Military Spending and GDP Growth: Is there a General Causal Relationship?

Abstract

The existence of a causal relationship between GDP growth and military spending has been a long-lasting debate. Earlier studies report mixed findings employing cross-sectional, pooled or fixed effects panel data. Recent studies employing causality tests also find mixed results. In this study, it is argued that structural changes in data (i.e. significant economic or military events) play an important role and should not be ignored. Also, treatment of stationarity around mean and around time trends needs special attention. Structural changes, if ignored, have the potential to cause bias in long-run tests through VECM. Mistreatment of trend stationarity may result in spurious results. With a sample of 65 countries, for the 1975–2004 period, a Granger-type causal relationship between military spending and GDP growth is analyzed with special emphasis on structural change and stationarity around time trends. While trend stationarity is a common trait of the variables, there are structural changes in the variables' time trends. Considering the bias towards rejecting stationarity in the presence of structural changes, the Zivot–Andrews unit root test is employed. A causal relationship is reported between military spending and GDP growth for 54 (of the 65) countries. For the overall sample, panel data Granger causality estimations provide evidence for a bi-directional positive causal relationship.