Replicating the Cost-Effectiveness Diagnostics of Ethiopian Field Research Station Systems: A Time-Series Forecasting Model for the 2021–2026 Period

Field research stations are critical infrastructure for agricultural innovation, yet their financial sustainability is often uncertain. Previous evaluations of their cost-effectiveness in the Ethiopian context have relied on static, retrospective analyses, limiting forward-looking planning. This study replicates and extends a prior methodological framework to develop a dynamic, time-series forecasting model. The objective is to assess the projected cost-effectiveness of a national network of research stations over a future five-year period. We employ a replication methodology, applying a seasonal autoregressive integrated moving average (SARIMA) model to historical operational and output data. The model, specified as φ(B)Φ(Bˢ)(1-B)ᵈ(1-Bˢ)D yₜ = θ(B)Θ(Bˢ)εₜ, forecasts key performance indicators. Forecast uncertainty is quantified using 95% prediction intervals. The model forecasts a declining trend in aggregate cost-effectiveness, with a projected decrease of approximately 18% over the forecast horizon. Station-level analysis revealed significant heterogeneity, with prediction intervals indicating high uncertainty for stations in drought-prone regions. The replication confirms the utility of time-series forecasting for ex-ante evaluation but highlights greater systemic financial pressure than previously estimated. The methodological approach provides a more nuanced, probabilistic assessment for resource allocation. Resource allocation should be prioritised towards stations with robust forecasted performance. Implementing real-time data capture systems is essential to improve model accuracy. Policy should integrate these forward-looking diagnostics into strategic planning cycles. agricultural research systems, forecasting, SARIMA, financial sustainability, operational research This study provides a novel, probabilistic forecasting tool for the ex-ante assessment of agricultural research infrastructure, moving beyond descriptive evaluation to inform proactive management and policy.