The foregoing discussion leads us to set up our core hypothesis for crisis behavior as a comparison between the classic bilateral explanations born out of rational deterrence models and brokered bargaining.
H0: Classic rational deterrence explanations set in the bilateral context continue to
account for crisis behavior between middle power nuclear states that operate without formal extended deterrence guarantees in a world with stronger hegemonic powers.
H1: The brokered bargaining model better accounts for crisis behavior between middle
power nuclear states that operate without formal extended deterrence guarantees in a world with stronger hegemonic powers.
Since we are most interested in explaining the mechanisms of crisis behavior, our alternative hypothesis does not necessarily amount to a wholesale rejection of the rational nuclear deterrence theory. However, we do rule out bilateral nuclear deterrence as the principal driver of crisis behavior and de-escalation. Nuclear weapons remain central to creating fear of unacceptable consequences of escalation and thereby imposing
constraints on the behavior of the three parties engaged in brokered bargaining. But we see the presence of nuclear weapons as endogenous to the crisis dynamic encompassed by our three-cornered bargaining framework. It therefore contributes to the process of interaction (along with other structural and situational factors that underpin our foregoing propositions) that ultimately shapes crisis behavior. Pressures to avoid escalation are a combination of both strands of the resolve-prudence trade-off described earlier. Since the presence of the third party is also endogenous to the crisis outlook of the principal
antagonists, the situation neither lends itself to establishing that nuclear weapons are the principal factor at play, nor to a straightforward counterfactual analysis which would see the absence of uncontrolled escalation solely as a function of third party presence. Studies that usually argue based on straight counterfactuals – escalation would likely have
occurred had the third party not been present – overlook the endogeneity problem in their analysis.67 The subtle but significant distinction between such analyses and what is argued here must be noted: we are contending that it is the process of trilateral interaction encompassed by our model that explains crisis behavior, and in turn, trajectories and outcomes. Studies of the sort referred to above on the other hand focus on the exogenous third party’s presence alone as central to their explanation. The fairest a priori assertion we can make is that in crises where behavior is shaped by brokered bargaining, the answer to whether classic nuclear deterrence would have held in the absence of the trilateral framework must remain indeterminate. Our contention challenges positions of both deterrence optimists and pessimists. We contest optimists in as far as they present nuclear deterrence as the principal variable driving crisis dynamics in nuclear dyads.68 The opposite is true of pessimists who tend to attribute escalation control to non-nuclear reasons like third party presence and see this as significantly weakening, if not
necessarily falsifying, optimist arguments.
2. Variables
The process marked by the recursive interplay of the perceptions, expectations, incentives, and strategies among the three parties that lies at the heart of the posited framework forms our dependent variable. If our hypothesis is to hold (if our theory of
67 Endogeneity refers to a situation whereby the values an explanatory variable takes on are a
consequence, rather than a cause, of the dependent variable. Such situations are common when observing political processes but create misleading conclusions about the direction of causality. Garry King, Robert O. Keohane, and Sidney Verba, Designing Social Inquiry: Scientific
Inference in Qualitative Research (Princeton: Princeton University Press, 1994), 185 (and 185-96 for a more in-depth discussion of the endogeneity problem).
crisis behavior is right), the process – and thus the expected crisis behavior – must
manifest itself any time a middle power nuclear crisis context appears and the triggers for our model to be set into motion are present: high sensitivity of the third party to the escalatory potential and the subsequent implications of escalation; and high sensitivity of the principal antagonists to third party preferences. The combination of these two
sensitivities forms our key explanatory variable. Table 2 presents the possible variability in the sensitivity levels of the antagonists and the third party on these two counts. Table 2: Possible Combinations of Sensitivity of the Third Party to the Escalatory
Potential and of the Antagonists to Third Party Preferences
Third party’s sensitivity to escalatory potential and subsequent implications
Principal antagonists’ sensitivity
to third party preferences High, Low Low, Low High, High Low, High Source: Author’s compilation.
As laid out in our propositions, cell 2,2 (High, High) in table 2 is likely to depict reality in the contexts most relevant to our discussion. This leads us to reiterate the rather deterministic prediction central to the theory of brokered bargaining: the three-cornered bargaining exercise that underpins the brokered bargaining model will play out in the expected manner every time a middle power nuclear crisis takes place in a world with stronger hegemons not intervening as competitor third parties.
3. Measurements
In any such crisis situation, four specific patterns of crisis behavior must be discernable for us to substantiate our hypothesis: (i) there should be clear evidence of the three-cornered interaction during a crisis; (ii) part of the behavior of the antagonists must be aimed at the third party actor(s) and there should be an observable influence of this
behavior on third party choices during the crisis; (iii) there should be an observable connection between the principal antagonists’ decisions to forego escalation and ultimately agree to terminate the crisis and the use of the third party’s “reward-
punishment leverage;” and (iv) in cases where uncontrolled escalation to the nuclear level takes place, there should be evidence that this occurred despite the conflicting parties’ desire to maintain positive evaluation and the third party’s fixation on the de-escalation goal rather than due to their indifference to third party preferences and de-escalation respectively. The last pattern also underscores the fact that the validity of the model does not hinge on successful de-escalation. Since the model explains and predicts the
mechanisms shaping crisis behavior, its explanatory power will be confirmed as long as the above-mentioned patterns are satisfied.
A stereotypical crisis that ends without uncontrolled escalation should exhibit autonomous decision making by the principal antagonists combined with deference to third party preferences even at the cost of political and reputational losses. The third party should not necessarily wait to be solicited into the mix, let alone be reluctant to intervene, and it should remain central to altering the preferences of the antagonists throughout the crisis. We should also find evidence of efforts by the conflicting parties to get third party backing for their stance while attempting to project the rival negatively and to isolate it but the third party should arduously work to attain swift de-escalation without alienating either of the antagonists completely. Ultimately, there should be sufficient evidence of the desire for positive evaluation by the conflicting parties rather than a situation where they ignore the third party. The primacy of the desire for positive evaluation would have
to hold even if both antagonists are left with sub-optimal results as far as their ideal crisis objectives are concerned.
IV. METHODOLOGY
We use the “structured focused comparison” method to test our hypothesis.69 The method involves devising a set of standardized instrumental questions in line with the research objectives and theoretical framework and applying them identically to each of the chosen case studies. The questions are general enough to allow for systematic comparisons among cases. At the same time, they must be “focused” on specific aspects of cases relevant to the research objective such that tangible and meaningful conclusions can be drawn. Structured focused comparison overcomes the weak cumulative potential of independent case studies while still retaining its strength of being able to conduct an in-depth investigation of each case. Since the questions are standardized, new cases that fit the conceptual framework and scope limitation of the study could be added and compared to the original set at any stage. This enhances the study’s relevance for policy makers. Moreover, the method is flexible enough to allow inclusion of additional
questions for a particular case if it has interesting unique aspects as long as they fit within the research frame.
69 For a basic introduction to the structured focused comparison method, see Alexander L.
George, "Case Studies and Theory Development: The Method of Structured, Focused
Comparison," in Diplomacy: New Approaches in History, Theory, and Policy, ed. Paul G. Lauren (New York: Free Press, 1979), 43-68; and Alexander L. George and Andrew Bennett, Case Studies and Theory Development in the Social Sciences (Cambridge: MIT Press, 2005), 67-72.
1. Research Questions
The research questions that form the basis of our inquiry for the case studies included here are listed below. These flow directly from our theoretical framework and will collectively allow us to test our core hypothesis.
1. What crisis objectives did the principal antagonists enter the crisis with and what functions of force did they employ in pursuing these?
2. What situational factors in the period since the previous crisis and at the time of the crisis most obviously impinged on the crisis outlook and objectives of the principal antagonists? How did these affect behavior patterns during the crisis?
3. If there is evidence of third party presence, what were the primary motivations for the third party to get involved? What strategy did the third party employ in pursuit of its objectives?
a. Was the third party a reluctant or eager entrant?
b. Did one or both principal antagonists target behavior to lure third party intervention?
4. Is there evidence of each actor behaving to affect the calculus of the other two actors (as opposed to just the conflicting parties affecting each other’s
behavior) during the crisis to further their respective objectives?
5. How did the behavior of the principal antagonists affect the third party’s expectations, strategy, and actions?
6. How did third party’s behavior affect the principal antagonists’ expectations, strategy, and actions?
7. Is there evidence of the principal antagonists conceding part of their decision making autonomy in deference to third party preferences?
a. Is there evidence that specific choices made by antagonists at ‘critical junctures’70 during crises were causally linked to the third party’s use of its “reward-punishment leverage” at those particular moments? 8. What is the implication of brokered bargaining for crisis stability?
9. How does experience with brokered bargaining affect behavior of the three parties after the crisis (the inter-crisis period) and during the next crisis iteration?
a. Is there evidence of “learning” by the principal antagonists that dictates preparations and behavior aimed at tilting the brokered bargaining dynamic in their favor during successive crises?
b. Is there evidence of “learning” on the part of the third party that alters its behavior in successive crisis iterations to increase the probability of successful crisis management on its part?
We will answer these questions by conducting a process tracing exercise.71 We will ‘soak and poke’72 in the details of each case study to decipher the presence of
70 ‘Critical junctures’ refer to phases or moments of institutional flux in an otherwise stable and
path-dependent environment. The term, as used here, implies a decision point in the crisis that can prove to be crucial in determining the trajectory and outcome of the crisis. On critical junctures, see Giovanni Capoccia and R. Daniel Kelemen, “The Study of Critical Junctures: Theory, Narrative, and Counterfactuals in Historical Institutionalism,” World Politics 59, 3 (April 2007): 341-369.
brokered bargaining. A process tracing exercise will allow us to examine the sequence of events, the linkage among choices and decisions of each of the three parties, and the motivations behind actions and signals of each actor. These are all crucial to identifying evidence that each of the three parties was influencing the other two’s behavior, that the antagonists’ behavior was partly aimed at the third party, that they were sensitive to third party preferences and the third party’s reward-punishment leverage was affecting their choices, and that the third party was consciously avoiding completely isolating either antagonist while pushing for de-escalation. In following this method, we would also have examined the pertinence of the principal rival
explanation – centrality of classic bilateral nuclear deterrence – for the observed crisis behavior. The analysis will also allow us to examine the implications of brokered bargaining for crisis stability.
In the interest of readability, we have not structured the sequence of our case studies according to the order in which the research questions have been presented. The sequence of developments and complexity of crises vary among the three cases and do not naturally lend themselves to an identical structure. Nonetheless, readers
71 For a basic introduction to the method of process tracing, see Stephen V. Evera, Guide to
Methods for Students of Political Science (Ithaca: Cornell University Press, 1997), 64-67; and John Gerring, Case Study Research: Principles and Practices (Cambridge: Cambridge University Press, 2007), 172-85.
72 ‘Soaking and poking’ is a methodological term brought to prominence by Richard Fenno’s
work on members of the U.S. Congress. It describes a technique which relies on ‘thick’ and detailed description of any mechanisms or occurrences relevant to the research question. Our aim to delve deep into the crisis behavior of the actors involved in our case studies and to
contextualize it approximates a soaking and poking exercise. For Fenno’s seminal work, see Richard F. Fenno, Home Style: House Members in Their Districts (Boston: Little, Brown and Company, 1978).
would be able to decipher that the narrative in each case study provides answers and explanations for each of the posed research questions.
2. Cases
This dissertation analyzes three case studies of middle power nuclear crises: the 1999 Kargil crisis; the 2001-02 standoff; and the 2008 Mumbai crisis. Each is a crisis between South Asian rivals, India and Pakistan. The U.S. operated as the principal third party in all three iterations although other external actors worked in tandem with it.
While we are studying only one rivalry, looking at three different crises helps us study behavior across time. Each crisis is a different ‘case’ within the overarching rivalry and can be compared and aggregated to provide cumulative knowledge for theory
development. The patterns of behavior in each case will form the within-case
‘observations.’ Focusing on one nuclear dyad also implies that background ‘noise’ – underlying conditions that, depending on the extent to which they vary across
observations, may affect our ability to determine causal linkages – can be minimized. At the same time, the variance over time on some of the situational ‘control’ variables allows us to test whether any of these were significantly correlated to crisis behavior such that they could bring into question the explanatory power of our theory.
Also to be noted, confirmatory evidence in line with our hypothesis would provide stronger-than-usual substantiation for the model. This is so since India is known to be obsessed with retaining its strategic independence and was traditionally wary of intervention from third party actors in India-Pakistan disputes. Pakistan, on its part, believes that the U.S. is actively interested in disarming its nuclear arsenal and is thus
hyper-sensitive to any situation that may allow U.S. say or intrusion in its nuclear matters. The South Asian dyad therefore constitutes a “least-likely” case in methodological terms: a case “that seems on a priori grounds unlikely to accord with theoretical predictions.”73 Finally, focus on a single rivalry lends itself to “aggregate learning” in terms of crisis behavior among the examined actors. Their behavior would indicate how their outlook towards crises and brokered bargaining has changed over successive iterations. Has brokered bargaining become more entrenched or have the antagonists moved to block the third party’s role during crises? How has the third party approached each crisis? Has its propensity to intervene and its crisis behavior changed over time? What has the collective learning of the three actors meant for crisis stability? These findings would help us generalize possible behavior patterns to other dyads as well.
A. Cases not Included in the Dissertation
Our case selection constitutes the “universe” of crises between two middle power nuclear states operating without any formal extended deterrence guarantees in a unipolar world. Every other nuclear crisis has either involved a global hegemon as one of the conflicting parties or has taken place under extended deterrence umbrellas.
There are four prominent cases that involved an antagonist other than the world’s hegemons and thus had a stronger third party potentially available for intervention: the India-Pakistan crises of 1987 and 1990; the Sino-Soviet border confrontation of 1969; and the simmering tensions on the Korean peninsula that have witnessed threats from a
nuclear North Korea from time to time since its first nuclear test in 2006.74 Each’s exclusion from the list of our case studies is driven by the scope limitation of our model.
The non-South Asian cases quite obviously fall outside the delimitations of the model. As already discussed, the Ussuri river incident involved one of the Cold War superpowers and this colored its hegemonic rival’s response. The Korean context, on the other hand, involves U.S. extended deterrence guarantees to South Korea and Japan. Exclusion of the South Asian cases needs more elaboration since a small strand of
literature focused on South Asia has latched on to the 1987 and 1990 episodes as the first two South Asian ‘nuclear crises.’75 Some have even gone on to argue that ‘existential’ deterrence most powerfully explained successful escalation management, at least in the 1990 crisis.76
Our decision to leave out these two cases flows from a key prerequisite for the posited three-cornered crisis engagement to be triggered. Since our model is specific to nuclear contexts and is predicated on the impact presence of nuclear weapons have on the outlook of the parties involved, all of them must perceive a nuclear environment to be present and internalize it in their outlook towards the crisis and their crisis choices. We therefore assume some evidence of an operational capability that can be successfully employed by both antagonists against each other as a necessary condition for crisis
74 For a timeline of related developments and tensions on the Korean peninsula dating back to
2002, see “Timeline: North Korea Nuclear Standoff,” BBC News, April 2, 2013, http://www.bbc.co.uk/news/world-asia-pacific-11811861.
75 See Hagerty, The Consequences of Nuclear Proliferation; Ganguly and Hagerty, Fearful
Symmetry, 68-115; and Chari, Cheema, and Cohen, Four Crises and a Peace Process, 39-117.
behavior to play out as predicted. Ambiguous, opaque deterrence/proliferation77
situations where even one of the three parties may not be convinced of either antagonist’s possession of an operational nuclear deterrent – and thus it cannot be expected to