Military Expenditure, Endogeneity and Economic Growth
The debate over the economic effects of military spending continues to develop, with no consensus, but a deepening understanding of the issues and limitations of previous work. A recent survey has suggested that the inclusion of post Cold War data has tended to make finding a negative effect more common, but issues remain (Dunne and Tian, 2013). One particularly important issue that has not been adequately dealt with, is the possible endogeneity of military spending in the growth equation, mainly because of the difficulty of finding any variables that would make adequate instruments. This paper considers the likely importance of endogeneity, using conflict onset as an instrument for military spending in an endogenous growth model for a panel of African countries 1989-2010. Following a brief review of the literature the theoretical and empirical models are outlined and the use of conflict onset as an instrumental variable for military spending in the panel estimates is justified. The empirical analysis suggests that endogeneity is likely to be an important issue and using IV estimation provides a larger significant negative effect for military spending on growth than OLS. It also identifies a further potential bias in the same direction in studies not including non-military spending in the growth equation. These results imply that the damaging effects of military spending on growth in Africa are being underestimated in most studies. While it is clear that conflict onset is a suitable and successful instrument in this analysis, the results are not directly generalisable. Conflict onset is unlikely to be applicable to a larger and more diverse panel of countries. What is of general concern is the finding that endogeneity is important and is likely to be influencing the results of studies of military spending and growth. It is important that future research tries to deal with endogeneity and the search for reasonable instruments is one that needs to engage researchers.
