimates the MVPF of a hypothetical government subsidy to the price of a PAYGo solar home system in Rwanda.
MVPF = 1.7
The paper expresses every component per US$1 of subsidy. That dollar costs the government US$1, plus US$0.07 of administrative cost. The administrative cost is the cost per dollar that Aker et al. (2016) estimate for a cash transfer delivered by mobile phone in Niger, which the paper considers feasible here because PAYGo customers already make their payments by mobile money.
The subsidy also changes tax revenue by drawing in new customers. The paper assumes the households a subsidy would bring in most resemble its lowest-demand customers, and so uses their price elasticity of −2.69 and the price they pay over the life of a contract, US$237 in present value. Each US$1 of subsidy then induces about 2.69/237 = 0.011 additional systems. In Kenya, solar home systems carry the same 16 percent value-added tax as the kerosene and disposable batteries they replace. Over the contract, the median rural household spends about US$101 on the energy services a system can replace, less than the US$237 the system costs. Each induced system therefore raises 0.16 × (237 − 101) = US$21.76 in tax, or US$0.247 per US$1 of subsidy.
The paper counts no other fiscal externality. Studies in Kenya, Zambia and Bangladesh find no lasting increase in children’s study time from solar lighting. Given how little solar home systems cut emissions relative to the world’s total, the avoided climate damage is unlikely to raise future tax revenue meaningfully.
The net cost is then US$1 + US$0.07 − US$0.247 = US$0.823 per US$1 of subsidy.
Households that would have bought a system anyway value each US$1 of subsidy at US$1. Households drawn in by the subsidy were close to indifferent about buying, so their own gain from the purchase is close to zero and is not counted.
The remaining benefit is the avoided carbon emissions of the induced systems. Fetter and Phillips (2019) estimate that each solar home system in East Africa avoids US$13.70 of carbon damage a year. Over a three-year system life with a 15 percent discount rate, that is US$35.97, which the paper rounds to US$36. At 0.011 induced systems per US$1 of subsidy, the carbon benefit is (36/237) × 2.69 = US$0.409 per US$1. The paper does not value improvements in local environmental quality, and notes that this may understate the benefits of solar home systems.
The willingness to pay is then US$1 + US$0.409 = US$1.409 per US$1 of subsidy.
The MVPF of a marginal subsidy for PAYGo solar home systems in Kenya is US$1.409 / US$0.823 = 1.71, which the paper reports as at most 1.7:
\text{MVPF} = \frac{1 + \frac{36}{237}(2.69)}{1 + 0.07 + \frac{-21.76}{237}(2.69)} = \frac{1.409}{0.823} = 1.71
The paper notes that this estimate is an upper bound. The experiment bounds the elasticity rather than pinning it down, and this estimate uses the most elastic demand (consistent with its lowest-demand customers not responding to the bonus).
The formula assumes perfect competition. The paper also reports an MVPF that allows for its estimate that the solar company loses money on its lowest-demand customers, about US$77 over a contract. If the government had to make up that loss for each customer the subsidy draws in, the net cost would rise to US$1.697 per US$1 and the MVPF would fall to 0.83 (0.8 in the paper’s main text), which the paper treats as a lower bound.
The MVPF uses the elasticity of how many days of access existing customers buy. The paper’s extrapolation of its demand estimates to new customers implies that a subsidy equal to the carbon benefit would raise adoption by about 8 percent. As US$36 is about 15 percent of the US$237 price used in the MVPF, that corresponds to an elasticity of adoption of about −0.53, which would give an MVPF of 1.06. The paper notes that this extrapolation probably understates what a subsidy would achieve, because demand is more price-sensitive among its lower-demand customers.
Aker, Jenny C., Rachid Boumnijel, Amanda McClelland, and Niall Tierney (2016). “Payment Mechanisms and Antipoverty Programs: Evidence from a Mobile Money Cash Transfer Experiment in Niger.” Economic Development and Cultural Change, 65(1): 1–37. DOI: https://doi.org/10.1086/687578
Fetter, Rob, and Jonathan Phillips (2019). “The True Cost of Solar Tariffs in East Africa.” Policy Brief NI PB 19-01, Energy Access Project, Nicholas Institute for Environmental Policy Solutions, Duke University. https://energyaccess.duke.edu/publication/the-true-cost-of-solar-tariffs-in-east-africa/
Hahn, Robert W., Nathaniel Hendren, Robert D. Metcalfe, and Ben Sprung-Keyser (2024). “A Welfare Analysis of Policies Impacting Climate Change.” NBER Working Paper 32728. DOI: https://doi.org/10.3386/w32728
Lang, Megan (2026). “Using Price Incentives to Bound Welfare from Pay As You Go Solar Electricity.” Journal of Public Economics, 259: 105656. DOI: https://doi.org/10.1016/j.jpubeco.2026.105656