ABSTRACT
Although the economic effects of defense expenditures have become an issue of intense interest over the recent decades, little is known about how the individual components of military spending affect the economy. Nevertheless, military spending is highly heterogeneous in its nature consisting of assorted categories that broadly encompass salaries’ payment, operations, training, research and development, and maintenance of equipment, arms and facilities. Naturally, this implies that military spending can affect the economy in various and probably contradictory ways. Hence, by considering this distinctive element of military spending, the present paper aims to uncover the economic effects of the most important components of the US defense budget focusing on a period from 1949 to 2021. Applying linear and non–linear methods on a Barro–style regression, the statistical evidence reported herein suggests that heavy reliance on the military sector entails potentially high opportunity costs.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Supplementary material
Supplemental data for this article can be accessed online at https://doi.org/10.1080/10242694.2022.2145717
Notes
1. U.S. Bureau of Economic Analysis, https://www.bea.gov/data/gdp/gross-domestic-product.
2. Stockholm International Peace Research Institute (SIPRI), https://milex.sipri.org/sipri.
3. Cumulatively, the reduction in total defense outlays reached 39.7% from 1968 to 1976. At a disaggregated level, spending on military personnel and procurement reduced by 41.1% and 77%, respectively, whereas the decrease in O&M and RDT&E activities was equal to 23% and 32.6% respectively.
4. See Note 2.
5. See Dunne, Smith, and Willenbockel (Citation2005) for an extensive overview of the theoretical models employed in the relevant literature.
6. The nonlinearity assumption is also supported by the DBS test results of Broock et al. (Citation1996). Under the null hypothesis the test assumes that the series are identically and independently distributed (i.i.d.), while rejection of the null hypothesis confirms the series’ nonlinear dependencies.