Principal investigator: Omar Barroso Khodr
Authors: Niccolò Meriggi; Erwin Bulte; Ahmed Mushfiq Mobarak
Original title: Subsidies for technology adoption: Experimental evidence from
rural Cameroon
Location of the Intervention: Cameroon (Country in Central Africa)
Sample Size: 199 villages (in the Adamawa region – northern Cameroon)
Sector: Economic Development & Environmental Economics
Primary Variable of Interest: The study examines three primary outcomes in different models: I) Adoption (solar lamps – binary); II) Usage per minute (solar lamps); and, III) Willingness to pay (continuous).
Type of Intervention: Subsidy for the adoption of sustainable energy (solar lamps)
Methodology: Ordinary Least Squares (OLS); Tobit; Instrumental Variables (IV); and Fixed Effects.
Summary
Economists have observed low adoption of well-being-promoting technologies in economically disadvantaged regions. This study investigates, through a two-stage experiment, the impact of temporary subsidies on the adoption, use, and future demand for modern solar lamps. Using an auction model with random variation in the reservation price among villages in Adamawa, northern Cameroon, the researchers generated variations in purchase prices and adoption rates. The results indicate that subsidies increase adoption without compromising future use. Even when they reduce future willingness to pay, this effect is overcome by learning about the benefits of the product, resulting in greater future appreciation. Although lower prices attract users with lower usage intensity, the allocative role of prices remains relevant. Finally, there is no evidence of social learning or anchoring effects beyond the initial sample.
- Policy Problem
The authors examine the transition from traditional lighting sources (such as kerosene) to modern, grid-independent lighting using solar and battery-powered LEDs in rural areas of Adamawa villages, where 47% of the population still lacks access to electricity. Considering the high costs of grid expansion, decentralized solutions, such as solar lamps and residential systems, play a crucial role in reducing the energy gap faced in the northern region of Cameroon.
In this context, during the study period (2013-2015), it was observed that rural families began to adopt LED lamps powered by batteries and solar energy, replacing kerosene. The study suggests that this change is related to the cheaper prices of solar lamps, making this alternative more accessible. Furthermore, early-stage energy transitions tend to involve scalable and low-cost investments—such as portable solar lamps. In more advanced stages, expansion to grid connections is possible as family income increases. However, the authors point to significant challenges in the implementation of solar energy, especially regarding the costs of expanding the electricity grid more broadly in rural areas.
Therefore, off-grid solar solutions become fundamental to guaranteeing access to energy in these locations. Regarding development, the authors highlight mixed evidence on the economic effects of solar electrification. Previous studies indicate reduced energy costs and increased study time for students, which could result in educational improvements.
However, these benefits tend to be limited or only noticeable in the long term. This discrepancy may be related to changing habits in the communities analyzed—such as children switching to night classes—or to external constraints, such as a shortage of teachers and difficulty accessing the job market. Still, the authors emphasize that other research shows positive impacts on family income after electrification, especially in the long term. In any case, they reinforce that the results may be modest if there is no adequate infrastructure and qualified professionals (Hamburger et al., 2019; Peters and Sievert, 2016). The sustainable expansion of solar energy, therefore, depends directly on how it is planned and executed.
Finally, according to the authors, the implications for policy and research indicate the need for more accurate counterfactuals. Many impact studies do not consider the basic lighting conditions present in communities—such as the difference in performance and demand between solar lamps compared to kerosene lamps versus battery-powered LEDs. This methodological limitation may compromise the true assessment of the benefits of the energy transition.
Furthermore, the effectiveness of solar lighting is directly linked to local production and support conditions—such as capacity-building systems, the presence of industries, and market access. These factors determine whether the technology will be successfully adopted or not.
In any case, solar lamps and innovation in home systems represent a viable alternative in the short term, especially before the complete expansion of the electricity grid in remote areas. However, the authors emphasize that these solutions only deliver significant results when accompanied by complementary investments, such as in education and the promotion of entrepreneurship. Such investments are crucial to maximizing the economic and social benefits of decentralized electrification.
- Policy Implementation Context
The study analyzed access to energy in rural communities, revealing that approximately 95% of families depended on kerosene lamps at the beginning of the research. These lamps, in addition to providing low-quality lighting, posed serious health and safety risks, with a high potential for fires. Simultaneously, they had a significant economic impact, consuming up to 5% of the family budget. In this context, the adoption of modern alternatives such as battery-powered or solar LED lighting was scarce, highlighting a significant gap in access to clean energy.
To understand purchasing behavior and encourage the adoption of solar technologies, the authors implemented a randomized auction based on the method. Becker-DeGroot-Marshak (BDM). The intervention was initially applied in 172 villages, later reduced to 90 due to budgetary constraints. Three discount levels were defined: 25%, 50%, and 75% on the market price of solar lamps. The values were adjusted according to price variations throughout the study (2013–2015). At the beginning of the period, the base price was approximately US$9 or R$24. Exercise prices were randomly assigned by village to avoid cross-treatment side effects. Transparency of the process was ensured through sealed envelopes, publicly opened at community events.
The analysis revealed that larger discounts significantly boosted the purchase of solar lamps, demonstrating high price sensitivity. The auction served as a valid metric for understanding the value attributed by families to the lamps, especially since most respondents were unaware of the market price, reducing the anchoring effect. Two years after the auction, self-reported usage data (not measured by sensors) indicated high product satisfaction among early adopters, continued use of the solar lamps, and the possible presence of a sunk cost effect, where higher prices initially paid stimulated greater usage—although this hypothesis requires further verification.
The role of social learning and spillover effects was also investigated. Researchers assessed whether knowledge of previous exercise prices influenced bidding decisions. The conclusion was that few participants knew the external prices, suggesting low anchoring effects. The study points to several recommendations for energy inclusion policies, in which subsidies are effective tools to overcome adoption barriers arising from high initial costs. Furthermore, purchasing behavior was not strongly influenced by external price references, which strengthens the use of random pricing mechanisms. Finally, the presence of early adopters can generate a positive influence on future demand, although peer effects have not yet been robustly proven.
- Evaluation Details
For empirical analysis to be necessary, the study required meticulous data collection combined with an experimental design. The first stage measured Willingness to Pay (WTP, or as the text describes it). Willingness-to-Pay – WTP) for solar lamps through the BDM auction with pre-defined random prices, as previously mentioned. In the second stage, another auction was held two years later (in 2015) to assess WTP, lamp usage habits, and behavioral consistency. The baseline data consisted of fundamental family characteristics such as education, age, and income. Subsequently, data randomness checks confirmed the balance between the treatment groups.
In this context, the first stage of the study was conducted with specific guidelines. Initially, price sensitivity was analyzed, observing an adoption rate of 70% for the lowest discounted prices, in contrast to only 40% for the highest prices. This exercise revealed that the demand for solar lamps was highly elastic: more aggressive discounts (or reduced prices) significantly stimulated consumer adoption.
According to the authors, the behavior of the bidders presented diverse characteristics. The so-called decliners (13,5%) offered bids above the exercise price, but did not make the purchase—possibly due to liquidity constraints. On the other hand, the bargainers (10,9%) made bids below the exercise price, but later sought to acquire the product—perhaps due to regret or access to new information.
To account for these inconsistent behaviors, the bids were adjusted: those of the decliners were reduced, while those of the bargainers were increased. However, these adjustments did not alter the main results of the study.
Finally, it was observed that reductions in lamp prices above 40% increased adoption by 20%, and reductions greater than 50% generated a 40% increase in adoption, evidencing a non-linear demand pattern.
Regarding behavioral consistency, no evidence was found that participants who refused or negotiated in Stage 1 made different bids in Stage 2. The absence of correlation between stages regarding refusal status suggests that there was no systematic bias in the BDM mechanism.
Furthermore, the inclusion of new participants — the so-called “co-villagers” — in Stage 2 made it possible to assess the indirect effects caused by the initial adopters on the rest of the community.
In summary, the two-stage experimental design provided robust evidence on price elasticity, barriers to adoption, and behavioral patterns in off-grid solar energy markets. While subsidies have proven effective in driving adoption, practical challenges such as liquidity constraints and post-purchase regret indicate the need for tailored policy solutions to ensure continued adoption. Finally, the results highlight the relevance of incentive-compatible mechanisms—such as BDM auctions—in accurately measuring demand in energy access interventions.
- Method
This study adopts a regression-based approach to analyze experimental data, using clustered standard errors at the village level (172 or 90 clusters, depending on the model applied). The analysis is conducted in two main stages, considering different dependent variables and hypotheses related to the adoption of technologies (such as solar lamps), usage patterns, and Willingness to Pay (WTP). The central methodological strategies include:
- Ordinary Least Squares (OLS) Regression: applied to most models.
- Tobit Models: used when the dependent variable is censored (for example, minutes of lamp use).
- Instrumental Variable (IV) Models: Implemented to address endogeneity in technology adoption, such as instrumenting adoption through exercise prices.
Furthermore, the models incorporate district fixed effects and baseline controls to capture socioeconomic characteristics of households, such as age, education level, and income. Three main outcomes are examined, in which the first stage assesses, through a binary variable, whether respondent i, in village j, acquired the solar lamp in the first stage. The regressors include dummies For medium and high exercise prices (low price as the base category). Immediately following, the second stage, measured by the number of minutes of lamp use in the 24 or 72 hours prior to the follow-up interview. This stage also analyzes the sunk cost effect, that is, whether higher initial prices induce greater use. Thus, the Willingness to Pay (WTP) assessed considers the impacts of prices and adoption in the first stage. Regression IV, with adoption instrumented by exercise prices, allows isolating the effects of anchoring versus social learning. Finally, it explores whether the second stage is influenced by factors such as the exercise prices previously practiced in the villages (anchoring) and the density of pioneer adopters (social learning).
In summary, the experiment employs two-stage OLS, Tobit, and IV regressions, with clustering at the village level. The main fixed effects and controls include age, education, and income. The central dependent variables are: adoption (binary), lamp usage (minutes), and DAP (continuous). The hypotheses tested address: sunk cost effect, screening effect, anchoring versus social learning, and social spillovers.
- Main results
The study presented mixed results. Firstly, there was no solid evidence of behavioral biases. The data indicate an absence of anchoring effects: participants did not base their bids on initial prices. Similarly, no sunk cost effects were observed, meaning that higher initial payments did not result in greater long-term use.
On the other hand, subsidies proved effective in increasing future demand as users discovered the benefits of the product. However, they also attracted consumers with limited use of the lamps, which compromised the cost-benefit ratio.
Willingness to pay (WTP) remained lower than cost. Even after exposure to the experiment, most households bid below the full cost of the light bulb in subsequent auctions. The authors suggest that market-based approaches alone are not sufficient to guarantee universal access to energy.
Although the amortization period was short—between 6 and 10 months, through kerosene savings—the adoption rate was low. Thus, liquidity constraints may be influencing the adoption, although the evidence is contradictory.
Finally, regarding social learning and spillover effects, minimal impacts were observed among peers: information about the benefits of the lamps did not spread significantly beyond the first users. These results are consistent with other studies that indicate passive and limited social learning in low-income contexts (Dupas, 2014b; Beaman et al., 2021; Sayinzoga et al., 2016).
- Lessons in Public Policy
The study indicates that subsidies can boost the adoption of solar lamps, provided they are accompanied by careful planning. Short-term subsidies, for example, help families recognize the value of the product, but can also attract users with low actual usage, requiring the implementation of targeting mechanisms.
In this context, access to credit can complement subsidies. If liquidity constraints are an obstacle to demand, microfinance models or pay-per-use systems may offer viable solutions. The authors highlight that expanding the electricity grid may not yet be feasible. If even small solar lamps face barriers to adoption, rural electrification may require long-term approaches.
Additionally, the study acknowledges limitations in the experiment: results may vary if participants have access to non-market options, such as prior knowledge of retail prices. This reinforces the importance of developing more refined consumption typologies, as behavioral biases—such as anchoring—can influence different user segments differently, especially when comparing durable and non-durable goods. The authors also emphasize the need for future research that further analyzes the product type. The observed effects may vary in technologies geared towards health, agriculture, or productive use.
In summary, the study concludes that subsidies can stimulate the adoption of solar lamps by facilitating consumer learning. However, behavioral biases, such as anchoring and sunk costs, proved to be insignificant in this context. Market demand alone is not sufficient to guarantee universal access to energy, and the effects of social learning remain limited.
Therefore, policymakers should seek a balance between short-term subsidies and targeting mechanisms, as well as explore complementary financing alternatives—such as credit—to expand welfare gains. Future research should investigate heterogeneous consumer responses and consider different product types to improve the design of subsidy policies.
References
Beaman, L., BenYishay, A., Magruder, J. & Mobarak, A. (2021) Can network theory-based targeting increase technology adoption? American Economic Review.
Dupas, P. (2014) Getting essential health products to their end users: subsidize, but how much? Science, 345(6202), pp.1279–1281.
Peters, J. & Sievert, M. (2016) Impacts of rural electrification revisited – the African context. Journal of Development Effectiveness, 8(3), pp.327–345.
Sayinzoga, A., Bulte, E. & Lensink, BW (2016) Financial literacy and financial behavior: experimental evidence from rural Rwanda. The Economic Journal, 126, pp.1571–1599.
