The manufacturing sector in Ethiopia has undergone significant structural changes, yet a persistent gap exists in robust, quantitative methodologies for evaluating systemic efficiency and forecasting future performance. Existing analyses often lack rigorous engineering-focused time-series frameworks. This study aims to develop and validate a novel time-series forecasting model to measure and project efficiency gains within Ethiopian manufacturing systems, providing a methodological evaluation of plant-level operational dynamics. A longitudinal dataset of key performance indicators from multiple manufacturing plants was analysed. The core forecasting model is an autoregressive integrated moving average (ARIMA) formulation: $y_t = \mu + \phi_1 y_{t-1} + \theta_1 \epsilon_{t-1} + \epsilon_t$, where parameters were estimated via maximum likelihood. Model diagnostics included checks for stationarity and residual autocorrelation. The ARIMA(1,1,1) model provided the best fit, forecasting a 17.5% aggregate efficiency improvement over the forecast horizon. Parameter estimates were statistically significant, with a 95% confidence interval for the autoregressive term $\phi_1$ of [0.42, 0.58]. The analysis identified energy utilisation as the most volatile systemic factor. The proposed time-series model offers a validated, quantitative tool for assessing manufacturing systems efficiency, demonstrating significant predictive capability for strategic planning within the industrial sector. Manufacturing plant managers should adopt similar forecasting methodologies for capacity planning. Policymakers are advised to integrate such models into national industrial performance monitoring frameworks to better target interventions. manufacturing systems, efficiency forecasting, time-series analysis, ARIMA modelling, industrial engineering, operational performance This paper introduces a novel application of ARIMA modelling for forecasting manufacturing efficiency gains in an emerging industrial context, providing a replicable methodological framework and a new longitudinal dataset for the region.