Jones Paintsil

Howard University · Economics · 2026–27 job market

Fields

Applied Micro, Development, Health

Job market paper

Outage Duration, Capital, and Firms’ Choice between Diesel Generators and Solar in Developing Countries

Unreliable electricity forces firms in developing economies to secure their own power, a need historically met by diesel generators. On-site solar is now cost-competitive with the generator over its lifetime, yet the generator still dominates. I ask whether outage exposure explains the adoption of either technology and, where it does not, what does. I develop a random payoff model of the firm's self-generation choice in which the two technologies differ in two respects. A generator is dispatchable and can be matched to an outage, and the model implies a threshold outage length beyond which running it pays, which makes outage duration rather than frequency the operative margin. Solar without storage follows the sunlight, and its large upfront cost makes adoption depend on firm capability and access to finance rather than on outage exposure. I test these predictions with the 2025 and 2026 World Bank Enterprise Surveys for 11,252 firms in 36 developing economies, with 17 high-income economies as a benchmark. I estimate linear probability and multinomial logit models with country and sector fixed effects and standard errors clustered by country, and I report the wild-cluster bootstrap as a robustness check. As a further check on omitted local conditions, measurement error in self-reported outages, and reverse causality, I instrument outage exposure with the outages reported by other firms in the same locality. The findings are as follows. First, generator ownership is associated with outage duration, while outage frequency shows no significant association. Second, the link between outage exposure and solar adoption is small and marginally significant. Third, longer outages raise the probability of holding a generator, on its own or with solar, but not of holding solar only. Fourth, solar adoption tracks firm capability, measured by size, exporting, and digital presence, while bank credit does not predict adoption once capability is controlled for, and firms' assessment of the courts does not change how either technology responds to outages. The paper makes three contributions. It extends the literature on self-generation, which has treated backup power as fossil generation, to the choice between the generator and solar. It identifies outage duration, not frequency, as the margin that drives generator investment. It separates the dispatchability and capital channels, which explains why falling panel prices have not displaced the generator. Firms appear to adopt the two technologies for different purposes, the generator as cover against outages and solar as a capital investment that lowers energy costs. Policies that ease the financing burden of solar would expand clean self-generation, whereas reducing reliance on the generator requires investment in reliable electricity infrastructure.