Vector Autoregressive Model of Maize Production in Northern Region of Ghana

Agricultural growth plays a crucial role in the Comprehensive African Agriculture Development Programme (CAADP, 2009) agenda. The program recognizes that increasing agricultural productivity is essential for reducing poverty, meeting food production targets, and lowering production costs and food prices for the impoverished. This study aimed to develop two types of models. The first model employed a vector autoregressive (VAR) approach, which involved regressing the production of maize in one district against the production of maize in other districts at various lags. The second model utilized a VAR framework where the maize production in each district was regressed against the production of maize in other districts and the corresponding climate conditions at different lag periods. The available data spanned from 1968 to 2018 and were recorded on an annual basis. Six climatic variables were included in the analysis. A lag order of 3 was selected for the models. The results of the autocorrelation test using the portmanteau test indicated no serial autocorrelation across all lag periods. The test for normality revealed that the residuals followed a normal distribution. Additionally, there was no evidence of heteroscedasticity in the data. Furthermore, a Granger causality test was conducted on the selected districts to explore causal relationships. Variance decomposition analysis was performed to assess the variance relation in the data and understand the contribution of different factors. Based on adjusted R-squared, mean absolute error (MAE), and root mean squared error (RMSE) values, the models that incorporated climatic variables were found to be the most suitable for forecasting maize production in the selected districts. The VAR model, which captures the interdependencies between the variables, was utilized in this analysis. All variables in the VAR model were treated symmetrically, meaning that their relationships were considered equally important.