Abstract:
This paper has considers the interpretation of the empirical results of the developing literature on the demand for military spending that specifies a general model with arms race and spillover effects and estimates it on cross-section and panel data. It questions whether it is meaningful to talk of an ‘arms race’ in panel data or cross-section data, and suggests that it may be more appropriate to talk about the relevant variables – aggregate military spending of the ‘Security Web’ (i.e. all neighbours and other security-influencing powers) and the aggregate military spending of ‘Potential Enemies’– as acting as proxies for threat perceptions, which will reflect both hostility and capability.