Ballistic Missiles and International Conflict

Abstract

Governments are concerned about the dangers posed by ballistic missiles. However, there is almost no theoretical or empirical scholarship on ballistic missiles. This article presents and tests the conventional wisdom that the spread of ballistic missiles makes conflict more likely. Original data on ballistic missiles and on crisis initiation is collected, and analysis using a variety of statistical models is conducted. It is found that among all directed dyads from 1946 to 2007, potential challengers possessing ballistic missiles are significantly more likely to initiate international crises. Further, potential targets armed with ballistic missiles are significantly less likely to be challenged. Crises are less likely to escalate when targets are armed with missiles. The results are obtained after accounting for several control variables. Analysis reveals that the findings are not affected by possible endogeneity bias. The analysis also reveals complex interactive effects between ballistic missiles and nuclear weapons on the onset of international crises.

Keywords

missiles proliferation nuclear weapons conflict

Ballistic missile proliferation gravely concerns policy makers. Ballistic missiles, along with nuclear, chemical, and biological weapons, are often categorized as weapons of mass destruction (WMD) that threaten international peace and stability. When the United Nations imposed sanctions on Iraq in 1991, it grouped together nuclear, biological, and chemical weapons with ballistic missiles, calling on Iraq to relinquish them. The international community has targeted other countries for their pursuit of ballistic missiles or export of materials supportive of missile programs, imposing sanctions on Iran, North Korea, China, Syria, and others. In June 2010, for example, the UN Security Council passed resolution 1929, calling on all states to prevent the transfer to Iran of ballistic missiles or any technology or assistance related to ballistic missiles. To constrain the proliferation of ballistic missiles, several dozen countries participate in the Missile Technology Control Regime (MTCR) to control the spread of technologies that might support missile programs.

This attention from policy makers begs an underlying question: How dangerous are ballistic missiles? Does the spread of ballistic missiles make international conflict more likely? If missile proliferation destabilizes the international system, then this attention to missile proliferation is justified and should continue. However, if the spread of missiles does not increase or possibly even decrease the threat of international conflict, then the costs of these policy efforts may not be worth bearing.

Surprisingly, there is very little scholarship of any kind, theoretical, qualitative empirical, or quantitative empirical, that assesses whether ballistic missiles make international conflict more likely. This is in contrast to the vast literature on the relationships between nuclear weapons and conflict (recent quantitative scholarship includes Gartzke and Jo [2009], Rauchhaus [2009], Horowitz [2009], Beardsley and Asal [2009a, 2009b], Kroenig and [forthcoming]). In this article, we aim to fill this gap by exploring the relationships between ballistic missiles and conflict. We present some of the first theoretical discussion as to how and under what conditions missiles might affect international stability. We execute quantitative tests on the effects of ballistic missiles on the initiation and escalation of crises among all directed dyads from 1946 to 2007, using probit and rare events logit models. We explore for the possibility of endogeneity bias by estimating a three-equation probit model, endogenizing the decision of each state in a dyad to acquire ballistic missiles. We collected new data on ballistic missile possession, ballistic missile range, missile fueling, and missile basing mode. We also collected new data on crisis initiation, building on the Interstate Crisis Behavior (ICB) data.

Our findings indicate that ballistic missiles increase military striking power, and that they can undermine or bolster deterrence. When potential aggressors possess ballistic missiles, they become significantly more likely to initiate crises, making deterrence more likely to fail. When potential defenders possess ballistic missiles, they become significantly less likely to be crisis targets. We find some consistent effects in the relationship between missile possession and conflict escalation. Using a Heckman model of crisis initiation and conflict, we find that missile possession by a target makes crisis escalation less likely, though missile possession by an initiator does not make crisis escalation more likely. We also find some complex interactive effects between ballistic missiles and nuclear weapons. Specifically, ballistic missiles make nuclear and nonnuclear states similarly more likely to initiate. Ballistic missiles confer deterrence advantages for potential nonnuclear targets but undermine deterrence for potential nuclear targets. We find that solid-fueled missiles are more likely than liquid-fueled missiles to deter crisis initiation, but that more survivable-based missiles are not more likely to deter crisis initiation than less survivable-based missiles.

This article proceeds in four parts. The next section presents theory and hypotheses connecting ballistic missiles and international conflict. The following section develops the research design. The next section presents and discusses the quantitative results and is followed by the Conclusion.

The Military Capabilities and Political Effects of Ballistic Missiles

A ballistic missile is an airborne projectile that can deliver conventional, nuclear, biological, or chemical ordnance. It is initially powered by rockets before it begins to follow an unpowered, free-falling trajectory toward its target.¹ Ballistic missiles were first used in combat by Germany against Britain in 1944–1945, killing about 9,000 (Pape 1996, 345). In the decades since 1945, many states have acquired and used ballistic missiles. The first post–World War II combat use of ballistic missiles occurred during the 1973 Yom Kippur War. Since then, missiles have been used sporadically in about a dozen conflicts.² In perhaps the most consequential combat use of missiles, Iraqi Scuds killed thousands of Iranian civilians during the Iran–Iraq War (Cooper and Bishop 2000, 262).

Today, around thirty countries possess ballistic missiles. Ballistic missiles vary in their characteristics. In contemporary arsenals, they have ranges from around 80 km to more than 10,000 km. They vary in the amount of weight they can carry, from around 150 kg to nearly 9,000 kg.³ Ballistic missiles are distinct from cruise missiles, as the latter function more like unmanned aircraft, making course corrections mid-flight. Few countries possess cruise missiles or, relatedly, unmanned aerial vehicles. The focus here is on missiles capable of delivering ordnance over ranges of at least 150 km. Such longer-range missiles can strike beyond border or combat areas, deeper into enemy territory, or against nonadjacent targets.⁴ The international community frequently focuses on longer-range missiles. For example, the 1991 UN Resolution on Iraq called on Iraq to abandon ballistic missiles with ranges of 150 km or more.⁵

Ballistic Missiles Increase Military Power

Some propose that ballistic missiles increase a state’s military striking power beyond that of other kinds of weapon systems, especially aircraft, the other primary means of delivering military ordnance over longer ranges.⁶ Missiles provide three advantages over aircraft, higher likelihood of reaching target, lower friendly casualty rates, and faster speed. First, a missile is more likely to deliver its payload than is an aircraft largely because antiaircraft technology has been and is more advanced than ballistic missile defense (BMD).⁷ The attrition rate of aircraft in combat is low but nonzero, ranging in the modern era from about 1 percent or 1.5 percent in the Yom Kippur War to 2 percent in the European theater in World War II to 3 percent in the Vietnam War. Antiaircraft defenses can have higher success rates. One World War II bombing raid over Germany suffered 20 percent attrition, and the US Air Force estimated in the 1970s that it might suffer as high as 15 percent attrition if it attacked Soviet airspace (Fetter 1991, 10n-11n).

The impact of antiaircraft capabilities may be underrated by these attrition numbers. Belligerents keep attrition lower by flying aircraft at higher altitudes, out of range of some antiaircraft assets, or by not flying at all when facing robust air defense assets. Aerial missions were reduced or redirected in the face of strong air defenses in the Iran–Iraq War, the 1991 Gulf War, and the 1999 Kosovo War, for example (Cordesman and Wagner 1990, 662; Gordon and Trainor 1995, 81; Byman and Waxman 2000). During the Gulf War, the UN coalition had just under three times as many aircraft as did Iraq but flew more than seventy-five times as many missions. Iraqi aircraft were generally kept grounded, reflecting Iraq’s respect for coalition air defenses (Clodfelter 2008, 633-34; Gordon and Trainor 1995, 224).

In comparison, ballistic missiles are much more likely to reach their targets. The gap between missile defense and air defense capabilities can be quite significant. For example, Syrian aircraft cannot reliably reach Israeli targets but Syrian missiles can (Nolan, 1991, 78). During the 1990–1991 Gulf War, coalition air superiority blocked Iraq from launching air strikes, but coalition BMD was largely unable to prevent Iraqi Scud missiles from striking Israeli cities or an American troop barracks in Saudi Arabia.

Only a few countries have BMD, and even fewer have used BMD in combat. The United States has the most advanced BMD system, and the effectiveness of its BMD is limited. The only American BMD system to see combat use is the Patriot. It enjoyed relatively little success against Iraqi Scuds in the 1991 Gulf War (Postol 1991/1992) but greater success against Iraqi FROG-7, cruise, and other missiles during the 2003 Iraq War, achieving an attrition rate of perhaps 39 percent (Boese 2003). However, the American success in 2003 should not suggest optimism about BMD. The 2003 successes were against shorter-range missiles, and the American system is as yet unproved against medium- or long-range missiles (G. Lewis and Postol 2010).

The lack of effective BMD means that usually missiles will have a higher likelihood of reaching their targets than will aircraft. This provides missiles with an operational advantage and may also give missiles greater psychological impact than aircraft. Even if ballistic missiles deliver the same ordnance as planes, ballistic missiles can be uniquely terrifying. The Israeli Defense Minister Shimon Peres told President John F. Kennedy in 1963, “A conventional warhead is very different from a bomb released from a plane. The main feature of a missile is that it is unmanned. It sows terror and enhances the sense of power of those who employ it, because there are no effective means of defence against it” (quoted in Karp 1996, 48). Greater fear of missiles may enhance their perceived striking power, encouraging missile-possessing states to attack and increasing missile-possessing states’ deterrent power.

A second advantage is that unlike aircraft attacks missile attacks risk no friendly casualties. Pilots can be shot down. In World War II, the US Air Force lost 80,000, many in bombing missions (Clodfelter 2008, 523). Unpiloted missiles risk no friendly casualties.

Third, missiles travel quickly, flying thousands of miles per hour compared to aircraft flying at hundreds of miles per hour (Carus 1990). Shorter flying times might allow a missile-armed initiator to destroy the target’s retaliatory capability, its command and control, or its own missiles or aircraft, before the target could initiate a retaliatory strike (Frye 1992). Lower risks of retaliation make aggression more tempting.

Missiles can strike both military and civilian targets. Against military targets, missiles are perhaps most useful against larger military bases, where even lower accuracy missiles are likely to hit something of value or at least to disrupt base activities. Paul Bracken (1998, 87) expressed the point that “Ballistic missiles are made to destroy bases. They can disarm an opponent before he can move to an offensive position. It is nearly impossible to engage in military operations where incoming warheads are bursting, because of the danger to troops, secondary explosions of ammunition, and the disruption to the maintenance of complex weapons.” And indeed, there are several instances of ballistic missiles being launched in combat against military targets, including Egyptian missile launches against Israeli ground forces during the 1973 Yom Kippur War, Iraqi missile launches against military targets, including an American base, in 1991, and nearly 20 Iraqi missile launches against coalition targets in 2003.⁹ Missiles can also be used against large civilian targets such as cities and industrial complexes. A lower accuracy missile can be effective because there are so many high-value targets within the large area of an urban zone. Further, the psychological impact of missile threats against cities may prove powerful.

If ballistic missiles are more likely to reach their targets, then the possession of ballistic missiles should enhance a state’s offensive military striking power, making it more willing to initiate conflicts that risk escalating to violence. A reduced risk of friendly casualties would also lower the expected costs of initiating conflicts, also tempting aggression. A missile-armed state might become tempted to threaten missile attacks on a target’s cities or military assets in an attempt to compel a target to make concessions (on air and missile strikes as compellence tools, see Pape [1996] and Horowitz and Reiter [2001]). Or, a state might envision missile attacks against a target’s military assets as a means of degrading the target’s military capabilities, making military victory for the initiator more likely. This was the concern expressed in the report of the 1998 Commission to Assess the Ballistic Missile Threat to the United States (the Rumsfeld Commission) that the acquisition of ballistic missiles would encourage rogue states to threaten the American homeland, deployed American forces, and American allies.⁹

Several theories of war forecast that increases in offensive striking power make a state more likely to initiate international conflicts. Deterrence theory predicts that potential attackers are more likely to challenge potential defenders as the potential attackers grow in military power (Jervis 1976; Huth 1988), though there have been debates about the role military technology plays in deterrence success (critics on the role of technology in war onset include Mearsheimer [1983], Biddle [2004], Lieber [2005]). The bargaining model of war forecasts that potential attackers are more likely to make threats if the balance of power shifts in their favor, knowing both that the potential target is more likely to make greater concessions in the face of a threat backed by more military power and that if war does come greater military power gives the attacker a greater chance of winning (Fearon 1995; Reiter 2003).¹¹

Hypothesis 1: States armed with ballistic missiles are more likely to initiate crises.

Increased striking power can benefit potential crisis targets as well as potential crisis initiators. Both deterrence theory and the bargaining model of war forecast that a state might be less inclined to aggression if it fears that the target might launch retaliatory missile strikes. In Glenn Snyder’s (1961) terms, missiles might facilitate “deterrence by punishment,” if a target could use missiles to inflict costs on an attacker especially by attacking cities, or “deterrence by denial,” if a target could use missiles to improve the balance of power by destroying the attacker’s military assets and/or disrupting the attacker’s military operations. American policy makers are concerned about the deterrent power ballistic missiles provide other states. The United States might not be able to exert military pressure on a missile-armed rogue state like North Korea or Iran, because of the possibility of retaliation against the United States or its allies (DeBiaso 2006, 160). In January 2011, Secretary of State Robert Gates expressed grave concern that North Korea’s burgeoning missile and nuclear programs were making it a “direct threat to the United States” (quoted in Bumiller and Sanger 2011). In short, ballistic missiles might make it easier for a state to deter crisis initiation.

Hypothesis 2: States armed with ballistic missiles are less likely to be the target of crisis initiation.

Ballistic Missiles Attract Crisis Initiation

An alternative perspective on missiles is that states that possess them are more likely to be crisis targets. If missiles increase striking power, then states are more likely to consider launching preventive or preemptive strikes against states with missiles. States might also consider initiating crises to compel a missile-armed target to abandon its missile arsenal. The fear that a state might lose its retaliatory capability to a missile strike creates an incentive to launch preemptive or preventive attacks against a missile-armed state (Fetter 1991, 29-30; Nolan 1991, 72-3). Conversely, there might be less fear of a potential initiator armed only with aircraft, as if an initiator launched its aircraft, there would be time for the target to retaliate before the initiator’s aircraft delivered their ordnance. Put differently, because ballistic missiles favor first strikes, they may exacerbate spirals of hostility and the security dilemma (Jervis 1976), and create commitment problems which foster preventive and preemptive motives for war (Fearon 1995).

In the past, ballistic missile programs have created international instability and crises. The rocket launch of the Sputnik satellite demonstrated Soviet long-range missile capabilities and sparked great fear in the West. The Soviet plan to deploy missiles to Cuba caused the 1962 Cuban Missile Crisis. More recently, North Korea’s 1998 missile test sparked an angry reaction from Japan and increased American anxiety over the North Korean nuclear threat (Mistry 2003, 12). That same year, the UN coalition forces launched attacks against Iraqi missile-related sites (Arkin 1999) And, in 2006, a North Korean missile test caused an international crisis.

Hypothesis 3: States armed with ballistic missiles are more likely to be crisis targets.

Not all missiles are equally likely to encourage preemptive or preventive strikes. One aspect of missiles affecting whether or not they will attract preventive or preemptive attacks is the mode of fueling. If a missile is liquid fueled, then for safety reasons the fuel is usually stored outside the missile. When a decision is made to launch the missile, it may take hours before the missile is fueled and ready. During that fueling period, there is an incentive for a threatened state to launch a preemptive attack against the missile before it is launched. Conversely, solid-fueled missiles are always fueled and hence can be fired very quickly after the launch order is given. Because they present less-attractive temptations for preventive or preemptive strikes, solid-fueled missiles may be more effective deterrents than liquid-fueled missiles (e.g., Wohlstetter 1959, 218).

Hypothesis 4: Among states with ballistic missiles, those armed with liquid-fueled missiles are more likely to be a crisis target than are those armed with solid-fueled missiles.

Likelihood of targeting can also be affected by mode of missile basing. Submarine-based missiles are less likely to be found, and missiles based on underground silos are less likely to be destroyed if found and attacked. As an attack on missiles becomes less likely to succeed, potential attackers become less motivated to target states with missiles (Jervis 1989).

Hypothesis 5: Among states armed with ballistic missiles, those with more survivable modes of missile basing are less likely to be crisis targets than those with less survivable modes of missile basing.

Ballistic Missiles Have No Effect on International Crisis Onset

Conversely, some argue that ballistic missiles are militarily inconsequential and ought not affect crisis initiation. Though ballistic missiles have some advantages over aircraft in terms of speed and likelihood of reaching their targets, aircraft have their own advantages. Munitions delivered are often more accurate if launched by aircraft rather than missiles, especially if aircraft can launch air-to-ground missiles that have targeting systems. Aircraft can often carry higher payloads than ballistic missiles. Aircraft may have higher rates of reliability, and aircraft may require less pretarget intelligence because pilots can adjust for target location errors. And, unlike an aircraft, a missile cannot be reused.

When missiles have been used in the past they have had little effect on combat or war outcomes. Aaron Karp (1996, 44) stated that “In no case were missile attacks alone enough to change the course of a war, although some were more consequential than others.” Janne Nolan (1991, 187n-88n) agreed: “Most Western analysts have concluded that missile strikes were not decisive in any of these actions, either in securing clear military advantages or in hastening the end of the war.” Ballistic missiles are on balance not a clearly preferred means for delivering conventional munitions, meaning that their possession might not substantially increase striking power or make conflict more likely (Nolan 1991, 71). Therefore, if missiles do not substantially increase a state’s level of combat performance, they might not make a state more willing to initiate a crisis or bolster a potential target’s deterrent power.

Research Design

Data

We examine the effects of ballistic missiles on the likelihood of conflict within a directed dyad for all directed dyads in the international system. We focus on the period 1946–2007, using the revised Gleditsch and Ward’s (1999) list of nation states.¹² We use directed dyad years, in which a pair of countries X and Y in a particular year Z generates two cases, X-Y in year Z and Y-X in year Z, in which the first named state is the potential challenger and the second named state is the potential defender. Before losing cases because of missing data, there are 1,404,410 cases.

Our dependent variable is the initiation of an international crisis. We start with the list of crises provided by the ICB project, version 10.¹³ ICB defines the two conditions of an international crisis as “(1) a change in type and/or an increase in intensity of disruptive, that is, hostile verbal or physical, interactions between two or more states, with a heightened probability of military hostilities; that, in turn, (2) destabilizes their relationship and challenges the structure of an international system—global, dominant, or subsystem” (Brecher and Wilkenfeld 1997, 4-5). ICB does not identify conflictual dyads within its crises, nor does it identify the challenger or initiator.¹⁵ We identified all the conflictual dyads, 374 in total, within all ICB crises from 1946 to 2007 (about 0.027 percent of all directed dyads over this period). We then identified the challenger in each of these dyads. We deemed the challenger to be the state that made the first threat, mobilized its forces first, or used violence first. If a substate actor initiated an act of violence against one state and that state retaliated against another state (perhaps by launching raids on the substate actor’s bases in the second state’s territory), we almost always coded the first state as challenger, even if there were suspicions that the second state was providing support to the substate actor. If one state challenged a second state, and a third state threatened, mobilized against, or attacked the first state, we also listed a dyad of the third state challenging the first state. We dropped ICB crises, which included only a single state and which were intrawar crises.¹⁶

For our principle independent variable, we coded whether each state in the dyad possessed ballistic missiles with range of at least 150 km. We used several sources to code the ballistic missile variable, including the Nuclear Threat Initiative (http://www.nti.org/e_research/profiles/), the Arms Control Association (http://www.armscontrol.org/factsheets/missiles), the Center for Defense Information (http://www.cdi.org/nuclear/database/nukestab.html), the Center for American Progress (http://www.americanprogress.org/issues/2007/05/ballistic_missile_tables.html), various issues of the Military Balance, and other sources. Table 1 lists the countries that possess ballistic missiles and the year in which they acquired them. We coded mode of fueling, developing a binary variable coded 1 if a missile-armed states possessed any solid-fueled missiles. We also developed a binary variable for survivability for missile-armed states and coded 1 if a missile-armed state had any missiles based on underground silos or on submarines.

Table 1.

Dates of Ballistic Missile Acquisition.

Country	Current BM capability	Since	Previous BM capability	Since
United States	ICBM	1959	IRBM	1957^a
China	ICBM	1978	IRBM	1972^a
Russia/Soviet Union	ICBM	1957	MRBM	1955^a
France	ICBM	1985	IRBM	1977^a
United Kingdom	ICBM	1994	IRBM	1967
India	MRBM	2004	SRBM	1994
Iran	MRBM	2003	SRBM	1985
North Korea	MRBM	1993	SRBM	1969
Pakistan	MRBM	1998	SRBM	1992
Israel	MRBM	1972	None
Saudi Arabia	MRBM	1986	None
Afghanistan	SRBM	1988	None
Armenia	SRBM	1994	None
Bahrain	SRBM	2001	None
Belarus	SRBM	1991	None
Egypt	SRBM	1968	None
Greece	SRBM	1996	None
Iraq	SRBM	1974	None
Kazakhstan	SRBM	1991	None
Libya	SRBM	1976	None
South Korea	SRBM	1978	None
Syria	SRBM	1974	None
Taiwan	SRBM	1983	None
Turkey	SRBM	1996	None
Turkmenistan	SRBM	1991	None
Ukraine	SRBM	1991	None
Vietnam	SRBM	1999	None
Yemen	SRBM	1979	None
West Germany^b	SRBM	1962–1988	None

Note. ICBM = intercontinental ballistic missiles, range around 12, 000 km and payload up to 2,000 kg; IRBM = intermediate-range ballistic missiles, range around 5,000 km and payload up to 1,000 kg; MRBM = medium-range ballistic missiles, range around 2,000 km and payload up to 1,000 kg; SRBM = short-range ballistic missiles, range no more than 500 km and payload up to 1,000 kg.

^aThese four countries had previous ballistic missile capabilities. The United States acquired SRBM capability in 1951, and the Soviet Union acquired it in 1947. China acquired MRBM and SRBM capability in 1966 and 1960, respectively. France acquired MRBM capability in 1971.

^bWest Germany had a few Pershing missiles (supplied by the United States) dismantled in 1988.

Ballistic missiles have finite ranges. We need information on whether the missiles possessed by one state in a dyad had the range to strike the other state in the dyad. The optimal way of accomplishing this task would be to include data between the closest point of launch a missile could be placed on a potential attacker’s territory, and the nearest target on the potential target’s territory (such as the nearest major city or military base). We unfortunately do not have such data for all pairs of states. As a proxy, we instead use data recording the minimum distance between all states in the international system, sometimes called the Cshape data. Minimum distance is the shortest distance between the territorial borders of states in a dyad (Weidmann, Kuse, and Gleditsch 2010). Among all directed dyad years in our temporal range, the potential challenger had ballistic missiles in the range of the potential defender in 2.61 percent of cases (36,598 directed dyad years).

A possible alternative distance measure is capital-to-capital distance. There are several reasons to prefer using minimum distance rather than capital-to-capital distance. For the potential initiator, there is no strategic or political reason to locate one’s ballistic missiles in the national capital. The potential initiator would want to move the missiles as close to the potential target as possible to maximize accuracy, and the attacker’s border is closer to the target than its national capital. Most ballistic missiles, especially shorter-range missiles, are mobile, meaning that they can be kept farther back on the attacker’s territory and out of harm’s way when not in use, and are moved closer to the adversary for launch. The difference between capital-to-capital distance and minimum distance is consequential. For example, during the 1991 Gulf War, the Iraqi al-Husayn missile could (and did) reach targets in Israel when launched from sites in Western Iraq but not when launched from Baghdad (Nolan 1991, 75).

As far as the potential target, a capital city is of course a tempting target for missile attack, but it is not the only target. Other cities can be targeted, as can large military bases. Perhaps surprisingly, there have been almost no examples in which ballistic missiles used in interstate conflicts were launched against the target’s capital city, other than some German attacks in World War II and some strikes in the Iran–Iraq War. Policy makers sense a threat when a potential attacker’s ballistic missiles can reach its territory, even if a missile cannot reach its capital. The Soviet Union was sufficiently concerned about the Chinese nuclear threat during the 1969 missile crisis to consider launching a preemptive attack against Chinese nuclear facilities, though China did not yet have a missile with sufficient range to strike Moscow from Beijing (Betts 1987, 79-81). The United States became much more concerned about the North Korean missile threat when the 1998 North Korean missile tests demonstrated that the North Korean Taepo-Dong 2 missile, if configured with a two-stage system, would have enough range to reach the Western edges of the territorial United States, including Alaska, Hawaii, and the west coast of the contiguous 48 states, though not Washington, D.C., see National Intelligence Council (2001, 9) The United States also became concerned when Chinese missiles (specifically, the CSS-4) acquired the ability to reach the western fringes of American territory, see Department of Defense (2007, 18-20)

Distance played a role in three of our independent variables. First, we coded the ballistic missile variable as zero if a state had ballistic missiles, but the minimum distance between it and the other dyad member was greater than the range of the state’s missiles. Second, we included as a control the square root of the minimum distance between the states, proposing that more distance between states means a lower chance of conflict. We also included a contiguity measure coded as 1 if the states are less than 10 km apart and 0 otherwise, as many interstate disputes emerge from border conflicts.

We included a measure of the power parity between the two states, following previous research that has found that states which are closely matched in power are more likely to come into conflict. We developed a variable ranging from 0 to 1, with higher scores representing greater power parity between states. For our power measure we used Correlates of War (COW) Composite Index of National Capability (CINC) scores scores from the COW National Material Capabilities data set, version 4.0.

Major powers are more likely to get involved in disputes because they tend to have global interests while minor powers do not. We used a dummy variable for major power status, using COW data.

There is substantial research finding that democracies are less likely to fight each other. We coded for democracy, using Polity IV data on regime type. If a state’s −10 to +10 democracy score is 7 or higher, we consider it a democracy. Our variable is a dummy variable and coded 1 if both states in a dyad are democratic and 0 otherwise.

We coded two dummy variables for the presence of nuclear weapons, one for each member of the dyad.¹⁹

Research Design

With a binary dependent variable, we need an appropriate estimator, such as probit. We suspect temporal autocorrelation within dyad years, so we use robust standard errors, clustered on the dyad and reverse dyad (that is, for an A-B dyad we cluster on A-B and B-A). We also employ the Carter and Signorino (2010) approach of including variables for the number of peace years (years since last crisis within the dyad), the number of peace years squared, and the number of peace years cubed.

A more nettlesome issue is the potential problem of endogeneity. Ballistic missiles are not randomly assigned to states in the international system, and states may choose to acquire ballistic missiles for reasons related to their propensity for conflict. This endogeneity may threaten the inferences we draw in a conventional probit framework. To address this concern, we employ a three-equation probit model, simultaneously endogenizing the decision of a potential initiator to initiate a crisis, the potential initiator’s acquisition of ballistic missiles, and the potential defender’s acquisition of ballistic missiles. The model is roughly similar to seemingly unrelated regression. It uses simulated maximum likelihood, specifically the Geweke–Hajivassiliou–Keane simulator (see Greene 2003, 931-33; Cappellari and Jenkins 2003).²⁰

To endogenize the acquisition of ballistic missiles, we elected not to rely directly on the findings of Barkley (2008).²¹ We use five regressors predicting the potential challenger’s and defender’s acquisitions of ballistic missiles.²² The first regressor concerns the MTCR. As mentioned, the MTCR is intended to slow the spread of ballistic missiles through monitoring and sometimes restricting international trade, especially the trade of dual-use technology. Some hypothesize that a more powerful MTCR will be more effective in constraining missile proliferation, in particular by controlling the spread of dual-use technology (Mistry 2003). We use Barkley’s (2008) measure of the fraction of the world gross domestic product (GDP) accounted for by MTCR members (though we collected our own data for this measure using www.mtcr.info/english/index.html for MTCR membership and World Development Indicators [http://data.worldbank.org/] for GDP data).

The second regressor is the regional ballistic missile threat. Nations often acquire ballistic missiles in reaction to their neighbors’ acquisition of ballistic missiles, in the classic spiral model dynamic of the arms race (Jervis 1976). We built a variable which counts the number of neighbors which have ballistic missiles which have the range to strike the territory of the potential challenger in one equation and the potential defender in the other equation, lagged by 1 year. We collected data on ballistic missile range ourselves. We used the Cshape minimum distance data.

The third variable is economic development. More economically developed states are more likely to have the scientific and industrial base to develop ballistic missiles. We use logged per capita energy consumption of the potential challenger in the potential challenger equation and of the potential defender in the potential defender equation. This measure is comparable to logged GDP/capita (Jackman 1973).

The fourth variable is major power status. Major powers are more likely to need the ability to project their power globally, and the longer distances of ballistic missiles enable power projection. We used the major power status of the potential challenger in the potential challenger equation, and the major power status of the potential defender in the potential defender equation.

The fifth is whether the state has nuclear weapons. Nuclear-armed states often desire to use ballistic missiles as nuclear delivery vehicles. We use the nuclear weapons status of the potential challenger in the potential challenger equation, and the nuclear weapons status of the potential defender in the potential defender equation.

Empirical Results

We first present the empirical results of a conventional probit analysis, in model 1 of Table 2. The analysis reveals that if state A, the potential challenger, possesses ballistic missiles, it is significantly more likely to initiate a challenge against a potential defender, state B. Further, if state B possesses ballistic missiles, then the potential challenger is significantly less likely to initiate a challenge. These results provide support for Hypotheses 1 and 2 and do not support Hypothesis 3. Many of the results for the control variables are significant and in the expected direction. However, the effect for the nuclear weapons status of the potential defender is statistically significant but not in the anticipated direction. The finding that nuclear weapons encourage the initiation of crises is consistent with other studies that find nuclear weapons encourage crisis initiation (Beardsley and Asal 2009b; Rauchhaus 2009). It is also consistent with the finding that nuclear-armed states are more likely to win their crises, in that nuclear states may be more likely to initiate crises because they know they are more likely to win them (Beardsley and Asal 2009b).

Table 2.

Effects of Ballistic Missiles on Conflict Initiation, 1946–2007.

	Model 1 Probit	Model 2 Three equation probit	Model 3 Rare events logit
Challenge DV
Distance	−.0140***	−.0140***	−.0509***
Distance	(.00174)	(.00174)	(.00407)
Contiguity	.357***	.358***	.780***
Contiguity	(.0817)	(.0819)	(.179)
Power parity	.153*	.143*	.454**
Power parity	(.0798)	(.0807)	(.173)
Major power, state A	.0973	.0976	.246
Major power, state A	(.138)	(.138)	(.291)
Joint democ.	−.172**	−.177**	−.433**
Joint democ.	(.0696)	(.0694)	(.151)
Nuclear, state A	.358**	.347**	.999**
Nuclear, state A	(.136)	(.138)	(.329)
Nuclear, state B	.589•	.555•	1.62•
Nuclear, state B	(.0969)	(.100)	(.257)
Ballistic missiles, state A	.353***	.366***	1.01***
Ballistic missiles, state A	(.0770)	(.802)	(.204)
Ballistic missiles, state B	−.231*	−.180*	−.831**
Ballistic missiles, state B	(.103)	(.107)	(.264)
Constant	−2.488***	−2.47***	−4.78***
Constant	(.0911)	(.0918)	(.218)
Ballis. msls. state A DV
MTCR	–	−.0522	–
MTCR		(.0503)
Neighbors of state A with missiles	–	.188***	–
Neighbors of state A with missiles		(.00878)
Development, state A	–	.143***	–
Development, state A		(.0181)
Major power, state A	–	1.87***	–
Major power, state A		(.0540)
Nuc. weapons, state A	–	1.79***	–
Nuc. weapons, state A		(.0472)
Constant	–	−4.09***	–
Constant		(.0783)
Ballis. msls. state B DV
MTCR	–	−.0520	–
MTCR		(.0501)
Neighbors of state B with missiles	–	.187***	–
Neighbors of state B with missiles		(.00874)
Development, state B	–	.143***	–
Development, state B		(.0181)
Major power, state B	–	1.87***	–
Major power, state B		(.0540)
Nuc. weapons, state B	–	1.79***	–
Nuc. weapons, state B		(.0473)
Constant	–	−4.08***	–
Constant		(.0786)
Rho21	–	−.00800	–
Rho21		(.00984)
Rho31	–	−.00286**	–
Rho31		(.0104)
Rho32	–	.298***	–
Rho32		(.0363)
Log like.	−2461.9649	−95235.463	–
R ²	.2771	–	–
n	1,295,374	1,280,930	1,295,374

Note. Results for the three peace-year variables not shown. *sig at .05 level; **sig at .01 level; ***sig at .001 level; •sig at .05 level, but not in predicted direction. All sig tests are one-tailed. Standard errors are robust for models 1 and 2, clustered on dyad.

We next explore for the possibility of endogeneity bias by employing a three-equation probit model, endogenizing the potential challenger’s decision to challenge, the acquisition of ballistic missiles of state A (the potential challenger), and the acquisition of ballistic missiles of state B (the potential defender). These results are displayed in model 2 of Table 2. They indicate that even when endogenizing the acquisition of ballistic missiles by both states in a dyad, the direction, statistical significance, and magnitude of the effect of ballistic missile possession by both potential challenger and potential defender on crisis initiation remain similar. The MTCR strength is not correlated with the likelihood of a state acquiring ballistic missiles, a result in opposition to previous findings (Barkley 2008).

Reassured that the results in model 1 are not biased by endogeneity, we now turn to another potential threat to inference, the lopsided distribution of the dependent variable, a very low percentage of ones. To account for potential bias introduced by such a skewed distribution, we employ rare events logit (Tomz, King, and Zeng 2003), reporting those results in model 3 of Table 2. The results do not change from those reported in model 1 or 2. The effects of missile possession on crisis initiation are substantively significant, as well. Holding the other variables at their means and modes, a state with ballistic missiles is 266 percent more likely to initiate a crisis as compared to a state without ballistic missiles. Further, a state is 67 percent less likely to initiate a crisis against a potential defender with ballistic missiles, as compared to a potential defender with ballistic missiles.

In the previous section, we argued that minimum distance is a more accurate proxy than capital-to-capital distance. As we expect, if we use a capital-to-capital distance measure in the Table 2 models, the two independent variables regarding missiles become statistically insignificant.²³ Closer examination suggests that the key reason for this difference in results is probably because there are 51 dyad years of crisis initiation in which the minimum distance measure codes the challenger as being within range of the defender, but the capital-to-capital measure codes the challenger as not being in range. However, in many of these cases, the challenger could have launched missiles from its territory against a strategic target in the defender’s territory, consistent with the coding of the minimum distance measure but not consistent with the capital-to-capital distance measure. For example, there are two India–Pakistan crises in which India had missiles, and though Islamabad is out of missile range from New Delhi, the Pakistani city of Lahore (population 6 million) is less than 50 km from the Indo-Pakistani border. There are three Iraq–Kuwait crises, and though the Kuwaiti capital is out of range from Baghdad, it is less than 100 km from the Iraq–Kuwait border. In the Turkey–Cyprus crisis, the Cypriot capital of Nicosia is out of missile range from the Turkish capital of Ankara, but it is in missile range (less than 150 km) from the south Turkish coast. There are two Iraq–Syria crises, and though Iraqi missiles cannot reach Damascus from Baghdad, they can reach Damascus from the Iraqi–Syrian border. There are two older, China–India crises, and when they occurred Chinese missiles could not reach New Delhi from Beijing, but they could reach New Delhi from the Sino-Indian border. In the Turkey–Syria crisis, missiles launched from Ankara could not reach Damascus, but missiles launched from Turkish territory could reach Syria’s largest city, Aleppo (population above 2 million). That is, there are several of these important cases in which crisis initiation occurred, minimum distance is a more accurate proxy for missile range than capital-to-capital distance, suggesting that the results produced with the minimum distance data are preferred.

Nuclear Weapons and Ballistic Missiles

There may be an interactive effect between ballistic missiles and nuclear weapons. The policy community is concerned that ballistic missiles armed with nuclear weapons are especially dangerous and that a state armed with such a combination might be more willing to initiate conflict.²⁴ One reason for this concern is that for nuclear missions, the operational qualities of ballistic missiles may indicate some advantages over aircraft. The lower accuracy of ballistic missiles becomes much less important for nuclear operations, as missing the target by dozens or even hundreds of meters can still mean that the nuclear mission (perhaps of destroying a city) can be accomplished, because nuclear weapons are so powerful. The atomic bomb dropped on Nagasaki missed its target by 1.5 miles but still accomplished the mission of killing tens of thousands of civilians and helping push Japan to surrender. Also, the likelihood of reaching the target becomes increasingly important for nuclear weapons. Most states have very few nuclear weapons, and if a leader decides to use a precious nuclear weapon, she would want to be highly confident that once launched the nuclear warhead would arrive at the target. A missile, once launched, is almost assured of not being intercepted. Conversely, launched aircraft have a nontrivial probability of being destroyed. Last, the lack of missile reusability becomes less of an issue for nuclear missions. That said, others have argued against the proposition that ballistic missiles make nuclear weapons especially dangerous, arguing that aircraft present advantages for executing nuclear missions such as higher reliability (Rubin 1991).

We tested for the possibility of a Ballistic Missiles–Nuclear Weapons interaction by analyzing a rare events logit model that explored the interactive effects between ballistic missiles and nuclear weapons. Specifically, there are four possible conditions each for challenger and defender: the state (whether challenger or defender) has missiles and nuclear weapons; it has missiles and no nuclear weapons; it has no missiles and nuclear weapons; and it has neither missiles nor nuclear weapons. We included three dummy variables each for challenger and defender to represent this array of states, excluding the condition of no missiles/no nuclear weapons for challenger and defender. In all cases, when a nuclear weapons state had missiles, the missiles had sufficient payload to deliver a nuclear weapon. That is, there were no cases of a state having nuclear weapons and ballistic missiles lacking the capacity to deliver a nuclear warhead.²⁵ We provide the results of this analysis in model 4 of Table 3.

Table 3.

Rare Event Logit Analysis of Interactive Effects of Ballistic Missiles and Nuclear Weapons on Crisis Initiation, 1946–2007.

	Model 4	Model 5	Model 6
Distance	−0.0517*** (0.00413)	−0.0508*** (0.00406)	−0.503*** (0.00404)
Contiguity	0.747*** (0.180)	0.773*** (0.179)	0.782*** (0.181)
Power parity	0.474** (0.175)	0.442** (0.173)	0.440** (0.171)
Major power, state A	0.361 (0.270)	0.266 (0.291)	0.327 (0.307)
Joint democracy	−0.438* (0.151)	−0.428** (0.151)	−0.435** (0.154)
Nucs/missiles, state A	1.87*** (0.275)	---	---
Nucs/no missiles, state A	1.80*** (0.350)	---	---
No Nucs/Missiles, State A	1.20*** (0.183)	---	---
Nucs/missiles, state B	0.645*** (0.195)	---	---
Nucs/no missiles, state B	2.12 (0.247)	---	---
No nucs/missiles, state B	−0.430* (0.240)	---	---
Missiles, state A	---	0.986*** (0.205)	0.964*** (0.199)
Missiles/liquid, state B	---	−0.572* (0.284)	---
Missiles/solid, state B	---	−0.805** (0.301)	---
Missiles/survivable, state B	---	---	−0.141 (0.227)
Missiles/vulnerable, state B	---	---	−0.528** (0.199)
Nuclear, state A	---	0.993*** (0.327)	1.03*** (0.321)
Nuclear, state B	---	1.55*** (0.251)	1.13*** (0.169)
Constant	−4.81*** (0.290)	−4.78*** (0.218)	−4.78*** (0.217)

Note. Results for the three peace year variables not shown.

*sig at .05 level. **sig at .01 level. ***sig at .001 level. All significance tests are one-tailed.

The results are interesting. Potential challengers are more likely to initiate crises when armed with missiles and nuclear weapons, missiles alone, or nuclear weapons alone, as compared to potential challengers armed with neither missiles nor nuclear weapons. Chi-square tests revealed that there is no significant difference between the Nucs/Missiles coefficient and the Nucs/No Missiles coefficient. That is, missiles do not make a nuclear-armed state more likely to initiate a crisis. However, the Nucs/Missiles coefficient is larger than the No Nucs/Missiles coefficient, and a chi-square test reveals that the difference has statistical significance (prob > χ² = .0256). Together, these findings indicate that nuclear weapons make both missile-armed and nonmissile-armed states more likely to initiate, missiles make conventionally armed states more likely to initiate, but missiles do not make nuclear-armed states more likely to initiate.

The results on the potential defender (state B) variables are also interesting. Among states without nuclear weapons, states with missiles are not significantly more likely to be targeted as compared to states without missiles. Among states with nuclear weapons, missiles make a state significantly less likely to be targeted. This finding argues against the simple existential deterrence proposition that the possession of any nuclear weapons is sufficient to deter (see J. Lewis 2008; Kroenig forthcoming), though interestingly nuclear-armed states with and without missiles are more likely to be targeted than states without nuclear weapons or missiles. Oddly, missile-armed states are more likely to be targeted if they have nuclear weapons then if they do not have nuclear weapons.

The findings in Table 3 provide nuance to the results in Table 2. One puzzle is that there is some inconsistency of effects across initiation and targeting. The initiation results indicate that missiles do not make nuclear-armed states significantly more likely to initiate, implying that missiles do not boost a nuclear state’s striking power or the credibility of its threats. The expectation would be that missiles would not make nuclear-armed states less likely to be targeted, if missiles do indeed fail to boost a state’s credibility or striking power, but the findings indicate that among states with nuclear weapons those with missiles are significantly less likely to be targeted (regarding the likelihood that those two coefficients are different, prob > χ² = .000). Further, for conventionally armed states, missiles do make a state more likely to initiate, implying that missile-armed states ought to be less likely to be targeted. But the findings indicate that among conventionally armed states, those armed with missiles are less likely to be targeted, evidence in support of Hypothesis 3. But among missile-armed states, those armed with nuclear weapons are more likely to be targeted, and those without nuclear weapons are less likely to be targeted. There is not a clean theoretical account for this array of diverging findings.

We are also interested in the possible effects of fueling and basing modes. We explore these possibilities in models 5 and 6 in Table 3. In model 5, we return to the rare events logit model in Table 3 but distinguish between targets armed with missiles fueled exclusively with liquid fuel, and those that possess at least some missiles with solid fuel. The analysis reveals that potential targets armed with solid-fueled missiles are less likely to be targeted than potential targets armed with liquid-fueled missiles only. Further, t tests indicate that the difference between the Missile/Liquid and Missile/Solid coefficients is statistically significant. This provides support for Hypothesis 4. In model 6, we test the hypothesis that potential targets armed with missiles based on more survivable modes are less likely to be targeted than potential targets armed with missiles in less survivable modes. The results do not support Hypothesis 5, as states armed with survivable missiles are not significantly less likely to be targeted than states without missiles. Interestingly, states armed with less survivable missiles are significantly less likely to be targeted than states without missiles, or states with survivable missiles.

Thus far, the discussion has focused on the effects of missiles on crisis initiation. A separate question is whether missiles have any effect on crisis escalation. A thorough understanding of the effects of missiles on crisis escalation can probably be best reached through a formal model of both crisis initiation and crisis escalation. However, we can speculate that the same forces that operate in the initiation of crisis might also operate in the escalation of crisis. That is, if potential initiators armed with missiles are more likely to initiate crises, they might be more likely to escalate crises. And, if potential targets armed with missiles are less likely to be targeted, then perhaps crises involving targets armed with missiles are less likely to escalate.

We test this speculation with a Heckman probit model, simultaneously analyzing the initiation of a crisis and the escalation of a crisis. For crisis escalation, we use the ICB VIOL variable, coded 1 if a crisis experiences no clashes, 2 if a crisis experiences minor clashes, 3 if a crisis experiences major clashes, and 4 if a crisis experiences war. We convert that variable into a dichotomous variable, coded 1 if VIOL is 3 or 4. For the selection model, we use the same variables as used in models 1 and 3 in Table 1. For the model predicting escalation, we use variables measuring joint democracy, power parity, nuclear weapons for each side, ballistic missiles for each side, and the ICB variable GRAVCR, which measures the gravity (or stakes) of the crisis on a 0–6 scale.²⁶

In Table 4, we present the analysis of the Heckman probit model. We find that if the crisis initiator has missiles, it does not make escalation significantly more or less likely. However, consistent with the findings in Tables 2 and 3, if the crisis target has missiles, it makes escalation less likely. That is, when the potential target has missiles, crisis initiation is less likely and (once it begins) a crisis is less likely to escalate to major clashes or war. These results hold if we instead run a Heckman regression analysis using the 1–4 categorization as the escalation-dependent variable. They also hold if we run a simple probit regression of just those dyads that experienced crisis. But to reiterate, these results are only suggestive, as a full understanding of the effects of missiles on crisis escalation would likely require the development of a formal model of the effects of missiles on crisis initiation and escalation.

Table 4.

Heckman Probit Analysis of Crisis Initiation and Escalation.

	Model 7
Escalation model
Gravity	.420***
Gravity	(.0530)
Ballistic missiles, state A	−.172
Ballistic missiles, state A	(.224)
Ballistic missiles, state B	−.894**
Ballistic missiles, state B	(.285)
Nuclear weapons, state A	−.153
Nuclear weapons, state A	(.226)
Nuclear weapons, state B	.281
Nuclear weapons, state B	(.285)
Power parity	.284
Power parity	(.274)
Joint democracy	−.216
Joint democracy	(.221)
Constant	.995*
Constant	(.493)
Initiation model
Distance	−.0140***
Distance	(.000956)
Contiguity	.357***
Contiguity	(.0535)
Power parity	.153**
Power parity	(.0629)
Major power status, state A	.0952
Major power status, state A	(.0865)
Joint democracy	−.172***
Joint democracy	(.0512)
Ballistic missiles, state A	.354***
Ballistic missiles, state A	(.0652)
Ballistic missiles, state B	−.232***
Ballistic missiles, state B	(.0748)
Nuclear weapons, state A	.360***
Nuclear weapons, state A	(.0895)
Nuclear weapons, state B	.589***
Nuclear weapons, state B	(.0723)
Constant	−2.49***
Constant	(.0664)
Rho	−.111
Rho	(.140)
Log likelihood	−2674.402

Note. Likelihood ratio test of independence (rho = 0): chi-square = .61, prob > chi-square = .4330.

Conclusion

This article has provided the first empirical tests of the effects of ballistic missiles on international conflict. It found that a state armed with ballistic missiles is significantly more likely to initiate an interstate crisis against another state, and that states armed with ballistic missiles are significantly less likely to be the target in an interstate crisis. It also found that solid-fueled missiles are more effective deterrents than are liquid-fueled missiles, though states with missiles based on more survivable basing modes are not less likely to be the target of crisis initiation than are states with missiles based on less survivable basing modes. There are also interesting interactive effects between ballistic missiles and nuclear weapons, as missiles make crisis initiation more likely only among nonnuclear states, and missiles make a state less likely to be targeted only among nuclear states. Crises involving targets armed with ballistic missiles are less likely to escalate.

The findings shed light on whether missiles are uniquely dangerous weapons. The results here indicate that missiles are at least perceived to have unique operational capabilities, as missile-armed states are more likely to initiate international crises than are states without missiles, and missile-armed states are less likely to be targeted. Consistent with this speculation, states armed with solid-fueled missiles are more likely to deter conflict initiation than states armed with liquid-fueled missiles. Unexpectedly, however, states armed with submarine- or silo-based missiles did not enjoy greater deterrence advantages compared with states armed with more vulnerable missiles.

These findings have other implications. Many prominent theories of conflict onset, such as deterrence, the security dilemma, the offense–defense balance, the power transition, and others propose that the nature of a state’s military affects the likelihood of conflict onset. Though many studies routinely emphasize the size of a state’s military and some studies focus on military strategy, relatively few focus on technology, the actual weapons systems a state has. Indeed, some scholars are critical of the idea that military technology has an independent causal effect on conflict onset. This study has demonstrated that at least one form of military technology, ballistic missiles, has a statistically and substantively significant effect on conflict onset.

The article also speaks to policy makers’ assumptions that ballistic missiles make international conflict more likely. Specifically, the article has mixed implications for this assumption. The findings tend to support the assumption of the policy community that ballistic missiles make a state more likely to initiate international conflict, but they also show that a state is less likely to be targeted if it possesses ballistic missiles. Hence, it is not completely clear whether the spread of missiles would make conflict more, less, or equally likely, given these contrasting effects.

Another area of missile policy is the advisability of the MTCR, assuming that constraining missile proliferation is a policy goal. The analysis does not support the conclusion that the MTCR significantly constrains missile proliferation. The three-equation probit model found that the MTCR had no effect on the acquisition of ballistic missiles. That finding might encourage reconsideration of the MTCR, given that its actions can constrain the development of civilian space programs (Early 2011). Because the rocket technology used to place satellites in orbit can also be used to develop ballistic missiles, the MTCR endeavors to constrain the spread of space programs. Civilian space programs have the potential to accelerate economic growth through the collection of (terrestrial) geological data from space and the development of indigenous telecommunication systems. For example, Argentina and South Africa were forced to abandon their civilian space programs as a condition of joining the MTCR. That being said, some are unconcerned about these costs, both because space programs generate relatively little economic revenue and because barriers to economic development might be avoidable, if as part of an agreement to enter the MTCR other states offered to place an MTCR entrant’s satellites into space on their own rockets, or to share data from their own satellites (Mistry 2003, 170-72).

If these results provide a mixed policy recommendation on ballistic missiles for the international community, they provide a more straightforward recommendation for American foreign policy: ballistic missile proliferation threatens American national interests. Ballistic missiles make states more likely to initiate conflict. This is consistent with the American fear that ballistic missiles might make rogue states more belligerent. Further, ballistic missiles appear to make states less likely to be the targets in international crises. The implication is that the deterrent effect of ballistic missiles might make it harder for the United States to exert political or military pressure on rogue states like North Korea, Iran, and Burma if they acquire ballistic missiles. Further, the findings might support the continued American search for effective BMD. If the United States can deploy missile defense sufficient to thwart smaller inventories of less sophisticated missiles (such as Chinese, Iranian, or North Korean missiles), then this may discourage those states from initiating conflict.

There is an important postscript to the preceding policy discussion. Aside from the effects of missiles on the likelihood of interstate conflict, missiles may make conflict between states and nonstate actors more likely. Some states have acquired ballistic missiles and then disseminated missiles to nonstate actors such as terrorist groups. Hezbollah employed Iranian-provided missiles with a range of about 45 km against Israel in the 2006 Lebanon War, and Syria is suspected of passing ballistic missiles of longer range to them, including Scud-D missiles with a range of 580 km (Gordon and Lehren 2010). If nonstate actors like Hezbollah acquire ballistic missiles of longer ranges, they may become more willing to launch attacks on targets from deeper within the territory of a host state, perhaps far enough away from the border to make retaliatory operations (especially ground operations to capture militants or occupy launching sites) by the targeted state less feasible. Such a prospect demands the attention of the international community.

There are several avenues for future research. Systematic data have not yet been collected on national air forces. With data on a country’s ability to launch airstrikes, especially data on aircraft range, scholars would be able to discern better the precise effect of ballistic missiles in comparison to aircraft. It may be that a state’s striking power is boosted equally with either aircraft or missiles and that would be borne out with data coding the power of a state’s ability to launch air strikes as well as missile strikes. Case studies might also begin to get at whether the operational striking power of missiles is exaggerated, as Peres speculated. If there is an unjustifiably great fear of missiles as compared to aircraft, this could help explain why missiles have significant effects on conflict initiation, given their operational drawbacks. Next, the analysis depends on minimum distance data, a proxy for a measure of the distance between the closest launch point and the closest strategic target. Better data on the true concept of interest rather than minimum distance data will reduce measurement error. Lastly, the uneven interactive effects of nuclear weapons and ballistic missiles demand greater attention. We need a theoretical account and empirical unraveling of why missiles make conventional initiation more likely and nuclear targeting less likely but have no effects on nuclear initiation or conventional targeting.

Footnotes

Acknowledgments

An earlier version of this article was presented at the annual meeting of the Southern Political Science Association in January 2011, and to the political science department at Vanderbilt University in Fall 2010. The authors acknowledge the helpful comments and assistance of Kyle Beardsley, Michael Horowitz, and Drew Linzer. Replication materials are available at .

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Arkin

William M

. 1999. “The Difference Was in the Details.” Washington Post, January 17, B1. Accessed July 5, 2012. http://www.washingtonpost.com/wp-srv/inatl/longterm/iraq/analysis.htm.

Barkley

Daniel

. 2008. “Ballistic Missile Proliferation: An Empirical Investigation.” Journal of Conflict Resolution 52 (June): 455–73.

Beardsley

Kyle

Asal

Victor

. 2009a. “Nuclear Weapons as Shields.” Conflict Management and Peace Science 26 (July): 235–55.

Beardsley

Kyle

Asal

Victor

. 2009b. “Winning with the Bomb.” Journal of Conflict Resolution 53 (April): 278–301.

Betts

Richard K.

1987. Nuclear Blackmail and Nuclear Balance. Washington, DC: Brookings Institution.

Biddle

Stephen

. 2004. Military Power: Explaining Victory and Defeat in Modern Battle. Princeton, NJ: Princeton University Press.

Boese

Wade

. 2003. “Army Report Details Patriot Record in 2003 War.” Arms Control Today, November. Accessed May 20, 2010. http://www.armscontrol.org/act/2003_11/Patriotmissile.

Bracken

Paul

. 1998. “America’s Maginot Line.” Atlantic Monthly, December, 85–93.

Brecher

Michael

Wilkenfeld

Jonathan

. 1997. A Study of Crisis. Ann Arbor: University of Michigan Press.

10.

Bumiller

Elisabeth

Sanger

David E.

. 2011. “Gates Warns of North Korea Missile Threat to U.S.” New York Times, January 12.

11.

Byman

Daniel L.

Waxman

Matthew C.

. 2000. “Kosovo and the Great Air Power Debate.” International Security 24 (Spring): 5–38.

12.

Cappellari

Lorenzo

Jenkins

Stephen P.

. 2003. “Multivariate Probit Regression Using Simulated Maximum Likelihood.” STATA Journal 3 (3): 278–94.

13.

Carter

David B.

Signorino

Curtis S.

. 2010. “Back to the Future: Modeling Time Dependence in Binary Data.” Political Analysis 18 (Summer): 271–92.

14.

Carus

W. Seth

. 1990. Ballistic Missiles in the Third World: Threat and Response. New York: Praeger.

15.

Clodfelter

Micheal

. 2008. Warfare and Armed Conflicts: A Statistical Reference to Casualty and Other Figures, 1500-2000. 3rd ed. Jefferson, NC: McFarland.

16.

Cooper

Tom

Bishop

Farzad

. 2000. Iran-Iraq War in the Air, 1980-1988. Atglen, PA: Schiffer Military History.

17.

Cordesman

Anthony H.

Wagner

Abraham R.

. 1990. The Lessons of Modern War, Volume II: The Iran-Iraq War. Boulder, CO: Westview.

18.

DeBiaso

Peppi

. 2006. “Proliferation, Missile Defense, and the Conduct of Modern War.” Comparative Strategy 25 (3): 157–71.

19.

Department of Defense. 2007. Annual Report to Congress: Military Power of the People's Republic of China. Washington: US Government Printing Office. Accessed July 5, 2012. http://www.defense.gov/pubs/pdfs/070523-China-Military-Power-final.pdf.

20.

Early

Bryan R.

2011. “Sharing Space? Great Powers and the Proliferation of Civil Space Capabilities.” Presented at the 2011 Annual Meeting of the Peace Science Society, Los Angeles, October 14-15.

21.

Fearon

James D.

1995. “Rationalist Explanations for War.” International Organization 49 (Summer): 379–414.

22.

Fetter

Steve

. 1991. “Ballistic Missiles and Weapons of Mass Destruction: What is the Threat? What Should be Done?” International Security 16 (Summer): 5–42.

23.

Frye

Alton

. 1992. “Zero Ballistic Missiles.” Foreign Policy 88 (Autumn): 3–20.

24.

Gartzke

Erik

Dong-Joon

. 2009. “Bargaining, Nuclear Proliferation, and Interstate Disputes.” Journal of Conflict Resolution 53 (April): 209–33.

25.

Gelpi

Christopher

Grieco

Joseph M.

. 2001. “Attracting Trouble: Democracy, Leadership Tenure, and the Targeting of Militarized Challenges, 1918-1992.” Journal of Conflict Resolution 45 (December): 794–817.

26.

Gleditsch

Kristian S.

Ward

Michael D.

. 1999. “Interstate System Membership: A Revised List of the Independent States since 1816.” International Interactions 25:393–413.

27.

Gordon

Michael R.

Lehren

Andrew W.

. 2010. “America Prods and Protests But Can’t Halt Arms Trade.” New York Times, December 7, A1.

28.

Gordon

Michael R.

Trainor

Bernard E.

. 1995. The Generals’ War: The Inside Story of the Conflict in the Gulf. Boston: Little, Brown.

29.

Greene

William H.

2003. Econometric Analysis. 5th ed. Upper Saddle River, NJ: Prentice Hall.

30.

Horowitz

Michael

. 2009. “The Spread of Nuclear Weapons and International Conflict: Does Experience Matter?” Journal of Conflict Resolution 53 (April): 234–57.

31.

Horowitz

Michael

Reiter

Dan

. 2001. “When Does Aerial Bombing Work? Quantitative Empirical Tests, 1917-1999.” Journal of Conflict Resolution 45 (April): 147–73.

32.

Huth

Paul K.

1988. Extended Deterrence and the Prevention of War. New Haven, CT: Yale University Press.

33.

Jackman

Robert W.

1973. “On the Relation of Economic Development to Democratic Performance.” American Journal of Political Science 17 (August): 17–21.

34.

Jervis

Robert

. 1976. Perceptions and Misperceptions in International Politics. Princeton, NJ: Princeton University Press.

35.

Jervis

Robert

. 1989. The Meaning of the Nuclear Revolution. Ithaca, NY: Cornell University Press.

36.

Karp

Aaron

. 1996. Ballistic Missile Proliferation: The Politics and Technics. Oxford, UK: Oxford University Press.

37.

Kroenig

Matthew

. Forthcoming. “Nuclear Superiority and the Balance of Resolve: Explaining Nuclear Crisis Outcomes.” International Organization.

38.

Lewis

George N.

Postol

Theodore A.

. 2010. “A Flawed and Dangerous U.S. Missile Defense Plan.” Arms Control Today (May). Accessed May 20, 2010. http://www.armscontrol.org/act/2010_05/Lewis-Postol.

39.

Lewis

Jeffrey

. 2008. “Minimum Deterrence.” Bulletin of the Atomic Scientists 64 (July/August): 38–41.

40.

Lieber

Keir A.

2005. War and the Engineers: The Primacy of Politics over Technology. Ithaca, NY: Cornell University Press.

41.

Mearsheimer

John J.

1983. Conventional Deterrence. Ithaca, NY: Cornell University Press.

42.

Military Balance . Various years. London: IISS.

43.

Mistry

Dinshaw

. 2003. Containing Missile Proliferation: Strategic Technology, Security Regimes, and International Cooperation in Arms Control. Seattle: University of Washington Press.

44.

National Intelligence Council. 2001. Foreign Missile Developments and the Ballistic Missile Threat through 2015. Washington: Central Intelligence Agency. Accessed July 5, 2012. http://www.dni.gov/nic/PDF_GIF_otherprod/missilethreat2001.pdf.

45.

Nolan

Janne E.

1991. Trappings of Power: Ballistic Missiles in the Third World. Washington, DC: Brookings Institution.

46.

Norris

Robert S.

Kristensen

Hans M.

. 2010. “Global Nuclear Weapons Inventories, 1945-2010.” Bulletin of the Atomic Scientists 66 (July-August): 77–83.

47.

Pape

Robert A.

1996. Bombing to Win: Air Power and Coercion in War. Ithaca, NY: Cornell University Press.

48.

Postol

Theodore A.

1991/1992. “Lessons of the Gulf War Experience with Patriot.” International Security 16 (Winter): 119–71.

49.

Rabinovich

Abraham

. 2004. The Yom Kippur War: The Epic Encounter that Transformed the Middle East. New York, NY: Schocken Books.

50.

Rauchhaus

Robert

. 2009. “Evaluating the Nuclear Peace Hypothesis: A Quantitative Approach.” Journal of Conflict Resolution 53 (April): 258–77.

51.

Reiter

Dan

. 2003. “Exploring the Bargaining Model of War.” Perspectives on Politics 1 (March): 27–43.

52.

Rousseau

David

Gelpi

Christopher

Reiter

Dan

Huth

Paul K.

. 1996. “Assessing the Dyadic Nature of the Democratic Peace, 1918-1988.” American Political Science Review 90 (September): 512–33.

53.

Rubin

1991. “How Much Does Missile Proliferation Matter?” Orbis 35 (Winter): 29–39.

54.

Snyder

Glenn H.

1961. Deterrence and Defense: Toward a Theory of National Security. Princeton, NJ: Princeton University Press.

55.

Tomz

Michael

King

Gary

Zeng

Langche

. 2003. “ReLogit: Rare Events Logistic Regression.” Journal of Statistical Software 8 (2).

56.

UN News Centre. 2010. “Citing Iran’s Failure to Clarify Nuclear Ambitions, UN Imposes Additional Sanctions.” UN News Service, June 9. Accessed July 5, 2012. http://www.un.org/apps/news/story.asp?NewsID=34970&Cr=iran&Cr1.

57.

Weidmann

Nils B.

Kuse

Doreen

Gleditsch

Kristian Skrede

. 2010. “The Geography of the International System: The CShapes Dataset.” International Interactions 36:86–106.

58.

Wohlstetter

Albert

. 1959. “The Delicate Balance of Terror.” Foreign Affairs 37 (January): 211–34.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

21.86 MB

0.00 MB