The inclusion of external variables in forecasting maize production: A dynamic regression model with ARIMA errors

Abstract This paper used time series analysis to forecast maize production by using regression with ARIMA (1, 0, 1) errors model. The aim was to examine the influence of external information in forecasting maize production in Tanzania using annual time series data from 1961 to 2022. The study examined the dynamic relationship between maize production and selected exogenous factors to generate medium-term forecasts. By integrating regression components with ARIMA disturbances, the approach improved predictive reliability beyond conventional univariate models and provides interpretable insights into the climate–production nexus. The results indicate a modest but steady upward trajectory in maize production over the next eight years, with relatively stable 80% and 95% forecast intervals. This suggests gradual productivity gains rather than rapid structural transformation. The findings imply that maize output remains sensitive to climatic conditions, underscoring the importance of incorporating exogenous information in agricultural forecasting models. From a policy perspective, the forecasts support calibrated grain reserve planning, targeted input subsidy allocation, and expanded investment in climate-resilient practices such as drought-tolerant seed adoption and small-scale irrigation. The evidence also highlights the need for stronger institutional coordination between meteorological services, agricultural research bodies, and planning authorities to operationalize data-driven decision-making. Notwithstanding its contributions, the study is limited by reliance on annual aggregate data, which may mask sub-national heterogeneity and nonlinear climate effects. Future research should explore higher-frequency and spatially disaggregated data, regime-switching dynamics, and scenario-based simulations to account for emerging climate risks. Overall, the study demonstrates that dynamic regression modelling offers a robust and policy-relevant framework for forecasting maize production under increasing climatic uncertainty.