Trading Arms and the Demand for Military Expenditures: Empirical Explorations Using New SIPRI-Data

Abstract

This paper analyzes the impact of arms imports and exports on national military expenditures. The recent literature on the determinants of military expenditures has mainly focused on countries’ external security environments and their regime type. Based on existing theoretical work, we argue that, in addition to these factors, arms trade flows may have an important role to play. First, we show that rising imports of major conventional weapons do not necessarily translate into higher defense spending. Rather, this relationship depends on political, economic, and contract conditions that influence different choices of financing imports. Therefore, the effect should be very heterogenous. Second, exports may have both a negative or a positive impact depending on regime type and the perceived impact of exports on national security. We empirically test these expectations for 156 countries from 1949 to 2013 using arms trade and new military expenditure data provided by SIPRI. Employing static and dynamic panel data models, we find that the effect of arms imports on defense budgets does indeed differ between regions and time periods. With respect to exports, there is evidence of a strategic substitution effect between military expenditures and arms exports in democratic countries: increases in arms exports are followed by a reduction in military expenditures. For non-democratic societies on the other hand, arms exports do not tend to be associated with lower military spending.

Keywords:

• Arms trade
• military expenditure
• dynamic panel data models

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

^1. There have only been some studies on the sub-national level that analyze the relationship between expenditures and the development of regional defense industries Short (1981); Malecki (1984).

^2. Of course, in certain conflict regions of the world this still seems highly prevalent. See, for example, a recent, new game-theoretic variant of the arms-race model that tries to explain arms races between Israel and Syria (Bar-El, Kagan, and Tishler (2010).

^3. It goes without saying that spending on the military may also serve other economic or political purposes. It may be necessary to ensure and solidify the backing of the military for a government, it may be used to satisfy certain constituencies, or it could be part of a larger economic stimulus program.

^4. Note that instead of looking at these different causal mechanisms independently, one could, alternatively, take a more integrated approach and analyze the nexus of arms trade and defense spending as contributions to a national and international defense good. From this perspective, decisions to spend, import, and export weapons are strategic decisions of countries to benefit from a desired security level taking other countries’ contributions into account. We leave a game-theoretic approach to future research.

^5. Note that the empirical analyses in this paper are not exhaustive and will not test all these possible links.

^6. We would like to thank Sam Perlo-Freeman to point out this example to us.

^7. The sensible assumption made here is that the international arms trade is a highly regulated and state controlled market where not firms but governments decide whether or not and how much to trade.

^8. Of course, one could argue that rising military spending could induce an arms race leaving a country’s security situation only better off in the short run but worse off in the long-run. At this point, we will not analyze this particular scenario, but would expect a further rise in spending therefore.

^9. In the original model by Levine (1994), there are actually four cases that cover all combinations of short and long-term enemy/ally. We simplify our analysis to two basic cases to keep the subsequent empirical analysis tractable.

^10. Detailed definitions and sources for all dependent and independent variables are provided in the Appendix 1.

^11. Major conventional weapons do not include weapons of mass destruction and small arms and light weapons (SALW). It covers a wide range of weapons systems such as armored vehicles, ship, submarines, and aircraft. This also includes components such as engines and sensors, but also satellites, anti-aircraft, and other defense systems (see http://www.sipri.org/databases/yy_armstransfers/background/coverage).

^12. TIVs are not meant to measure the financial value of an arms transfer but are based on the estimated production cost of a core weapon’s system. The intention is to measure the amount of military resources that are being transferred through the trading of arms.

^13. See the Norwegian Initiative on Small Arms Transfers (NISAT) that collects trade data on SALW (http://nisat.prio.org/Trade-Database).

^14. Conflict scores range from 0 (no conflict) to 10 (highest intensity) and capture the societal and systemic impact of a conflict (Marshall 2014).

^15. Version Sep 9, 2014.

^16. Note that values smaller than 500.000 TIV are coded as zero in the data-set. Hence, we chose to use the minimum export volume captured in the data for identifying potential exporters.

^17. Countries are defined as democracies if they have a Polity-IV score of five or higher, otherwise they are classified as non-democracies.

^18. We would like to thank an anonymous referee for suggesting this possibility to us.

^19. The sample mean for arms export is 147 and the standard deviation is 924.