French Arms Exports and Intrastate Conflicts: An Empirical Investigation

Abstract

The aim of this article is to evaluate how French Major Conventional Weapons (MCW) exports impact on the conflict intensity of recipient countries. The recent increase in French arms exports seems to contradict the French political discourse on the promotion of regional stability. We run zero-inflated ordered probit model in order to analyze the role of the arms trade on the intensity of civil conflicts in 144 countries from 1992 to 2014, using SIPRI and UCDP/PRIO data. Our results suggest that French MCW exports tended not to exacerbate intrastate conflicts during this period. This finding is robust to changes in the empirical framework. We propose two lines of explanations: France seems to be prone to choosing partners that respect human rights and selling more ‘defense-oriented’ MCW than the rest of the world.

Keywords:

• Major conventional weapons
• arms trade
• intrastate conflicts
• France
• zero-inflated ordered probit models

JEL Codes:

• C25
• D74
• F5
• H56

Acknowledgments

We thank Amaury Barra, Elvire Guillaud, and two anonymous referees for helpful comments. This version also benefits from feedback from participants of the 7th Congrés des associations francophones de science politique and the 2018 meeting of the European Public Choice Society as well as in various seminars in France. The usual disclaimer applies.

Notes

1. Some papers deal with the detection of illegal traffic (see DellaVigna and Ferrara (2010) or McDougal et al. (2015)) but quantifying its extent appears difficult to implement. The SIPRI database only covers legal trade based on deliveries. In the case of France, these deliveries are well-documented because of control in arms exports and several public reports.

2. See http://www.smallarmssurvey.org/weapons-and-markets/transfers/exporters.html.

3. This market has followed an upward trend since 2001 and is dominated by Austria, Belgium, Brazil, Germany, Italy, Russia, South Korea, Switzerland and the USA. Moreover, the assault rifle of the French armies (FAMAS) has to be replaced but French firms are no longer able to produce such equipment, so a contract with a German company (Hecker and Koch) has been officially notified by the French procurement agency.

4. For a recent empirical analysis, see Martinez-Zarzoso and Johannsen (forthcoming).

5. According to the Uppsala Conflict Data Program (UCDP), an intrastate conflict is a conflict between a government and a non-governmental party, with no interference from other countries.

6. Interested readers may find more details about theoretical channels in Baugh and Squires (1983). They propose five ‘models’ linking up arms trade and conflict.

7. It should be noted that the view of the stabilizing school presented in this paper is slightly different from the one analyzed by Anderton (1995). Indeed, his analysis based on the restoring of the balance of power seems more accurate for cases of international or regional conflicts (e.g. Intriligator and Brito (1984)). By focusing on the infra-national level, we almost systematically observe unbalanced situations in the sense that the conflict opposes an organized State and rebel groups that do not have equivalent military capacities.

8. One originality of Moore (2012) lies in the fact that he takes into account both the government and the rebel organization. In particular, he finds ‘strong evidence […] that the provision of MCW to rebels increases the severity of civil wars’ (Sambanis 2004, 339).

9. The commitments of France are listed in the Report to the Parliament on arms exports (Ministère des Armées 2017, 23).

10. See for example the publication of Amnesty International: https://www.amnesty.fr/controle-des-armes/actualites/ventes-darmes-une-hypocrisie-francaise (in French).

11. See http://discours.vie-publique.fr/notices/123002283.html (in French).

12. See https://www.sipri.org/databases/armstransfers/sources-and-methods

13. It leads to the exclusion of 9 countries: Belize, Bhutan, Comoros, Costa Rica, Guyana, Marshall Island, Palau, Solomon Islands, Tonga and Tuvalu.

14. See http://www.eia.gov/cfapps/ipdbproject/IEDIndex3.cfm?tid=5&pid=53&aid=1

15. We also run our test with the variable from the World Bank which represents the share of natural gas resources related to GDP. But this variable is not significant.

16. One could also say that ethnic fractionalization matters. Results presented in the next section also hold when we replace Language _i by the variable of ethnic fractionalization available in the QoG database. However, for the sake of clarity, we choose to only expose tables with Language _i . It is supported by Bormann, Cederman, and Vogt (2017) who considers that language is the most conflict-promoting variable of fractionalization.

17. It should be noticed that Language _i is time invariant and could capture other country specificities. Therefore, one should remain cautious in the interpretation of its coefficient.

18. For this variable some values are missing due to the code ‘-66’ which represent cases of foreign ‘interruption’' are treated in PolityIV database as ‘system missing.’

19. A likelihood ratio test allowed us to discriminate between a negative binomial regression and a ZIOP model.

20. It should be noticed that, for ZIOP models to be identified, we use NGIntensity _i,t on the inflated equation, but not in the outcome equation. Indeed, if conflicts in neighboring countries may spread (Bara forthcoming), it is less clear how it directly affects the intensity of the conflict. We also choose not to include $X_{i,t}^{\mathit{FR}}$ on the inflated equation because the paper deals with the impact of French arms exports on conflict intensity, and not on probability of incomplete peace.

21. Same kind of results have been found in Dunne and Tian (2016) in their ZIP models.

22. For countries which never experienced intrastate conflict, we estimate the peace period since the end of World War II.

23. This variable relies on information collected by SIPRI. See https://www.sipri.org/databases/embargoes.

24. Results obtained with a Probit and a Logit specifications are similar.