Abstract:
This study attempted to determine the relationship between real defence expenditure and real aggregate output for Ghana. The limited nature of existing literature for the case of Ghana and reducing defence expenditure over the years in the absence of pressing security issues are some of the reasons for undertaking this study. To achieve this, the Auto-Regressive Distributed Lag (ARDL) cointegration approach for time series analysis is employed on a data set from1980 to 2017. Before the ARDL model was employed, seasonality and unit-root tests were used to determine the specific characteristics of the variables. A cointegration relationship was found among the estimating variables, hence long and short run ARDL models were estimated. The error correction term in the short run model was negative, supporting the ARDL bounds test for cointegration. The results indicated an insignificant impact of defence expenditure on aggregate output in both periods. Although insignificant, it supported the findings of other studies.
Diagnostics test were used to confirm the stability and robustness of the model. A granger causality test was also employed to determine the direction of the causal relationship existing between the variables. This test supported the findings of the ARDL estimations by discovering that defence and government expenditure do not Granger cause output. However, in the long run model, interest rates, labour and the lag of output were the only positively significant variables. Whilst the past values of aggregate output were the only significant variable in the short run. It was recommended that the GAF increases services such as manufacturing, construction, education, and health. Since it has physical and human capital that may not be available to the civilian sector. The government was also advised to create an environment that promotes growth and development which will influence government expenditures which will then affect the economy of Ghana as postulated by Wagner’s law.