The systematic measurement of efficiency gains in industrial settings is critical for sustainable development, yet robust, longitudinal datasets and forecasting methodologies tailored to the specific infrastructural and operational contexts of emerging economies are scarce. This data descriptor presents a novel, curated panel dataset and evaluates a bespoke time-series forecasting model designed to quantify and project manufacturing plant efficiency gains, with the objective of providing a reproducible analytical framework for structural engineering and industrial policy analysis. The methodology integrates plant-level operational data with national industrial statistics. A seasonal autoregressive integrated moving average with exogenous variables (SARIMAX) model, specified as (B) (Bˢ) ᵈₛD yₜ = (B) (Bˢ) ₜ + Xₜ, was developed and validated. Model parameters were estimated using maximum likelihood, with robust standard errors computed to account for heteroskedasticity. The forecasting model indicates a positive secular trend in aggregate manufacturing efficiency, with projected gains averaging 2. 3% per annum over the forecast horizon. A key finding is the significant role of capital reinvestment cycles, which account for approximately 40% of the explained variance in efficiency improvements. The constructed dataset and validated model provide a reliable empirical foundation for analysing industrial efficiency trajectories, demonstrating the utility of context-specific time-series approaches in engineering economics. Future research should incorporate granular energy consumption data and machine-level telemetry. Practitioners are advised to adopt similar modelling frameworks for capacity planning and to inform maintenance scheduling for optimal plant performance. industrial efficiency, time-series forecasting, SARIMAX, manufacturing, panel data, engineering economics This work provides the first open-access, plant-level longitudinal dataset for the region's manufacturing sector and introduces a validated forecasting model that explicitly incorporates local operational cycles, filling a critical gap in infrastructure performance analytics.
Bekele et al. (Sat,) studied this question.