indo-pakistani relations: reciprocity in long-term perspective

14
Research Notes Indo-Pakistani Relations: Reciprocity in Long-Term Perspective SHEEN RAJMAIRA East Carolina University This article examines the foreign policy behavior of two rival states, India and Pakistan. Previous studies of this dyad reveal competing causal claims concerning the nature of Indian and Pakistani relations. I argue that Indian and Pakistani foreign policy behavior exhibits strong short-term relations in the context of long-term “memories” that shape future expec- tations of their bilateral relations. The results indicate that reciprocity in Indo-Pakistani relations is shaped by a long-term equilibrium and suggest a reconceptualization of the nature of the Indo-Pakistani relations. The findings highlight the legacy of suspicions between these two countries, providing sober insights into the possibilities for reducing conflict and promoting cooperation in South Asia. Whether rival states respond to each other’s actions represents a true “puzzle” for international relations research (Zinnes, 1980). Studies of arms races, foreign policy behavior, and the development of cooperation among adversaries all focus, to some degree, on reciprocity. Some of these studies seek to determine if reciprocity exists, while others examine the role reciprocity plays in foreign policy interactions. This research reveals mixed findings regarding reciprocity, leading some to doubt the utility of reciprocity. These doubts may be a result of the varying conceptions of “reciprocity” that have been used in the international relations literature (Goldstein, 1991), despite the fact that these studies use similar data and empirical methods to examine reciprocity across dyads. As a pattern of action that characterizes the exchange of behavior between rivals (Keohane, 1986; see also Moore, 1995), reciprocity is variable in terms of the equality and nature of that exchange. This article examines short-term interactions of two rivals, India and Pakistan, in the context of their ongoing rivalry. This rivalry is rooted in long-term “memories” and exhibits itself in the short-term interactions of the two rivals. International Studies Quarterly (1997) 41, 547–560 ©1997 International Studies Association. Published by Blackwell Publishers, 350 Main Street, Malden, MA 02148, USA, and 108 Cowley Road, Oxford OX4 1JF, UK. Author’s note: This research was conducted with support from the Department of Political Science at East Carolina University and the Department of Political Science and the Institute of Behavioral Science at the University of Colorado. An earlier version of this article was presented at the 1995 Annual Meeting of the Midwest Political Science Association. I would like to thank a number of people, many of whom read different versions of this article, for their helpful comments and suggestions: John Freeman, Joshua Goldstein, Sean Q Kelly, David Kinsella, Michael D. McGinnis, Will H. Moore, Maury Simon, and Michael D. Ward. I would also like to thank Robert Durr for his advice on a methodological matter, and Henrik Bjarheim for sharing his thoughts on Indo-Pakistani relations with me. Marnie Pasciuto provided excellent research assistance. Downloaded from https://academic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

Upload: khangminh22

Post on 23-Mar-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Research Notes

Indo-Pakistani Relations:Reciprocity in Long-Term Perspective

SHEEN RAJMAIRA

East Carolina University

This article examines the foreign policy behavior of two rival states, Indiaand Pakistan. Previous studies of this dyad reveal competing causal claimsconcerning the nature of Indian and Pakistani relations. I argue thatIndian and Pakistani foreign policy behavior exhibits strong short-termrelations in the context of long-term “memories” that shape future expec-tations of their bilateral relations. The results indicate that reciprocity inIndo-Pakistani relations is shaped by a long-term equilibrium and suggesta reconceptualization of the nature of the Indo-Pakistani relations. Thefindings highlight the legacy of suspicions between these two countries,providing sober insights into the possibilities for reducing conflict andpromoting cooperation in South Asia.

Whether rival states respond to each other’s actions represents a true “puzzle” forinternational relations research (Zinnes, 1980). Studies of arms races, foreign policybehavior, and the development of cooperation among adversaries all focus, to somedegree, on reciprocity. Some of these studies seek to determine if reciprocity exists,while others examine the role reciprocity plays in foreign policy interactions. Thisresearch reveals mixed findings regarding reciprocity, leading some to doubt theutility of reciprocity. These doubts may be a result of the varying conceptions of“reciprocity” that have been used in the international relations literature (Goldstein,1991), despite the fact that these studies use similar data and empirical methods toexamine reciprocity across dyads. As a pattern of action that characterizes theexchange of behavior between rivals (Keohane, 1986; see also Moore, 1995),reciprocity is variable in terms of the equality and nature of that exchange. This articleexamines short-term interactions of two rivals, India and Pakistan, in the contextof their ongoing rivalry. This rivalry is rooted in long-term “memories” and exhibitsitself in the short-term interactions of the two rivals.

International Studies Quarterly (1997) 41, 547–560

©1997 International Studies Association.Published by Blackwell Publishers, 350 Main Street, Malden, MA 02148, USA, and 108 Cowley Road, Oxford OX4 1JF, UK.

Author’s note: This research was conducted with support from the Department of Political Science at East CarolinaUniversity and the Department of Political Science and the Institute of Behavioral Science at the University of Colorado.An earlier version of this article was presented at the 1995 Annual Meeting of the Midwest Political Science Association.I would like to thank a number of people, many of whom read different versions of this article, for their helpful commentsand suggestions: John Freeman, Joshua Goldstein, Sean Q Kelly, David Kinsella, Michael D. McGinnis, Will H. Moore,Maury Simon, and Michael D. Ward. I would also like to thank Robert Durr for his advice on a methodological matter,and Henrik Bjarheim for sharing his thoughts on Indo-Pakistani relations with me. Marnie Pasciuto provided excellentresearch assistance.

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

Previous studies of the Indo-Pakistani dyad have only focused on the short-term,tit-for-tat interactions between the countries. Most of these studies apply Richard-son’s (1960) action-reaction framework only to reach contradictory conclusionsabout Indo-Pakistani relations. For instance, Hollist (1977) and Oren (1994) bothfind evidence of negative reaction coefficients in their analyses of Indian andPakistani arms spending. Ward and Mahajan (1984) conclude that India is notresponsive to changes in its rival’s military expenditures. Majeski and Jones (1981)discover that Indian and Pakistani military expenditures are best represented bymoving average models not Richardsonian models and find no relationship betweenthe states’ arms spending. It appears that “factors internal [i.e., technology, costconstraints, fatigue] to each nation play a comparably greater role than does thebasic action-reaction factor” (Hollist, 1977:338).

Other studies have reached different conclusions. Majeski (1985) discovers thatPakistan develops expectations of India’s current military expenditures and factorsthis into its own spending decisions. India, on the other hand, does not respond toPakistani spending (Majeski, 1985:237). By contrast, Ferejohn (1976) concludesthat India reacts positively to increases in Pakistani spending. Taking a slightlydifferent approach, Freeman (1983) finds that past values of Indian spendinginfluence Pakistani spending, and concludes that Indian behavior fuels the conflictbetween the two rivals. Freeman, Kellstadt, and Williams (1996) reanalyzed Free-man’s (1983) study and found similar qualitative results. Rather than focus solelyon the short-term interactions of these states, I conceptualize the short-termbehavior of these rivals in the context of the long-term equilibrium that sustainstheir rivalry. This conception of foreign policy behavior among these rivals hasimplications for conflict resolution and confidence-building not only in South Asia,but in other regions as well.

Short-Term Interactions and Long-Term Memories

In the analysis of foreign policy interactions as in the arms race literature, manystudies examine superpower and regional power relations through reciprocity. Indoing so, many scholars examine the diplomacy, bargaining, and negotiationbehavior of states. Evidence of action-reaction processes is found primarily amongregional rivals in the Middle East or in South Asia rather than between the U.S. andthe Soviet Union (Ward, 1984:247). This highlights an important point about thenature of reciprocity in foreign policy behavior. Is reciprocity a short-term phe-nomenon that is best explained by daily vigilance and tit-for-tat behavior, or is it along-term, more diffuse phenomenon that represents a norm of behavior betweenrival states? Most existing research examines reciprocity as a short-term phenome-non and uses Richardsonian arms race models to examine foreign policy behavior.However, there is merit in conceptualizing reciprocity in a more long-term manner.The works of Axelrod (1984, 1986), Axelrod and Keohane (1985), and Goldstein(1995) provide a framework for understanding various interactions between statesin the context of iterated games such as the prisoner’s dilemma. Axelrod finds thatdaily vigilance is important in a tit-for-tat world of international politics. Dailyvigilance facilitates cooperation between self-interested actors, and has an evolu-tionary impact on the emergence of cooperative norms between adversaries. Reci-procity regulates conflict by establishing a norm of behavior over time, playing animportant role in the short-term, day-to-day interactions between states and in thelonger-term relations between these actors (Keohane, 1986; Rajmaira and Ward,1990; Ward and Rajmaira, 1992).

There are two factors that are likely to shape the frame of reference in whichforeign policy is formulated and implemented. First, is asymmetry. India is largerin terms of both population and territory and has more military might, as measured

548 Indo-Pakistani Relations

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

in manpower and technology, than Pakistan. Thus, the foreign policy behavior theserivals exhibit toward each other may not be symmetric. As the more powerfulcountry, India may be reluctant to decrease conflict or to cooperate with a weakerPakistan. A weaker Pakistan, however, may be willing to reduce conflict with India,but may not be willing to increase cooperation particularly in the context of theirestablished rivalry. Second, as regional rivals, the foreign policy agendas of Indiaand Pakistan are not as great in scope as would be superpower agendas. Indian andPakistani leaders seldom concern themselves with the international repercussions oftheir bilateral foreign policies; primarily, they are concerned with the impact oftheir bilateral relations in the South Asian region.1 Their intense rivalry and conflicthas culminated in three wars since their independence from Britain in 1947. Indiaand Pakistan are at the center of each other’s foreign policies and “official relations[between India and Pakistan] are based on a permanent state of paranoia and azero-sum mentality” (Thakur, 1992:168). This paranoia is a manifestation of theimmediate danger both sides perceive from the other, and is augmented both byideological and religious differences and by the disputed territory of Kashmir. It isthis perception of an immediate threat coupled with sustained hostility that shapesthe day-to-day interactions of the two rivals.

While it is evident that rival countries often respond to each other’s foreignpolicies, it is not clear that this response is always immediate. Certain critical issuesin foreign policy persist over several months or even years, as there are “manyforeign policy actions that are purportedly ‘reactive’ and ‘proportional’ that occuroutside of the week-to-a-month time scale preferred by many analysts” (Rajmairaand Ward, 1990:459). The Indo-Pakistani case is no exception, and the Kashmirissue is a case in point. In 1990 Pakistan carried out a military exercise dubbedZerb-e-Momin (“the believers’ counterpunch”) in Pakistani-controlled Kashmir. Thiswas a direct response to India’s major military exercise “Brass Tacks” whichoccurred in the fall of 1986, nearly four years earlier. The intent “was to deliver aclear message to India: If attacked, Pakistan still possessed the military strength todeliver a blow to India” (Burki, 1991:195). This example of the loose coupling (orlinking) of actions and reactions over a long period of time suggests that the foreignpolicy relations of India and Pakistan are shaped by a long-term memory. This isconsonant with what is known about Indo-Pakistani bilateral relations; Indian andPakistani foreign policy behaviors are “long-memoried” and are characterized byboth short and long-run dynamics.2 That is, the short-term, day-to-day policies ofIndia and Pakistan are made within the context of the long-term norm of behaviorthat exists between the two rivals. For over forty years this norm, or equilibriumlevel of behavior, has been characterized by high levels of sustained hostilitypunctuated by an intense ideological, religious, and political rivalry. This underly-ing norm provides the context for day-to-day foreign policy behavior; the short-runadjustments of foreign policy behavior occur within this context.

It is expected that the foreign policy behavior of these rivals move together ortrend together rather closely over time. Between these variables an equilibriumexists; this state is one “in which there is no inherent tendency to change” (Banerjee,Dolado, Galbraith, and Hendry, 1993:2) and it represents the “expected” but notconstant, level of conflict (or cooperation) that each state anticipates from the other.Indian and Pakistani behavior is attracted to a long-run equilibrium to which thesystem converges over time (Banerjee et al., 1993:2), and it fluctuates around a

1 This is not to say that these countries are never concerned with the role they play in world politics. My point hereis that in their bilateral relations, India and Pakistan are seeking primarily to maximize their security vis-à-vis eachother; this has resulted in relatively myopic foreign policies on both sides. See Thakur (1992).

2 Beck (1993:238) defines a process as “long-memoried” if “predictions about the distant future of the process aredependent on the current value of that process.” See also Smith (1993) on this point.

SHEEN RAJMAIRA 549

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

“moving equilibrium” (Smith, 1993; see also Ostrom and Smith, 1993). As Indianconflict toward Pakistan changes, this alters the “environment” of Indo-Pakistanirelations thus altering the appropriate level of Pakistani conflict toward India. Inshort, foreign policy behavior is not at rest but is a moving equilibrium that isreached whenever levels of conflict or cooperation are altered by either side. Indianand Pakistani foreign policy behavior will not drift far from each other; they arecoupled (Durr, 1993a; Ostrom and Smith, 1993), interdependent, and mutuallyresponsive.

Empirical Analysis

Data

The data utilized in this analysis are event data that measure the bilateral political,diplomatic, and military interactions of both India and Pakistan. The events weretaken from the Daily Reports of the Foreign Broadcast Information Service (FBIS).3While existing event data sets such as the World Event-Interaction Survey (WEIS)or the Conflict and Peace Databank (COPDAB) are a good source for data on GreatPower interactions, the paucity of events implies that they are a less than ideal choicefor the study of Indo-Pakistani relations. While FBIS has a coverage bias—itmonitors mostly areas of strategic importance to the United States—these DailyReports provide a rich yet often underutilized source for events (Schrodt, 1993:134).The events are scaled according to event type and affective content, cooperation orconflict,4 with 15 representing the most conflictual event, all-out war, and 1representing the most cooperative event, unification into one state (Azar and Sloan,1975).5 Next, these actions are weighted based on their intensity according to themethod developed by Azar and Sloan (1975) and aggregated into monthly conflictand cooperation time series.6 A net cooperation series is created by subtracting theconflict series from the cooperation series. Thus, there are six variables (Indianconflict, Pakistani conflict, Indian cooperation, Pakistani cooperation, Indian netcooperation, and Pakistani net cooperation) and 132 cases from 1980 to 1990 (12months for 11 years).

The graphs of the conflict, cooperation, and net cooperation data series arepresented in Figures 1, 2, and 3, respectively. In these figures, Indian and Pakistaniforeign policy behavior track together well, never drifting too far from each other.As they trend together, both the conflict and cooperation series do not return to aconstant level of behavior, rather the series move together, attracted to each other.Conflict behavior (Figure 1) corresponds nicely to what is known about Indo-Paki-stani bilateral relations. First, Indian levels of conflict are usually higher than those

3 FBIS is a composition of texts of foreign reports, newspaper stories, broadcasts, speeches of world leaders, and thelike, that are monitored by the Central Intelligence Agency.

4 Related to problems of bias is the issue of inter-coder reliability. Based upon a random sample of the data, a testof inter-coder reliability of the data used here indicates the data were highly reliable. Reliability was assessed bycalculating Cronbach’s α, the most widely used measure of reliability. Cronbach’s α is based on the inter-item correlationand is bound by 0 and 1 and where, as a rule of thumb, researchers strive for an α > .70 (see Carmines and Zeller,1979:44–8). Calculation of the α yielded a score of .94.

5 Categories 15 through 9 correspond to conflictual events, while categories 7 through 1 correspond to cooperativeevents. Events of COPDAB category 8 (no comment, neutral statements) were excluded.

6 The appropriate aggregation unit for event data is in dispute (Freeman, 1990; Goldstein, 1991; McClelland, 1983).The annual aggregations used by most studies risk overaggregation, and may cloud statistical and substantive inferences(Freeman, 1990). Nevertheless, only through theoretical guidance is it possible to determine the appropriate temporalaggregation. The monthly aggregation used here allows us to probe the long-term nature of reciprocity with little dangerof the problems of overaggregation.

550 Indo-Pakistani Relations

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

of its rival Pakistan. Second, there is a trend of increasing conflict for both seriesover the data period, culminating in late 1989–early 1990 with the beginning of apowerful, armed Muslim uprising in Indian-held Kashmir. The coupling of behav-ior is also evident in the cooperative relations of these two states (Figure 2). Pakistanexhibits somewhat higher levels of cooperation than does India. The very largeincrease in cooperation that emerges in late-1986 and early-1987 stems from theflurry of diplomatic activity that occurred as India and Pakistan sought to reduceborder tensions following the rapid escalation of hostilities which occurred withBrass Tacks in late fall of 1986. During this time, the two countries also held severaldiscussions on monitoring and curbing illegal border crossings and reducing drugtrafficking and smuggling. In addition to examining conflict and cooperation asdistinct types of foreign policy behavior, these variables were rescaled into a singlenet cooperation variable. Figure 3 displays the net cooperation series for the twocountries, reinforcing the findings in Figures 1 and 2. Corresponding to Figure 1,India exhibits larger troughs in net cooperation than does Pakistan. This figuredepicts a substantial decrease in the level of net cooperation for both states followingflare-ups in Kashmir in 1986, 1988, and 1990.

Nonstationarity

The cointegration of two series means that the linear combination of these twointegrated series produces a series that is stationary, or invariant with respect totime. If it is determined that two integrated series are cointegrated, this suggests (1)the existence of an equilibrium between the two series, and (2) that the two seriestrend together over time. Deviations from this underlying equilibrium of expectedbehavior reveal the nature of reciprocity in foreign policy behavior. If a shockdisturbs the moving equilibrium by forcing the series further apart (or closer

FIG. 1. Indian and Pakistani conflict, 1980–1990.

SHEEN RAJMAIRA 551

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

FIG. 3. Indo-Pakistani net cooperation, 1980–1990.

FIG. 2. Indian and Pakistani cooperation, 1980–1990.

552 Indo-Pakistani Relations

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

together) than what is expected, then this error is corrected over the long run asthe interdependent process seeks a new level consistent with equilibrium state (Durr,1993a:186). This error correction process reveals the dynamics of short-run adjust-ments in the context of the long-run equilibrium (Baghestani and Mott, 1991) andreciprocity is captured in the foreign policy adjustments to deviations from thenorms of expected behavior between rivals.

A series that is integrated is one in which the sum of past contributions is thesource of the process that generates the series (Banerjee et al., 1993:39). Specifically,it is said that the series is integrated of order d, or I(d), meaning that it needs to bedifferenced d times in order to produce a stationary series.7 The Dickey-Fuller (DF)or Augmented Dickey-Fuller (ADF) tests allow one to determine whether a series isI(1), which is a test for a unit root (Kennedy, 1992:253). The issue at hand is howto illustrate that the two series, X and Y, are variant with respect to time (i.e.,nonstationary). Typically, this is tested by employing DF or ADF tests which test fora unit root, that is if a series is I(1). While these unit root tests do have limitations,they remain the most widely employed tests for determining the nonstationarity oftime series.8

For the Indian and Pakistani foreign policy behavior series, the following ADFmodel was estimated.

∆yt = α + β1T + β2yt–1 + β3∆yt–1 + . . . + β15∆yt–12 + et , (1)

where T is the time trend.9 The test for a unit root is a test on the coefficient of yt-1,ß2. If this coefficient is significantly different from zero, the hypothesis that ycontains a unit root is rejected and the hypothesis is accepted that y is stationary.Greene (1993:564–6) presents a slightly different procedure for analyzing an ADFmodel with a time trend. In the above model, for example, the test for a unit rootis carried out by testing the joint hypothesis that β1 = β2 = 0 through the calculationof an F-test.10 Again, if these coefficients are significantly different from zero, thehypothesis that y contains a unit root is rejected. The results of these tests aredisplayed in Table 1. As indicated, both the t-test on the coefficient of yt-1 and theF-test conducted on the six series allow acceptance of the null hypothesis of a unitroot (p < .05) for all the series except Indian conflict to Pakistan. These results leadus to conclude that five of the six series, with the exception of Indian conflict, arenonstationary.11

Methods

The cointegration and error correction methodology consists of three regressionequations that are fit with ordinary least squares beginning with equation (1), thecointegrating regression. First, one determines if there is an equilibrium level of

7 This is typically what is considered a nonstationary series, although this is different than a nonstationary series,i.e., one that has a changing mean or variance.

8 Excellent treatments of the problems associated with the DF and ADF tests can be found in Cromwell, Labys, andTerraza (1994:10–9), Greene (1993:549–70),Ostrom and Smith (1993:177–81), and Williams (1993). However, thereseems to be a general agreement that these tests tend to be biased toward finding a unit root.

9 MacKinnon (1991) suggests including a time trend in the DF regression when there is significant drift in a series,as is the case for all series in the model. See also Ostrom and Smith (1993).

10 This F-statistic differs from the conventional F; critical values are found in Dickey and Fuller (1981) and Greene(1993).

11 A growing debate surrounds the appropriate properties of time series data that are necessary for conductingcointegration and error correction analysis. The logic of cointegration analysis maintains that two nonstationary seriesform a linear combination which is stationary, but whether the prerequisite of nonstationarity is a valid one is receivingmuch attention in the literature (see Banerjee et al., 1993; Beck, 1993; Durr, 1993b).

SHEEN RAJMAIRA 553

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

foreign policy behavior (conflictual or cooperative) between India and Pakistan. Lettwo series Xt and Yt represent Indian conflict to Pakistan and Pakistani conflict toIndia, respectively. The resulting cointegrating regressions regress each series onthe other plus a constant.

xt = α1 + C1yt + ut (2)

yt = α2 C2xt + vt (3)

When ut and vt are 0, X and Y can be assumed to be in equilibrium (Durr, 1993a).In addition, ut and vt measure the deviation of Xt and Yt, respectively, from theequilibrium, and represent the errors that are corrected in the long run. Hence, utand vt capture the long-run relationship between the two variables, X and Y.Following Engle and Yoo (1987), I employ the Augmented Dickey-Fuller test to testfor the cointegration of the series. The ADF methodology is represented below.

∆ut = –a1ut–1 + a2∆ut–1 + . . . + ak+1∆ut–k + eut (4)

∆vt = –b1vt–1 + b2∆vt–1 + . . . + bk+1∆vt–k + fvt (5)

Note that ut and vt are the error terms from equations (1) and (2) above and krepresents the number of lags. If the t-statistics for a1 and b1 are significant, the nullhypothesis of no cointegration is rejected. In other words, there is a long-runequilibrium that exists between the two series; over time these series will not driftfar from each other.

The next step is to examine the deviations or fluctuations of foreign policybehavior from this base, equilibrium level. Short-run changes or adjustments maycause these series to drift slightly, but they are brought back in line so that foreignpolicy behavior continues to meet the expectations of the moving equilibrium. Thisoccurs through an error correction mechanism12 that takes into account the errorsin such a way that short-run changes are incorporated into the long-run propertiesof the system (Baghestani and Mott, 1991). The importance of the error correctionmodel is that it focuses on change; short-run dynamics are included in the long-run

12 Given the endogeneity of Indian and Pakistani behavior, exogenous variables are not included in the model. SeeKennedy (1992:259) on the issue of general specification of cointegration models.

TABLE 1. Augmented Dickey-Fuller Tests for Unit Root

India PakistanNet Net

Conflict Cooperation Cooperation Conflict Cooperation Cooperation

yt-1a –4.07*** –2.94 –3.22* –2.71 –2.70 –2.07

F b 8.34** 4.35 5.33 3.94 3.66 1.25

N = 119.aCritical values for the coefficient of yt-1 are from MacKinnon (1991), where H0: β2 = 0, i.e., the exist-ence of a unit rootbCritical values for the F-test are from Dickey and Fuller (1981) and Greene (1993), where H0:ß1 = ß2 = 0, i.e., the existence of a unit root*p < .10; **p < .05; ***p < .01

554 Indo-Pakistani Relations

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

processes. The model seeks to uncover the long-run process of change in therelationship of the series.

∆xt = C3 + A1ut–1 + A2∆xt–1 + . . . + Ak+1∆xt–k +Ak+2∆yt–1 + . . . + A2k+1∆yt–k + ext

(6)

∆yt = C4 + B1vt–1 + B2∆yt–1 + . . . + Bk+1∆yt–k +Bk+2∆xt–1 + . . . + B2k+1∆xt–k + eyt

(7)

Of primary interest are the fluctuations, both increases and decreases, of conflictualand cooperative foreign policy behavior and the implications these fluctuations havefor Indo-Pakistani relations. In each case the error correction coefficients A1 andB1 (from equations (5) and (6) above) must be negative and lie in the interval(-1, 0), indicating that X and Y will respond negatively to the error introducedin the system, ut-1 and vt-1, respectively, in an effort to reestablish equilibrium(Durr, 1993a:194). Consider equation (1) above. If the shock to series Y is positiveindicating a negative value for ut-1, then a negative A1 implies that X will adjustupward in order to reestablish equilibrium at a new level. Similarly a positive ut-1implies that series X is already too high and a negative A1 means X should adjustdownward in order to reestablish equilibrium (Durr, 1993a:194). A positive errorcorrection coefficient A1 or B1 implies there is no equilibrium and in factindicates “error propagation” because X and Y are unwilling to adjust behaviorto correct for the errors, ut-1 or vt-1, that have been introduced into the system.In short, error correction behavior requires statistically significant and negativecoefficients, A1 and B1 (Durr, 1993a).

Results

The results of the cointegration and error correction analysis are presented inTable 2.13 The null hypothesis of no cointegration is rejected when the t-test for a1

TABLE 2. Long-Term Reciprocity in Foreign Policy Behavior

India PakistanError Error

Cointegration Correction Cointegration Correction N

Conflict –.62** –.54** –.78** –.80** 120(–3.21) (–1.79) (–3.42) (–1.96)

Cooperation –.99** –1.47** –.89** –.51 127(–4.98) (–2.19) (–4.96) (–.83)

Net cooperation –.56** –.61** –.69*** –.69** 124(–3.60) (–2.12) (–3.78) (–1.98)

Values in parentheses are t-statistics. The models include 10 lags for conflict, 4 for cooperation, and 7for net cooperation. Critical values for the ADF test for cointegration are taken from Engle and Granger(1987), whereas critical values of the standard t-distribution are used for the error correction model.Critical values are 2.84, 3.17, and 3.77 for the cointegration model and 1.66, 1.98, and 2.62 for the errorcorrection model, for α = .10, α = .05, and α = .01, respectively.*p < .10; **p < .05; ***p < .01

13 The columns labeled “Cointegration” present the coefficients and t-statistics for a1 and b1 from the ADFcointegration analysis (equations [3] and [4]), while the “Error Correction” columns present the coefficient and t-statisticsfor A1 and B1 from the error correction model (equations [5] and [6]).

SHEEN RAJMAIRA 555

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

and b1 (from equations [3] and [4] above) is greater than the value of the criticalt-statistic listed at the bottom of each table; these statistics are represented in thecolumn labeled “cointegration.” The coefficients and t-statistics for A1 and B1 (fromequations [5] and [6] above) are presented in the column labeled “error correction”and indicate whether there is an error correction process at work. After testing forup to 12 lags for each variable, the results displayed in Table 2 present the modelsthat performed the best, at 10, 4, and 7 lags for conflict, cooperation, and netcooperation, respectively.

For Indian and Pakistani conflict behavior, the cointegration test is significant ata 10-month lag indicating that both states formulate their foreign policy toward oneanother in the context of the previous 10 months of relations, thus suggesting arather long-memoried process. While there are short-term actions and reactions(e.g., mobilization of troops along the line of control in Kashmir), these short-termpolicies are being formulated in the context of longer-term (previous 10 months)foreign policy relations. If, for example, India receives conflictual foreign policybehavior that deviates from the level of conflict behavior it expects from Pakistan,the error correction mechanism indicates whether India adjusts (corrects) its conflictualforeign policy behavior in response to behavior it received from Pakistan and therate at which this adjustment occurs. Assume, for example, that Pakistan exhibitsconflictual foreign policy behavior toward India that deviates from India’s expecta-tion. According to the results presented in Table 2, the coefficient of .54 means thatthis equilibrium error is corrected by India at a rate of 54 percent per period (month)so that its behavior falls back in line with that of Pakistan and establishes a newequilibrium level of foreign policy behavior (Durr, 1993a:186). If there is a “shock”(a sudden increase or decrease in Pakistani conflict behavior toward India) to thesystem in period t-10, for Indian conflict behavior, 54 percent of the initialdisequilibrium remains at t-9, 29.2 percent remains at t-8, 15.8 percent at t-7, andso forth. A similar dynamic exists in the Pakistani case; an error correctioncoefficient of .80 means that Pakistan corrects for any disequilibrium (from a“shock” on the Indian side) in its bilateral relations with India at a rate of 80percent per month.

While the cointegration analysis indicates that Indian and Pakistani coopera-tive behavior shares a common trend, the error correction results from cooperativebehavior suggest a growing disequilibrium results from shocks to the system. Forinstance, at t-4, the error correction coefficient for India is 1.47, suggesting that anydisequilibrium is adjusted at the rate of 147 percent per month. This implies thatfor a shock that occurred at t-4, 147 percent of the initial disequilibrium remains att-3, 216 percent at t-2, 317.6 percent at t-1, etc. Thus while India does respond toshocks (increases or decreases) in Pakistani cooperative behavior by adjusting itsbehavior, it does not endeavor to bring Pakistani behavior back in line with theexpected equilibrium. Instead, India amplifies the disequilibrium in cooperativeforeign policy; India is not willing to alter its cooperative foreign policy to maintainequilibrium with the long-run dynamics of Pakistani cooperative foreign policy.Similarly for Pakistan, no error correction coefficients are significant; Pakistanibehavior is not shaped by changes in Indian cooperative behavior since Pakistandoes not respond to Indian cooperation at all.

The net cooperation results may link the disparate results of conflictual andcooperative foreign policy behavior.14 For both rivals, the cointegration analysis

14 States may condition future cooperation on the attenuation of conflict, and perhaps it is the case that examiningconflict and cooperation separately misses the relationship between these types of foreign policy behavior. However,simple correlational analysis of the Indo-Pakistani foreign policy behavior series does not support this notion. Indianconflict and cooperation have a correlation of -.05 (p = .58) while Pakistani conflict and cooperation have a correlation

556 Indo-Pakistani Relations

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

indicates that Indian and Pakistani net cooperation share a common trend and thatboth states formulate policy toward each other in the context of the previous sevenmonths of relations. In addition to this long-term trending of Indo-Pakistan netcooperation, each country adjusts its behavior in response to the behavior it receivesfrom the other. These error correction coefficients are also significant at sevenmonths for both India and Pakistan. Interpretations about the rate at which thedisequilibrium in conflictual behavior is corrected are as follows: the rate at whichPakistan corrects for any shocks related to net cooperative behavior is higher thanthe rate at which India corrects for shocks to net cooperative behavior. In short,these results indicate that India and Pakistan do correct for errors in each other’snet cooperative behavior.

Let us consider the substantive implications of these results. The rate at whichPakistan corrects for any shock related to conflict behavior is generally higher thanthe rate at which India corrects for Pakistani shocks to conflict behavior. Pakistanis more willing to alter the equilibrium and reestablish a new level of expectedbehavior by moving more quickly to alter its conflict behavior, while India appearsto take things slower in terms of recalibrating conflictual foreign policy behavior.As the more powerful nation, India is not motivated to move too quickly toreestablish a new equilibrium in its conflictual relations with Pakistan. As the lesspowerful nation, Pakistan is quicker to respond seeking to reduce the disequilibriumat a faster rate. In the case of cooperative behavior, India adjusts its behavior inresponse to changes in Pakistani cooperation after four months, seeking not toreestablish equilibrium but rather to augment the disequilibrium. Pakistan, how-ever, does not respond to changes in Indian cooperative behavior at all. The conflictand cooperation results in Table 2 indicate that the error correction model isapplicable only to bilateral conflictual foreign policy behavior between India andPakistan but not for cooperative relations. It appears that India and Pakistan treatconflict and cooperation as relatively distinct types of foreign policy behavior.

Summary and Conclusions

This paper examined the bilateral relations of two important regional rivals, Indiaand Pakistan, by analyzing short-term foreign policy interactions in the context ofa long-run relationship that is shaped by the ongoing rivalry between the two states.Much of the extant research on foreign policy behavior has focused on (1) super-power relations and (2) the short-term, tit-for-tat relations between rivals. Thisanalysis of the Indo-Pakistani case provides a more nuanced understanding of theforeign policy interactions of two increasingly important actors in the post–ColdWar world. It provides a new approach with which to conceptualize reciprocity inIndo-Pakistani foreign policy relations. In particular, the results shed some light onthe longer-term relationship between Indian and Pakistani conflict and cooperationsuggesting that short-term, tit-for-tat reciprocity occurs within the context of a morediffuse, longer-term framework.

The results have important implications. First, these results lend support to theutility of analyzing reciprocity in terms of its long-term properties as suggested byAxelrod (1984, 1986) and put forth by Rajmaira and Ward (1990) and Ward andRajmaira (1992). As researchers shift their focus to regional rivalries in thepost–Cold War world, this approach to the study of reciprocity is strengthened by

of -.01 (p = .95). In addition, Indian net cooperation is correlated strongly with Indian conflict (-.93, p = .00) while itis correlated moderately but significantly with Indian cooperation (.42, p = .00). Similarly, Pakistani net cooperation iscorrelated with Pakistani conflict and cooperation with values of -.83 (p = .00) and .56 (p = .00), respectively. Thus, netcooperation variables are not capturing anything different than the conflictual relations between the two states.

SHEEN RAJMAIRA 557

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

the insights it provides into the Indo-Pakistani case, a case that is vastly different incultural, historical, and geographical context than the U.S.-Soviet case. Second, thislong-term manifestation of reciprocity necessitates a reconceptualization of theforeign policy dynamics between regional rivals such as India and Pakistan. Indianand Pakistani treatment of conflict and cooperation as distinct types of foreign policybehavior suggests that efforts to link cooperative initiatives with conflict reductionmay be unsuccessful. This is underscored by the unwillingness of Pakistan to respondto the changes in cooperative foreign policy behavior it receives from India whileIndia responds by worsening the disequilibrium in cooperative relations.

On the other hand, this separation of conflict and cooperation implies that Indiaand Pakistan can begin to cooperate on issues of less consequence than Kashmir.Agreements on cultural exchange of music or movies may spill over into furthertalks on water rights, drug trafficking, or smuggling, for instance, all occurring,however, in the context of continued conflict over enduring issues such as Kashmir.Although there will be short-term shocks that alter day-to-day relations, theseday-to-day policies of India and Pakistan will be made within the context of thelonger-term conflict equilibrium that exists between the two rivals. This underlyinglevel of behavior continues to be characterized by high levels of enduring hostility,an ongoing rivalry, and limited cooperation. While there is short-term cooperationbetween the two rivals (i.e., meetings of government officials, official culturalexchanges, limited agreements on technological and economic matters, etc.), as ofyet, there is no evidence of sustained, long-term cooperation between India andPakistan. Both countries conceive of cooperative behavior primarily in the shortterm; cooperation is viewed as a short-term phenomenon which has little long-term,evolutionary impact. This suggests a potentially dangerous dynamic in Indo-Paki-stani bilateral foreign policy relations; a dynamic in which the memory of conflictis long-term, while the memory of cooperation is short. This focus on conflictcoupled with the legacy of suspicions between the two rivals makes the region appearripe for future confrontation. An extreme sensitivity to conflict coupled with therelative inattentiveness to cooperation might fuel a rapid and unpredictable esca-lation of conflict between the two rivals.

There are two factors which may be important in the future of Indo-Pakistanirelations. The first of these is China and its relations with both India and Pakistan.Sino-Indian relations have been historically tense. Sino-Pakistani relations, on theother hand, have been friendly as exemplified by reports of the Chinese sale ofnuclear materials and technology to Pakistan in recent years. In considering securityin the region, therefore, it is important to consider the role China plays inIndo-Pakistani relations (Rajmaira, 1997). Second, efforts to expand confidence-building measures (CBMs) may also play an important role in South Asian securityand stability. While CBMs have not been absent in Indo-Pakistani relations, theyhave been few and far between.

The development of CBMs is likely to be affected by several factors. First is thestability of democracy in Pakistan. The rapid regime changes that have plaguedPakistan since its inception in 1947 continue to influence the Indo-Pakistanirelations. Existing Indian threat perceptions are exacerbated by fears of anothermilitary regime coming into power in Pakistan. However, it is also possible that theconsolidation of democracy in Pakistan will establish the conditions for continuedconflict with India. In an effort to maintain his political support, Nawaz Sharif maybe compelled to continue to pursue hard-line policies in Pakistan’s bilateral rela-tionship with India, thus reducing possibilities for the development of long-termcooperation. Second, the ethnic and separatist movements within both countriesaugment the paranoia of both sides, as each struggles to shore up its militarydefenses to protect against internal and external threats to security. Third, and mostimportant, is Kashmir. A resolution of the Indo-Pakistani conflict is impossible

558 Indo-Pakistani Relations

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

without addressing the outstanding issue of Kashmir. The problem has grown moreintractable over the years as both sides have become entrenched in their positionsfueled in large part by the increase of religious fundamentalism in both countriesand the end of the Cold War. Nevertheless, CBMs may work to develop a long-termmemory for cooperative behavior between these two rivals, in which case long-termconflict resolution could take root. Given the volatility of the region and thepredisposition to conflict, it appears that hopes for reduced conflict and sustainedcooperation, however limited, are linked to the development of sustained CBMsbetween the two rivals and the role China chooses to play in the ongoing rivalrybetween India and Pakistan.

ReferencesAXELROD, R. (1984) The Evolution of Cooperation. New York: Basic Books.AXELROD, R. (1986) An Evolutionary Approach to Norms. American Political Science Review 80:1095–1111.AXELROD, R., AND R. O. KEOHANE (1985) Achieving Cooperation Under Anarchy. World Politics

38:226–254.AZAR, E. E., AND T. J. SLOAN (1975) Dimensions of Interaction: A Source Book for the Study of 31 Nations

from 1948 Through 1973. Studies of Conflict and Peace, Department of Political Science, Universityof North Carolina at Chapel Hill.

BAGHESTANI, H., AND T. MOTT (1991) Cointegration and the Analysis of the Money Supply Process. Paper.University of Colorado.

BANERJEE, A., J. DOLADO, J. W. GALBRAITH, AND D. F. HENDRY (1993) Co-integration, Error Correction andthe Econometric Analysis of Non-Stationary Data. New York: Oxford University Press.

BECK, N. (1991) Comparing Dynamic Specifications: The Case of Presidential Approval. Political Analysis3:51–87.

BECK, N. (1993) The Methodology of Cointegration. Political Analysis 4:237–247.BURKI, S. J. (1991) Pakistan: The Continuing Search for Nationhood, 2nd ed. Boulder, CO: Westview Press.CARMINES, E. G., AND R. A. ZELLER (1979) Reliability and Validity Assessment. Paper #17 in Sage Series in

Quantitative Applications in the Social Sciences, editor J. L. Sullivan. Newbury Park, CA: Sage.CROMWELL, J. B., W. C. LABYS, AND M. TERRAZA (1994) Univariate Tests for Time Series Models. Paper #99

in Sage Series in Quantitative Applications in the Social Sciences. Newbury Park, CA: Sage.DICKEY, D. A., AND W. A. FULLER (1981) Likelihood Ratio Tests for Autoregressive Time Series with a

Unit Root. Econometrica 49:1057–1072.DURR, R. H. (1993a) An Essay on Cointegration and Error Correction Models. Political Analysis

4:185–228.DURR, R. H. (1993b) Of Forests and Trees. Political Analysis 4:255–258.ENGLE, R. F., AND C. W. J. GRANGER (1987) Co-integration and Error Correction: Representation,

Estimation, and Testing. Econometrica 55:251–276.ENGLE, R. F., AND B. S. YOO (1987) Forecasting and Testing in Co-integrated Systems. Journal of

Econometrics 35:143–159.FREEMAN, J. R. (1983) Granger Causality and the Time Series Analysis of Political Relationships. American

Journal of Political Science 27:327–358.FREEMAN, J. R. (1990) “Systematic Sampling, Temporal Aggregation and the Study of Political Relation-

ships.” In Political Analysis, vol. 1, edited by J. A. Stimson, pp. 61–98. Ann Arbor: University ofMichigan Press.

FREEMAN, J. R., P. KELLSTADT, AND J. T. WILLIAMS (1996) Unit Roots and Causal Inference in PoliticalScience. Paper presented at the Political Methodology Summer Meetings, Ann Arbor, Mich., July18–21.

GOLDSTEIN, J. S. (1991) Reciprocity in Superpower Relations: An Empirical Analysis. International StudiesQuarterly 35:195–209.

GOLDSTEIN, J. S. (1995) Great-Power Cooperation Under Conditions of Limited Reciprocity: FromEmpirical to Formal Analysis. International Studies Quarterly 39:453–477.

GREENE, W. H. (1993) Econometric Analysis, 2nd ed. New York: Macmillan.HOLLIST, W. L. (1977) Alternative Explanations of Competitive Arms Processes: Tests on Four Pairs of

Nations. American Journal of Political Science 22:313–340.KENNEDY, P. (1992) A Guide to Econometrics, 3rd ed. Cambridge, MA: MIT Press.

SHEEN RAJMAIRA 559

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022

KEOHANE, R. O. (1986) Reciprocity in International Relations. International Organization 40:1–27.MACKINNON, J. G. (1991) “Critical Values for Cointegration Tests.” In Long-Run Economic Relationships:

Readings in Cointegration, edited by R. F. Engle and C. W. J. Granger, pp. 267–276. New York:Oxford University Press.

MAJESKI, S. J. (1985) Expectations and Arms Races. American Journal of Political Science 29:217–245.MAJESKI, S. J., AND D. L. JONES (1981) Arms Race Modelling: Causality Analysis and Model Specification.

Journal of Conflict Resolution 25:259–288.MCCLELLAND, C. A. (1983) Let the User Beware. International Studies Quarterly 27:169–178.MOORE, W. H. (1995) Action-Reaction or Rational Expectations? Reciprocity and the Domestic-Interna-

tional Nexus During the “Rhodesia Problem.” Journal of Conflict Resolution 39:129–167.OREN, I. (1994) The Indo-Pakistani Arms Competition: A Deductive and Statistical Analysis. Journal of

Conflict Resolution 38:185–214.OSTROM, C. W., JR., AND R. M. SMITH (1993) Error Correction, Attitude Persistence, and Executive

Rewards and Punishments: A Behavioral Theory of Presidential Approval. Political Analysis4:127–184.

RAJMAIRA, S. (1997) Enter the Dragon: China’s Role in the Indo-Pakistani Rivalry. Political ResearchQuarterly. Forthcoming.

RAJMAIRA, S., AND M. D. WARD (1990) Evolving Foreign Policy Norms: Reciprocity in the SuperpowerTriad. International Studies Quarterly 34:457–475.

RICHARDSON, L. F. (1960) Arms and Insecurity: A Mathematical Study of the Causes and Origins of War.Pittsburgh, PA: Boxwood Press.

SCHRODT, P. A. (1993) “Machine Coding of Events Data.” In International Event-Data Developments:DDIRPhase II, edited by R. L. Merritt, R. G. Muncaster, and D. A. Zinnes, pp. 125–148. Ann Arbor:University of Michigan Press.

SMITH, R. M. (1993) Error Correction, Attractors, and Cointegration: Substantive and MethodologicalIssues. Political Analysis 4:249–254.

THAKUR, R. (1992) India After Nonalignment. Foreign Affairs 71:165–182.WARD, M. D. (1984) Differential Paths to Parity: A Study of the Contemporary Arms Race. American

Political Science Review 78:297–317.WARD, M. D., AND A. K. MAHAJAN (1984) Defense Expenditures, Security Threats, and Governmental

Deficits: A Case Study of India. Journal of Conflict Resolution 28:382–419.WARD, M. D., AND S. RAJMAIRA (1992) Reciprocity and Norms in U.S.-Soviet Foreign Policy. Journal of

Conflict Resolution 36:342–368.WILLIAMS, J. T. (1993) What Goes Around Comes Around: Unit Root Tests and Cointegration. Political

Analysis 4:229–236.ZINNES, D. A. (1980) Three Puzzles in Search of a Researcher. International Studies Quarterly 24:315–342.

560 Indo-Pakistani Relations

Dow

nloaded from https://academ

ic.oup.com/isq/article/41/3/547/1796508 by guest on 13 July 2022