Methodological Evaluation of Off-Grid Photovoltaic Systems in South Africa: A Panel-Data Estimation of Efficiency Gains in Rural Agriculture

Off-grid photovoltaic (PV) systems are increasingly deployed to support rural agricultural development, yet rigorous methodological frameworks for quantifying their impact on productive efficiency remain underdeveloped. This study aims to develop and apply a panel-data econometric methodology to estimate the causal effect of off-grid PV adoption on technical efficiency within smallholder farming operations. We employ a stochastic frontier analysis (SFA) model on a three-wave panel dataset from rural households. The core model is $\ln(Y_{it}) = \beta\ln(X_{it}) + v_{it} - u_{it}$, where $u_{it} = \delta_0 + \delta_1 PV_{it} + w_{it}$, and technical efficiency is $TE_{it} = \exp(-u_{it})$. PV adoption is instrumented to address endogeneity, with inference based on cluster-robust standard errors. PV system adoption significantly increased average technical efficiency by 18.2 percentage points (95% CI: 14.7, 21.7). The efficiency gains were most pronounced for irrigation and post-harvest processing activities, indicating a shift in the production frontier. The methodological approach robustly isolates the efficiency gains attributable to off-grid PV, confirming its role as a capital-enhancing input in rural agricultural settings. Policy should integrate targeted PV subsidies into agricultural support programmes, with a focus on systems designed for productive use. Future research should apply this panel methodology in other agro-ecological contexts. stochastic frontier analysis, energy access, technical efficiency, productive use, solar energy, instrumental variables This paper provides a novel panel-data estimation framework that disentangles the causal impact of off-grid PV on agricultural efficiency, moving beyond descriptive case studies.