Recouping after Coup-Proofing: Compromised Military Effectiveness and Strategic Substitution

ABSTRACT

In order to counter the threat of a coup, states often undertake a number of strategies to “coup-proof” their militaries, such as creating institutional redundancy, severely limiting interbranch communications, and basing promotions on loyalty rather than merit. As a result of such policies, however, the fighting effectiveness of these armed forces is degraded, and the marginal return on military investment is greatly reduced. We argue that leaders who have coup-proofed their militaries undertake several substitution policies in order to offset their military weakness when faced with external threats. These policies include pursuing chemical, biological, or nuclear weapons and forging alliances. We find support for these theoretical predictions in quantitative tests on data with global coverage between 1970 and 2001.

KEYWORDS:

• Alliances
• civil-military relations
• international security
• nuclear proliferation

Acknowledgments

The authors wish to thank Ulrich Pilster and Tobias Böhmelt for allowing us to use their data set on military forces and Michael Horowitz and Neil Narang for allowing us to use their data set on WMD proliferation. We also extend our appreciation to Erik Gartzke, David Lake, Daniel Maliniak, as well as participants at ISA-West, MPSA, the 2011 Peace Science Conference, and UCSD’s IR workshop for their comments on earlier drafts of this article. Brown would like to thank Tel Aviv University’s Institute for National Security Studies (INSS). Replication materials are available at a dataverse archive maintained by the authors at http://dx.doi.org/10.7910/DVN/29145 and cross listed at the dataverse page maintained by International Interactions at http://thedata.harvard.edu/dvn/dv/internationalinteractions.

Funding

Authors Brown and McMahon gratefully acknowledge support from the Institute on Global Conflict and Cooperation. McMahon also acknowledges funding from a National Science Foundation Graduate Research Fellowship. The views expressed in this article are those of the authors and do not reflect the official policy or position of the United States Air Force, Defense Department, or the United States Government.

Supplementary Material

A supplemental appendix providing additional details about the analyses presented in this article is available on the publisher’s Web site at http://dx.doi.org/10.1080/03050629.2015.1046598.

Notes

¹ We recognize that other strategies exist, such as the use of proxy forces, but data limitations prevent us from exploring these alternatives in this article.

² Throughout this section, we rely on illustrative examples from the Middle East because this region is particularly relevant for our research question. Regimes in Iraq, Syria, Egypt, Libya, Iran, Tunisia, and Yemen all saw governments successfully overthrown—and in all cases but Tunisia and Iran, by their own military officers. Syria and Iraq experienced several such takeovers before a strongman finally established himself well enough to maintain power. All told, there were over 40 coup attempts in 11 countries, with 27 occurring during the 1960s alone (Brooks 1998:13, 81). The civil-military stability of Middle Eastern regimes since 1970 (that is, until the Arab Spring of 2011) can be explained by the extensive use of coup-proofing strategies in these countries (Rubin and Kearney 2001).

³ Our approach toward factors that influence military disposition is consistent with the framework that Feaver (1999) discusses. Changing the disposition to intervene involves: “(a) adjusting the ascriptive characteristics of the military so that it will be populated by people inclined to obey, and (b) adjusting the incentives of the military so that, regardless of their nature, the members will prefer to obey” (Feaver 1999:226). The decision to focus on these strategies is warranted, given their prominence in the historical record and numerous previous studies of coup-proofing (for example, Pilster and Böhmelt 2011, 2012; Powell 2012; Quinlivan 1999).

⁴ “Additional Captured Documents Reveal Again the System of Money Transfers to Terrorist Squads, Personally Authorized by President Yasir Arafat, with the Deep Involvement of Marwan Barghouti,” Israeli Defense Forces document TR6-498-02, June 24, 2002.

⁵ For an alternative perspective on this issue, see McMahon and Slantchev (2015).

⁶ Ironically, once the Iranians had repelled the Iraqi advance and a new cadre of military leaders had gained valuable experience, Khomeini again purged his military of officers who had previously served for the Shah.

⁷ Turkey and Israel are not exceptional, as both have also faced insurgencies. See Rubin’s chapter in Rubin and Kearney (2001).

⁸ Leeds (2003) argues that alliances commitments are most likely to be broken when the costs of formation and violation of these agreements are low or when conditions have changed significantly between when alliances are formed and when they are triggered.

⁹ Frisch (2002) has argued previously for a connection between coup-proofing and alliances. Frisch’s argument, however, puts causality in the opposite direction: He purports that regimes that have a great power (especially American) patron can then afford to increase the level of institutional division in their armed forces.

¹⁰ The temporal scope is limited by the availability of data for certain key variables. The data for counterbalancing begin in 1970 (Pilster and Böhmelt 2012). Alternatively, we need to control for the threat environment, including those interactions that do not devolve into full-scale war. The most comprehensive data for measuring coercive interactions among states—the Militarized Interstate Dispute data set—is available in the form needed for our analysis only through 2001 (Ghosn, Palmer, and Bremer 2004). As a result, we have limited all tests to the 1970–2001 time frame, since this is the maximum period for which data on all variables are available.

¹¹ Given the nature of WMD, in particular how hard it is for foreign intelligence and others to detect these weapons, any data are going to be suspect. This opacity should lead to primarily Type II error in the data, which should bias against findings that support our hypothesis. Moreover, Horowitz and Narang (2014) and Way (2011) have gone to great lengths to develop accurate information for chemical, biological, and nuclear weapons pursuit and acquisition.

¹² The only state-years that are censored from the analysis of weapons pursuit are those for which a state is in possession of the munitions. If states give up their nuclear or chemical/biological weapons, they are again eligible to pursue these weapons after the stockpiles have been eliminated. For example: In the case of South Africa, we again allow for the pursuit of nuclear weapons for the years following the dismantling of its stockpile.

¹³ In an alternative specification, we develop a “Powerful Allies Count” measure that sums only the defense-pact allies that are materially stronger than the state of interest. Material strength is determined by comparing states’ Composite Index of National Capabilities (“CINC”) scores (Singer 1987). The results for these tests do not differ significantly from those presented in the following.

¹⁴ Such measures have a long history in comparative politics (for example, Taagepera and Shugart 1989). The Pilster and Böhmelt (2011) variable is coded using the Military Balance data from the International Institute of Strategic Studies. While some question the reliability of certain aspects of this data set (see Colgan 2011), Pilster and Böhmelt (2011) show that the states with high values for Effective Number tend to be among the states with the highest coup risk, a strong indicator of convergent validity (see also Pilster and Böhmelt 2012).

¹⁵ See generally Enloe (1976) and Vanhanen (1999). In the Middle East, see Droz-Vincent (2011:5), Zisser (2001:13–25), al-Marashi (2002), and Bligh (2001). In Africa, see Welch (1986), Decalo (1989), Goldsworthy (1981:58), Anderson (1999:93), Good (1974:13), and N’Diaye (2002:624). In Nepal, see Nepali and Subba (2005:99). For Latin America and the Caribbean, see Enloe (1978) and Kaufman and Haklai (2008).

¹⁶ Concurrent validity is a type of construct validity designed to assess the ability of a variable’s operationalization to distinguish between groups that it should be able to distinguish between (Trochim and Donnelly 2008).

¹⁷ We adopt the most expansive view of militarized competition and include the dyad-years for states that undertake interactions labeled as “isolated” in the Klein et al. (2006) data within the set of potential rivals.

¹⁸ We conduct similar analysis on tests that omit these dependency variables. The results are largely consistent in terms of the direction and statistical significance of the key regressors.

¹⁹ For both the Oil and Mountainous Terrain, ln variables, we extend the figures for 1999 (the last year of Fearon and Laitin’s data) through 2001.

²⁰ The percentage of mountainous terrain is multiplied by 100, such that a state with mountains on half of its territory is represented as 50% before the log transformation, rather than 0.5.

²¹ Iraqi President Saddam Hussein feared that growing the country’s conventional forces to face the country’s Iranian enemy would endanger the regime from within. The Iraqi government began ordering the use of chemical weapons—the only WMD available to the regime at that time—to augment its conventional forces, which were trying desperately to cope with a dangerous enemy (Hiro 1991).

²² For more on interaction terms, see Brambor, Clark, and Golder (2006).

²³ The predicted probability of Nuclear Weapons Possession for Nuclear Weapons in Minority Regimes is 98.4% versus 42.6% for non-Minority Regime states.

²⁴ While some might counter that South African Defense Forces were considered highly effective relative to their neighbors, this does not mean that they could not have been even more effective had they been able to draw from a larger pool of talent.

²⁵ This relationship is not driven by the need for additional forces to maintain a WMD stockpile. The Effective Number (ln) variable is based on the number of ground combat-capable troops within different branches of the military. The measure is not inclusive to personnel whose primary job is to maintain the weapons or to personnel like pilots and submariners who are tasked with employing air- or sea-launched munitions. In addition, the vast majority of ground troops in WMD-armed militaries train for and conduct conventional warfare. Even among the few ground combat organizations capable of employing WMD, most are tasked primarily with conventional kinetic warfare. As a result, the pursuit and possession of WMD should not strongly influence the distribution of personnel that the Effective Number (ln) variable measures.

²⁶ This comparison is again made with covariate values set to Iraq in 1986.

²⁷ The claim of Peruvian use comes from Reuters Television, February 10, 1995, Ref: 605160364, “Peru: Ecuador Claims to Have Shot Down Two Peruvian Planes Attacking Its Military Posts.” Peru’s military officially denied having used chemical weapons in its conflict with Ecuador. Peru, Joint Command of the Armed Forces, Official communique No. 011 CCFFAA, Lima, February 24, 1995, as cited by the ICRC (https://www.icrc.org/customary-ihl/eng/docs/v2 cou pe rule74). For sources that do include Peru among suspected proliferators, see Horowitz and Narang (2014) and the Office of Technology Assessment, Appendix 2-A: Sources on Tables Listing Countries of Chemical and Biological Weapon Concern Proliferation of Weapons of Mass Destruction: Assessing the Risks, OTA-ISC-559 (1993) (http://www.wws.princeton.edu/cgi-bin/byteserv.prl/ ota/disk1/1993/9341/934105.PDF). It should be noted that there is some doubt whether Peru did, in fact, pursue chemical weapons, and few sources on the topic include Peru as a likely proliferator. We consulted several of the top CW experts on this question, but none could definitively confirm or disconfirm whether Peru had a program. One fact that cast some doubt is that Peru did not disclose a CW program when it ratified the CWC.

²⁸ In terms of “ethnic stacking,” it should also be noted that until 2006, Peru was essentially a minority regime, whereby indigenous peoples and Afro-Peruvians were almost entirely disenfranchised from political and economic power, as well as from the upper echelons of the military.

Additional information

Funding

Authors Brown and McMahon gratefully acknowledge support from the Institute on Global Conflict and Cooperation. McMahon also acknowledges funding from a National Science Foundation Graduate Research Fellowship. The views expressed in this article are those of the authors and do not reflect the official policy or position of the United States Air Force, Defense Department, or the United States Government.