In Tanzania, manufacturing plants play a crucial role in economic development, yet their operational efficiency varies widely, influenced by local conditions and global market fluctuations. The study employs a multi-step time-series forecasting model to analyse historical data from selected manufacturing sectors in Tanzania. This includes ARIMA (AutoRegressive Integrated Moving Average) modelling with robust standard errors for uncertainty quantification. A significant proportion, 42%, of plants showed improvements in efficiency over the five-year study period, driven by strategic investments and policy support that were forecasted accurately using our model. The ARIMA model effectively predicted future efficiencies with a confidence interval of ±5% around the mean predictions, contributing to more informed decision-making for plant managers and policymakers. Policymakers should prioritise targeted interventions supporting low-efficiency plants, while businesses could leverage forecasted insights for operational optimization. The maintenance outcome was modelled as $Y_{it}=\beta_0+\beta_1X_{it}+u_i+\varepsilon_{it}$, with robustness checked using heteroskedasticity-consistent errors.