Abstract
Military alliances are an important feature of major power competition. One way for a major power to increase its relative power is to drive a wedge in a rival’s alliance. Why and when does a major power do so? I propose a theory of strategic opportunity, which holds that a major power seeks to drive a wedge in the rival’s alliance by offering economic aid to a rival’s protégé when that country experiences a rapid deterioration in its relations with its patron. I test my theory on major power rivalry dyads and find empirical support for my argument on a set of directed dyad-year observations between 1960 and 2010. The findings contribute to our understanding of alliance politics and strategic considerations behind using economic carrots in major power rivalry.
Introduction
Alliance management is one of the most important features of major power competition. As such, scholars have widely studied how major powers manage their alliances (Blankenship 2020; Cha 2010; McManus and Nieman 2019; Morrow 1991). Although much ink has been spilled on how major power patrons manage their minor power protégés (e.g., Morrow 1991), less attention has been paid to how adversarial major powers seek to create divisions within opposing alliances. Under what conditions would a major power seek to drive a wedge in a rival alliance? 1 There are considerable risks in conducting a wedge strategy; the divider’s economic aid may serve no purpose in increasing the divider’s influence over the target protégé since the latter is a member of a rival alliance, or even worse, the aid might even strengthen a potential adversary. What explains when and why major powers offer economic concessions to the protégés of a rival alliance?
The question of wedge strategies has both theoretical and practical importance. “Divide and rule” is an adage as old as time (Crawford 2008; Izumikawa 2002). Several renowned strategists both in the East and the West had written long ago on the importance of dividing enemies. Already back in the sixth century B.C. Sun Tzu thus wrote: “when he is united, divide him…sometimes drive a wedge between a sovereign and his ministers; on other occasions separate his allies from him. Make them mutually suspicious so that they drift apart” (Griffith 1963). Likewise, Machiavelli asserted “whenever there are many powers united against another power…one ought always to put more hope in that one alone…[for] by using a little industry, he will be able to disunite the very many and to weaken the body that was mighty.” (Machiavelli 1998, 245). Such ideas are by no means outdated. In fact, they are very much alive within the minds of policymakers today. For example, Biden stated at the 2021 NATO summit that “Russia and China are both seeking to drive a wedge in our transatlantic solidarity” (Siebold, Holland and Emmott 2021). Similarly, the National Security Strategy document under the Biden administration notes that Beijing seeks to “erode U.S. alliances in the region and around the world” (Biden 2022).
Given the self-evident importance of the topic in the alliance literature, it is surprising that relatively scant attention has been paid to the concept. 2 When a major power seeks to aggregate its power through an alliance, we can expect that a competing major power would seek opportunities to disrupt such alliances. Yet, much of the IR scholarship has focused only on the first of these two aspects, focusing on balancing and bandwagoning (e.g., Schweller 1994, 2004; Walt 1987). 3 However, as Crawford (2011) notes, it is difficult to understand behaviors such as balancing or bandwagoning without a proper understanding of wedge strategies: bandwagoning is a type of outcome in alliance behavior while driving a wedge among potential or realized allies is the process that leads to such an outcome. We would only be able to gain a complete picture of the outcome only when we have a deeper understanding of the process that lies behind it (Crawford 2011).
In this paper, I propose a theory of strategic opportunity: I argue that a major power seeks to drive a wedge by offering economic aid to a protégé of the rival alliance when the protégé experiences a rapid deterioration of relations with its patron. Using wedge strategies involves a degree of risk for the divider since it is generally unlikely that the divider can exert influence over the target protégé of a rival alliance (Baldwin 1971, 29–30; Walt 1987, 221–242). In the worst case, the divider’s aid may only serve to strengthen a potential adversary. As such, a major power divider is more likely to use windows of opportunity and seek to exploit a rift in a rival alliance by offering economic inducements. This would increase the divider’s likelihood of success in influencing the target protégé’s behavior.
I test my theory on a set of directed dyad-year observations from 1960 to 2010 consisting of major power dividers as the sender states and the protégés of the rival alliance as the target states. I find empirical support for my argument: a higher difference in ideal point distance between the target protégé and its patron is positively associated with greater aid from the sender state. The finding is robust to a wide range of specifications controlling for observed confounders.
The contribution of this paper is twofold. First, this study deepens our broader understanding of alliance politics. Although there has been no shortage of works on the reliability of alliances and reassurance measures within alliances (e.g., Blankenship 2020; Leeds 2003; Leeds, Long and Mitchell 2000; McManus 2017; McManus and Nieman 2019), a proper understanding of the alliance dynamics would be incomplete without considering the strategy of the divider. Understanding how major powers seek to divide rival alliances has significant implications for the broader literature on alliances − as Crawford (2021, 3) correctly notes, it would be difficult to fully understand the question of alliance reliability without a deeper understanding of this topic.
Second, this study makes important contributions to the broader International Relations literature on political influence by showing that major power politics is an integral part in explaining the conditions under which major donors give aid. Seeking influence in international politics is a common theme explored across multiple subfields within International Relations (e.g., Flores-Macías and Kreps 2013; Malis and Smith 2021). In particular, International Political Economy scholars have examined how major power donors seek to exert influence by buying UN votes through economic aid (e.g., Carter and Stone 2015; Dreher, Nunnenkamp and Thiele 2008). However, much of the theoretical framework proposed in these studies is approached from the perspective of International Political Economy. Similar to Walt (1985) and Baldwin (1985), I show that strategic opportunity between rival alliances is an important factor to be considered when analyzing aid transferred between rival alliances.
The rest of the paper proceeds as follows. I first elaborate on the concept of wedge strategies in major power rivalry and why major powers may want to use them. I then explain the conditions under which major powers are likely to use economic aid to drive a wedge in a rival alliance. Next, I test my theory by conducting a panel analysis consisting of directed dyad-year observations. The final section concludes with the contributions of the study and suggestions for future research.
Concept of Wedge Strategy
At the most fundamental level, wedge strategies refer to the use of “diplomacy and statecraft to move or keep a potential adversary out of an opposing alliance” (Crawford 2021, 1). Wedge strategy is a broad concept that operates in “many different contexts” (Cooley and Nexon 2020, 59). As such, there are various tools that could be used for wedge strategy. For example, Goddard (2009) traces how Prussia used rhetoric in the form of “legitimation strategies” to drive a wedge between potential coalition partners. Similarly, Chai (2020) examines how China has recently used persuasion to drive a wedge between Washington and Canberra. Creating divisions within societies through propaganda has also been described in the context of wedge strategies (Goddard and Nexon 2016; Kim and Simón 2025).
The wedge strategy literature identifies two types of tools used in wedge strategy, namely carrots (Crawford 2008, 2021; Dian and Kireeva 2022; Yin 2022; Yoo 2015) and sticks (Izumikawa 2013; Lee and Kwon 2024; Ross 2025; Vu 2024). Carrots refer to the use of positive inducements (Drezner 1999; Nincic 2010, 2011) while sticks refer to the use of (military) coercion (Izumikawa 2002; Taffer 2020) to change the behavior of the target. The relative merits of a carrot-based wedge strategy has been the subject of debate between Crawford and Izumikawa (Crawford 2011; Izumikawa 2013).
Why would a major power ever seek to use a (carrot-based) wedge strategy in major power competition? Given the conventional wisdom that minor powers can add little to the overall fighting capability, some might argue that there is little incentive for a major power to use wedge strategies. However, such an argument overlooks the strategic advantage that minor powers could provide to major powers. As Crawford (2021, 11) notes, the “determinants of strategic weight are varied and context-specific.” 4 Minor powers are often geographically located in strategically critical parts for major powers and serve as sites for military power projection (Morrow 1991, 914). For example, Washington paid attention to the extent to which Ankara would cooperate with Moscow regarding the use of the Turkish straits (FRUS 1970). 5
Although there are various tools that could be analyzed for wedge strategy, I focus on economic aid for the sake of analytical parsimony. Focusing on economic aid is justifiable, since aid appears to be the predominant tool used in wedge strategies and is widely examined in much of the scholarship (e.g., Crawford 2008; Crawford 2021; Huang 2020; Izumikawa 2013; Yin 2022). 6 Economic aid as a tool for seeking political influence has been widely discussed by political scientists and economists. In particular, International Political Economy scholars have widely noted that states use economic aid as a tool of economic statecraft to exert political influence (Alesina and Dollar 2000; Bearce and Tirone 2010; Dreher, Nunnenkamp and Thiele 2008). International Relations scholars have also widely noted that the United States often uses economic aid as an important tool to seek influence (Carter and Stone 2015; De Mesquita and Smith 2007). In short, it is plausible to assume that major donor countries use economic aid as a tool to exert their influence over other countries.
Although the concept of using economic aid as a tool of wedge strategies is obviously related to the broader literature on using aid to seek influence, there is a subtle difference in that wedge strategies are conceptualized as a zero-sum influence over the target protégé. In other words, if we characterize a country (A) offering economic aid to another (B) as “A seeking influence over B,” wedge strategy involves “A seeking influence over B at the expense of C,” where C is a rival major power. In essence, a divider seeks influence over the target protégé at the expense of the latter’s patron. By signaling that C has outside options, A would seek to reduce B’s dependence on C and replace such a dependency with A’s own influence. Thus, a wedge strategy involves influencing the target to adopt a behavior that is contrary to the rival patron’s security interests. As such, I assume that when a major power divider offers economic aid to a target protégé, it is at least partly motivated to drive a wedge in the rival alliance. 7
There is certainly a wealth of qualitative evidence suggesting that the United States used economic aid as a tool to drive a wedge in a rival alliance. Historical accounts show that major powers have been keen to reward members of a rival alliance when they suffered a deterioration in the relations with the patron. 8 If the assumption that aid can be used as a tool of wedge strategy in the context of major power rivalry is a plausible one, then we could conceptualize any instance of a major power offering economic aid to a protégé of a rival alliance as being at least partly motivated to drive a wedge in the rival alliance. 9
Windows of Opportunity and Carrots
When would a major power use wedge strategies against a rival alliance? At the core of my theory is the element of strategic opportunity: I argue that a major power seeks to drive a wedge by offering economic inducements to a protégé of the rival alliance when the latter experiences a rapid deterioration of relations with its patron.
Implementing wedge strategy during such windows of opportunity would minimize the risk of economic aid being wasted or used to strengthen a potential adversary. Historical cases provide some face-level evidence for this theory. For example, Hui (2005, 121) argues that it was the animosity among Louis XIV’s opponents that provided the “fertile ground for the divide-and-conquer strategy.” Why are such windows of opportunity necessary for a divider to implement wedge strategies? Obviously, sufficient materialistic capabilities are necessary for dividers to be able to implement wedge strategies with respect to a rival alliance, but they are not by themselves sufficient in terms of incentives. In other words, a major power divider does not conduct wedge strategies merely because it has the capabilities to do so. Providing aid to a protégé of a rival alliance generally makes little strategic sense since the divider would be able to exert little influence over the target protégé. After all, the two are likely to have conflicting interests; the target protégé most likely joined the rival alliance because it perceived security threats from the divider in the first place. As such, a carrot-based wedge strategy is likely to be a futile attempt in influencing the behavior of the target.
In such a context, a major power divider would need some credible signal to believe that it is worth investing in a carrot-based wedge strategy. A rapid deterioration in the relations between the target protégé and its patron may provide such an incentive for two reasons. First, it is less risky for a major power to exploit a rift in the rival alliance than to attempt such a strategy when the relations between the rival patron and its protégé are relatively stable. Providing aid when there is little chance of influencing the target is likely to be futile. In contrast, a marked rapid deterioration in the relations would be a signal to the divider that there could be an opportunity to further worsen the relations between the two through the use of carrots. Second, the fact that there has been a deterioration in the relations signifies that the protégé would depend less on its patron. Subsequently, the divider is likely to believe that this reduced dependence on the patron would increase the divider’s likelihood of success in influencing the target protégé’s behavior.
What do I mean by the prospect of further worsening of the relations? There are two aims the divider may hope to achieve. First, the divider may hope to create a sense of distrust or betrayal between the target protégé and its patron. The target protégé’s act of receiving positive inducements from the divider could make the patron reassess the protégé’s loyalty, given that the protégé had been relying on the patron for its security. 10 This sense of betrayal might make it more difficult for the rival patron and its protégé to repair the rift. Second, the divider may aim to reduce the dependence of the target protégé on its patron, or, even better still, hope that the target protégé may instead rely on the divider for its security. The target protégé’s recent experience of a deterioration in the relations with the patron presents an opportunistic timing for the divider to demonstrate its goodwill. By sending a costly reassurance signal that it has no hostile intentions, the divider might hope to dealign the target protégé from its patron.
The argument I propose here represents a refinement of the theories on wedge strategy proposed in past research. Past works emphasize the relative reward power of the divider in explaining the conditions under which carrot-based wedge strategies are likely to be attempted. For example, Crawford (2021, 11) argues that a “divider’s willingness to pay costs to influence the target is related to how much advantage it expects to gain by doing so.” Similarly, Izumikawa (2013, 508) argues that the divider is likely to attempt a carrot-based wedge strategy when it enjoys a “reward power advantage” over the main adversary. The theory proposed here suggests that the discussion on the relative reward power of the divider alone is insufficient; the divider would be concerned about the possibility of a “leakage” since the target protégé belongs to a rival alliance. 11 What I argue and subsequently demonstrate with empirical evidence is that the divider would have a greater incentive to pay the costs of influencing the target when the conditions are ripe.
Some readers might argue that the divider would prefer to abstain from taking any initiative and let the relations further deteriorate rather than seek to exploit the rift. The logic of competitive outbidding would suggest that the divider would be concerned with the possibility of the rival patron seeking to offset any wedge initiatives from the divider and such reasoning could induce the divider to abstain from implementing any wedge initiatives. Although such an argument is reasonable, I believe that the mechanism I propose is more plausible. Given that the relations between the target protégé and its patron have recently deteriorated, the latter would not be willing to make concessions to the former lest such actions might be perceived as a sign of weakness. The relationship between the rival patron and its protégé is characterized by its asymmetric nature; the rival patron would not want to be perceived as a loser in its dispute with its protégé. Indeed, as I elaborate below in the subsequent section, a divider may in fact have the opposite concern; in the wake of the Tito-Stalin split, for example, Washington was concerned that Stalin would seek to oust Tito in the absence of Washington’s support.
Others may wonder whether it would be possible for a target protégé to suffer a deterioration in the relations with its patron as well as the divider. I suggest that although this is theoretically possible, it is highly unlikely. A formal minor power protégé that is a member of an asymmetric alliance is likely to have pressing security concerns. In such a context, it does not seem strategically wise to alienate both its patron and the divider at the same time. While alienating the patron does not necessarily mean that the target protégé would befriend the divider, it would at least seek to maintain lukewarm relations. Indeed, as I elaborate in the subsequent section, Tito was cautious to maintain at least lukewarm relations with Washington despite the mutual suspicion.
Given the extant discussion, I propose the following hypothesis:
The deterioration in the relations between the rival major power patron and its protégé is positively associated with the quantity of economic aid offered by the divider.
Research Design
Scope Conditions
I test my argument on triads involving major power rivalries from 1960 to 2010. I consider the major powers of the post-WWII era to be the United States, Russia (the Soviet Union), the United Kingdom, France, and China. The definition used here is broadly consistent with how International Relations scholars conceptualize a “major power.” For example, many studies in the conflict literature studying extended deterrence and alliance focus on these five major powers (Fordham 2010; Huth, Bennett and Gelpi 1992; Huth and Russett 1984). Most recently, Gibilisco and Montero (2022) use this definition in analyzing major power interventions in the post-WWII period. Similar to Gibilisco and Montero (2022), I exclude Germany and Japan from my analysis since they do not have any protégés of their own.
I define a rival as an adversarial major power (and their respective major power allies) that experienced rivalries based on a criterion similar to that of Thompson and Dreyer (2011). Based on this criterion, I identify fourteen directed pairs of major power rivalry, namely the United States-Russia (1960–2010), the United Kingdom-Russia (1960–2010), France-Russia (1960–2010), the United States-China (1960–1971, 2000–2010), the United Kingdom-China (1960–1971, 2000–2010), France-China (1960–1971, 2000–2010), the Soviet Union-China (1960–1989) and their reverse counterparts. 12
Some readers might wonder why I do not examine all rivalries in the post-WWII era. There are a couple of reasons why I limit the scope conditions of the argument. First, there is the issue of data availability. The data for economic aid are available up to only 1960 as the use of economic aid was not a major feature of international politics before the Second World War. Although economic concessions were an important feature of wedge strategies during the Second World War (Crawford 2008, 2021), such data are sparse.
The focus on economic aid as a tool of wedge strategy also means that this study necessarily excludes many regional rivals that do not have developed economies. Clearly, not all states in the international state system have the ability to use economic aid as a foreign policy tool. For example, although Pakistan and India have been in intense rivalry, neither has the economic capacity to use economic aid as a foreign policy tool. The same can be said with respect to many other regional rivalries, such as the rivalry between Nigeria and Cameroon or the rivalry between Iran and Iraq.
Second, many of the regional rivalries take place under the shadow of major power patrons. For example, Argentina and Brazil are both members of the Rio Pact, 13 sharing the United States as the major power patron. Security competition may be less severe as a regional system dominated by a major power may be less anarchic (Butt 2013). Moreover, the overlap in the alliance membership makes it difficult to assess whether a country such as Brazil would have been driven by wedge motivations when giving aid to a Rio Pact member who is allied to both Brazil and Argentina. In contrast, the mutual exclusivity in the alliance networks managed by major power rivals makes statistical analysis more amenable for examining wedge strategies.
I consider the minor power protégés to be the members of a defensive alliance managed by the major powers per Leeds et al. (2002). 14 The five major powers have a number of protégés to varying degrees. Washington has notably possessed the largest number of protégés in the post-WWII era. For this paper, I consider the United Kingdom and France to be both protégés of the United States and major powers in their own right. For countries such as Belgium that are allied to both the United Kingdom or France and the United States, I consider the United States to be the patron. Past studies essentially adopt this framework as they consider the United States to be the security provider to the NATO alliance members. For example, Fuhrmann (2020) considers the conditions under which NATO members (including the United Kingdom and France) would be more willing to free-ride on the defense spending of the United States. There are also some minor power countries that were allied to the United Kingdom or France, but not allied to the United States. For example, France had allies in Africa such as Central African Republic and Senegal (Leeds et al. 2002). The United Kingdom also had allies of its own not allied to the United States such as Nigeria and South Africa. For such countries, I consider France or the United Kingdom to have been the major power patrons as appropriate.
Data and Statistical Model
I conduct a panel analysis consisting of directed dyad-year observations from 1960 to 2010. The sender state refers to a major power divider. The target states consist of protégés of a rival alliance. 15 Based on the preceding discussion, the dependent variable is differenced (logged) economic aid. The data on economic aid for the United States, the United Kingdom and France are drawn from the World Bank. The data on Soviet aid are drawn from Bach (2003). The data on Chinese aid are drawn from Lin (1993) and Dreher et al. (2022). 16
The ideal point distance measures the extent to which two countries share similar preferences, with greater distance denoting more dissimilar preferences. I draw my data on the ideal point distance from Bailey, Strezhnev and Voeten (2017). This measure is appropriate as it denotes “single dimension that reflects state positions toward the US-led liberal order” (Bailey, Strezhnev and Voeten 2017, 430).
The main independent variable of interest is the deterioration in the relations between the adversarial major power patron and its protégé. As highlighted in the description of the dispute between Washington and Ankara, it is the deterioration in the relations over time, rather than the fixed state of relations at any given point in time, that is the main theoretical variable of interest. From the perspective of the divider, how the relations between the target protégé and its patron change is likely to be a more informative signal because protégés with weak linkages are likely to have disharmonious relations with their patron to begin with. This is consistent with the theoretical implications from recent works (Beardsley 2024). For example, in discussing the trilateral dynamics among the United States, Russia and Ukraine, Beardsley (2024, 1428) argues that the “change in congruence is also important to take into account, not just the levels of congruence.”
One way to take into account of both the change and the level of the ideal point distance is to use a standardized measure of ideal point distance and observe the change in this standardized measure from 1 year to the next. Thus, I first the standardize the ideal point distance between a patron and its protégé and then take the difference between the standardized measures at time t and t − 1 to calculate the (standardized) change in relations over time. 17 Intuitively, a greater positive difference in the ideal point distances between the 2 years for a given pair of patron-protégé would signify a greater deterioration in the relations. In regression specifications, I lag this measure (as well as the control variables) by 1 year to alleviate concerns of reverse causality.
I control for various observed confounders that could be correlated with both the deterioration of relations between the patron and the protégé as well as the aid from the divider. The deterioration in the relations between the rival major power patron and its protégé and the divider’s wedge initiatives do not take place in a vacuum: the rival major power patron and its protégé obviously have agency. I thus control for various observed confounders that pose a challenge for inference in this regard. It is likely that the protégé of the rival patron may seek to play off its patron against the divider to receive more aid. Moreover, as noted by Izumikawa (2018), a rival major power patron might use binding strategies, and potentially offer more aid to its own protégé to keep it loyal. I thus control for various economic variables that pertain to such incentives. I first control for the (logged) gross domestic product of the divider and the target protégé states as well as the (logged) total trade between the two countries. These variables could be plausibly correlated with both the tendency of the sender state to offer aid as well as the relations between the target state and its patron. The data on these variables are drawn from IHME (2022) and Barbieri, Keshk and Pollins (2009). For similar reasons, I also control for the aid given by the patron since this could be correlated with the degree of deterioration between the patron and its protégé as well as affecting how much aid the divider gives to the target protégé in response.
Next, I control for the Composite Index of National Capability (CINC) scores of the divider and the target protégé states. While the gross domestic products of the divider and the target protégé take account of the economic variables that could confound the relations, some might argue that these are insufficient since this study is focusing on observations pertaining to security alliances. For example, it is possible that a divider with an overwhelming military strength might not be inclined to attempt wedge strategies because it does not perceive the need to do so. At the same time, a target protégé of the rival alliance might be less willing to let its relations with the patron deteriorate in the presence of such a threatening divider (Walt 1987). Similarly, a target protégé that has sufficient military capabilities might have more incentives to let its relations with the patron deteriorate as it could afford to pursue a more autonomous policy. At the same time, a divider might be inclined to induce such protégés to break away from the patron by offering aid as this would represent a greater loss for the rival patron.
I control for the regime type of the sender state since western democracies such as the United States or the United Kingdom might be more inclined to give aid. At the same time, target protégés that were once illiberal may suffer a deterioration in the relations with their autocratic patron as they become more democratic. I thus control for the regime type of both the sender and the target countries. I draw the data on the democracy score from Pemstein, Meserve and Melton (2010). Higher values denote that a country is more democratic.
Next, I control for the (logged) arms transfers from the divider to the target protégé. Recent research suggests that military aid could be highly correlated with economic aid from the sender state (Boutton 2021). At the same time, the target protégé might become more emboldened after receiving military assistance from the divider and be willing to condone the deterioration in the relations with its patron. The data on arms transfers are drawn from the Stockholm International Peace Research Institute (SIPRI 2023).
I additionally control for variables relevant to the relations between the rival patron and its protégé. First, I control for whether the rival patron initiated a militarized conflict against the target protégé state. Such disputes could potentially be correlated with the deterioration in the relations between the patron and its protégé as well as the divider’s aid. For similar reasons, I also control for sanctions imposed by the major power patron. The data on these two variables are drawn from the Correlates of War MID dataset and Morgan, Bapat and Kobayashi (2014), respectively.
There are three models in the main analysis. In all the models, I use unit fixed effects to control for time-invariant unit heterogeneity. Scholars have warned against the blind use of two-way fixed effects, showing that the interpretation is often unclear (Imai and Kim 2021; Kropko and Kubinec 2020). Thus, the subsequent interpretation relies on models using only unit fixed effects. However, I show that the results are robust to two-way fixed effects in sign and statistical significance in the results below. I cluster the standard errors by directed dyads to account for intra-dyad correlation of errors per the standard practice in the conflict literature (e.g., Sechser and Fuhrmann 2017). 18
Empirical Findings
Main results
Clustered standard errors in parentheses.
+p < 0.10, *p < 0.05, **p < 0.01.
Effect of deterioration in patron-protégé relations on Divider’s economic aid.
Placebo Test
Some readers might wonder whether the above posited relations would hold for non-aligned countries in the international system as the target countries. The theory I propose here does not make any specific predictions with respect to countries that are not aligned to the major powers. Consider, for example, a triadic setting involving the United States, Russia and Kenya. Both the United States and Russia might seek influence over Kenya by offering economic aid. However, it is unlikely that the logic of strategic opportunity would hold in such a scenario since Kenya was neither allied to the United States nor Russia. From the perspective of the United States, there is no strong reason to believe that Kenya would somehow misuse U.S. economic aid in an adversarial manner as it was not of a member of a rival alliance. Thus, we have no a priori theoretical expectation that the deterioration in the relations between Russia and Kenya, for example, should be a strong predictor for explaining the amount of aid given by the United States.
Placebo test: Non-aligned targets
Clustered standard errors in parentheses.
+p < 0.10, *p < 0.05, **p < 0.01.
Placebo test: Non-aligned and aligned targets
Clustered standard errors in parentheses
+p < 0.10, *p < 0.05, **p < 0.01.
Reverse Causality?
Although I have adopted the appropriate lag structure to alleviate concerns of reverse causality, such concerns remain an important issue in observational conflict studies. For example, scholars have difficulty in identifying whether causality flows from mutual trade dependence to peace or vice versa (Hegre, Oneal and Russett 2010; Keshk, Pollins and Reuveny 2004; Kim and Rousseau 2005). This study obviously suffers from similar limitations since it is not a randomized controlled trial. I address these concerns both theoretically and empirically. First, using economic aid to drive a wedge between a rival patron and its protégé is an uncommon behavior in general as there are obvious risks in providing positive inducements to members of a rival alliance (Baldwin 1971; Walt 1987). As an extreme example, it is difficult to imagine that Russia would be able to drive apart the relations between Washington and its NATO allies by offering the latter economic aid. As the case studies below show, the divider is more likely to be responding to external events rather than manipulating them.
The placebo test also highlights why reverse causality seems to make little sense in light of our understanding based on the alliance literature. If reverse causality is indeed the dominant mechanism driving the results, the placebo test results in Table 3 suggest that a divider’s aid to the target state is more effective when the target state is a formal ally of the rival. Given past research arguing that security alliances often comprise states with similar preferences (Benson and Smith 2023; Smith 1995), it seems unlikely that wedge strategy could be more effective for allies than non-allies.
Moreover, if reverse causality is the driving mechanism behind the empirical findings and wedge strategy is as effective as supposed, we should observe more instances of states seeking to implement wedge strategies. However, such predictions seem to be inconsistent with the empirical pattern in past studies. For example, in a slightly different context, Kinne and Bunte (2020) note that Washington failed to provide a single bilateral loan to the Kyrgyz government because of the latter’s deep ties with Russia.
Effect of Divider’s aid on change in standardized ideal point distance between patron and protégé
Clustered standard errors in parentheses
+p < 0.10, *p < 0.05, **p < 0.01.
Robustness Tests and Sensitivity Analysis
I further conduct robustness tests to probe the sensitivity of my results in the appendix. First, scholars might argue that the quantity of economic aid matters little in the context of wedge strategies since it is the symbolic act of transferring economic aid to a protégé of a rival alliance that has political significance. As such, scholars might be interested in whether the act of giving aid itself, rather than the amount, is positively associated with the rapid deterioration of patron-protégé relations. Such an approach would also alleviate concerns about measurement issues regarding the quantification of aid across different dividers. I conduct analysis where I code the variable of giving aid as 1 if a divider offer aids to a target protégé state in a certain year and 0 otherwise, essentially treating it as a binary dependent variable. I conduct analysis using logistic regression with year fixed effects as well as with correlated random effects. Next, given the potential problems of using logistic regression with unit fixed effects, I conduct additional analysis with the linear probability model with unit fixed effects as well as two-way fixed effects. The results are shown to be robust to such specifications.
Second, readers might be concerned that the results shown above might be driven by the behavior of a particular divider. This is especially a big concern since I am only examining five major powers. Moreover, the United States, the United Kingdom and France exhibited similar foreign policy preferences throughout the sample period examined. As such, I conduct further sensitivity analysis to probe the robustness of my results by sequentially dropping each of the five major powers with the same regression specifications to analyze whether any single particular divider is driving the results. I also try excluding observations with missing values of Russian aid. Overall, the results are consistent with those presented in the manuscript. 23
Third, scholars might be concerned about unobserved confounders since this is an observational study. As such, I have conducted sensitivity analysis to examine the extent to which my independent variable of interest is robust to unobserved confounders (Cinelli and Hazlett 2020). The crux of the logic behind this method is to assess the “minimum strength of association that unobserved confounding would need to have, both with the treatment and with the outcome, to change the research conclusions” (Cinelli and Hazlett 2020, 39). Using the patron’s aid as the benchmark covariate, I show that the results are robust and consistent with the theoretical expectations given here.
Fourth, I include additional control variables that may pose problems for inference. First, I control whether the major power divider and the target protégé is jointly democratic. 24 I also control for whether there have been leadership and/or regime transitions in the protégé (Goemans, Gleditsch and Chiozza 2009; Mattes, Leeds and Matsumura 2016). I also include additional control variables relevant to the relations between the rival patron and its protégé such as the power differential, 25 trade (Barbieri, Keshk and Pollins 2009), 26 joint regime type (Pemstein, Meserve and Melton 2010), 27 and defense cooperation agreements between the rival patron and the protégé (Kinne 2020). Some readers might also argue that countries such as France, the United Kingdom and China have been only regional players and that treating these countries as independent major power dividers is problematic even after accounting for homophily in aid endowment. I further conduct tests using standard errors accounting for the interdependence among dyads (Aronow, Samii and Assenova 2015; Carlson, Incerti and Aronow 2024) with the full set of control variables. The results are overall shown to be consistent with the main results provided in the manuscript.
Case Studies
Washington-Belgrade-Moscow, 1948–1959
In this section, I briefly describe three cases highlighting the use of wedge strategy by Washington and Moscow. The first case examines how Washington responded after the rift between Belgrade and Moscow became public in 1948. The second case examines Moscow’s attempt to drive a wedge between Ankara and Washington in the aftermath of the Cyprus crisis in 1963. The third case examines Washington’s attempt to exploit the rift between Tbilisi and and Moscow in the wake of the Cold War. In these case studies, I seek to show that there was a deterioration in relations in a rival alliance and that the major power divider was incentivized in response to use a carrot-based wedge strategy to exploit this rift.
Washington’s use of wedge strategy with respect to Belgrade in the early years of the Cold War has been examined by various historians (e.g., Lees 2010). Moscow and Belgrade formed a defensive alliance in April 1945 (Leeds et al. 2002; Perović 2007). 28 Belgrade was assessed to be “Moscow’s most loyal ally” (Perović 2007, 32). However, the relations between Moscow and Belgrade soon became severely strained as Tito sought to pursue a more autonomous and expansionist foreign policy (Lees 1978; Mehta 2011; Perović 2007). The split became public in June 1948 when the Cominform announced the expulsion of Yugoslavia (Mehta 2011). This represented a rift between Moscow and Belgrade which Washington could potentially exploit. As Robert Reams, the Chargé to the US embassy in Belgrade, remarked in his letter to the Secretary of State on June 30th, 1948, “[n]o event could be more momentous for the attainment of our foreign policy objectives than the permanent alienation from the Soviet of this key regime” (FRUS 1948). The National Security Council made a similar assessment in August 1948, stating that Washington should “promote by means short of war the gradual retraction of undue Russian power and influence from the present satellite area and the emergence of the respective eastern European countries as independent factors on the international scene” (Mayers 1986, 26–27).
Washington was shocked by the split and carefully formulated its policy. There were multiple concerns in Washington. Washington fully recognized that the split did not mean that Tito became Washington’s “friend” (Mitrovich 2009, 37). Kennan argued for a moderate policy toward Belgrade, warning that Washington should neither “beseech” Tito’s favor nor be “too cold toward Tito” as being too cold could present an opening for Moscow to exploit as a propaganda opportunity (Mitrovich 2009, 37).
The possibility of competitive outbidding by Moscow does not seem to have been a significant concern when Washington considered supporting Belgrade. On the contrary, Washington was concerned that Moscow could seek to replace Tito with a “stooge” (Mehta 2011, 122). Such a concern prompted Washington to support Tito. As Alex Dragnich, the Cultural Attaché to the US embassy in Belgrade, remarked, “the only alternative was either to help Tito survive, or let him perish. So the recommendation out of the Embassy was to help Tito survive.” (Mehta 2011, 122) Dragnich’s recount is consistent with the recommendation in the CIA report published in June 1949 which assessed that while Belgrade could probably withstand pressure from Moscow, “[m]ilitary pressure from the East may eventually necessitate, for the purposes of US policy, military and economic assistance which would provide Yugoslavia with some means of self-protection.” (CIA 1949)
The fact that Tito could still be a potential adversary, albeit potentially independent of Moscow’s influence, meant that Washington had to tread carefully. The Policy Planning Staff in the Truman administration recommended the shipping of material “that did not endanger American security” (Lees 1978, 415). Washington also realized that the leverage over Belgrade through its aid would be limited. The Policy Planning Staff concluded in its report in February 1949 that Tito’s attitude “has softened only slightly” as Tito refused to make political concessions (Lees 1978, 414).
In summary, Washington offered aid to Belgrade only after the relations between Belgrade and Moscow deteriorated. There was little political incentive for Washington to offer aid before the rift became public. Moreover, even after the rift became public, Washington was aware of the potential security concerns and the limited influence of its aid. However, given Washington’s security competition with Moscow, the benefit of exploiting the rift outweighed the cost. As Dean Acheson remarked in his conversation with the British Foreign Secretary on September 14, 1959, Tito was now “our son-of-a-bitch” (Gaddis 1987, 159).
Moscow-Ankara-Washington, 1963–1965
Moscow’s attempt to drive a wedge between Washington and Ankara is another important case for probing the validity of my theory. It is well-documented that the Soviet Union offered Turkey aid during the Cold War (Bach 2003). At first, it might be puzzling that the Soviet Union offered aid to Turkey in the 1960’s given that Turkey has been a member of NATO since 1952. What explains Moscow’s motivation in offering aid to Ankara? Past scholarship suggests that Moscow was motivated to drive a wedge between Washington and Ankara (İşçi, Hirst and Bayraktar 2024). Moscow hoped that Ankara would become more like Helsinki in terms of the bilateral relations (İşçi, Hirst and Bayraktar 2024). 29
The relations between the United States and Turkey rapidly deteriorated in the 1960s over the issue of Cyprus. Ankara wanted to intervene in Cyprus when internal violence broke out between the Turkish and Greek Cypriots in 1963. However, Washington expressed its vehement opposition to such an intervention. President Johnson used strong language in his letter in criticizing Turkey’s handling of the crisis, stating “I hope you will understand that your NATO allies have not had a chance to consider whether they have an obligation to protect Turkey against the Soviet Union if Turkey takes a step which results in Soviet intervention without the full consent and understanding of its NATO Allies” (Johnson and Inonu 1966, 387). İsmet İnönü, the Turkish prime minster at the time, responded by stating that Johnson’s “message, both in wording and content, has been disappointing for an ally like Turkey who has always been giving the most serious attention to its relations of alliance with the United States” (Johnson and Inonu 1966, 388). Interestingly, the United States’ own intelligence service assessed that “Johnson’s letter has done more to set back United States Turkish relations than any other single act” (CIA 1964).
After learning of the Western bloc’s opposition to Turkish intervention, Rijov, the Soviet ambassador to Turkey, approached Ankara and offered assistance (Гасымлы, 2008).
30
As Crawford (2003, 119) points out, “the Soviets were more than happy to capitalize on Turkey’s new mood.” The deterioration in the relations between Washington and Ankara were apparent in the exchange of letters between President Johnson and Prime Minister İnönü. Consistent with this account, we observe a large spike in Soviet aid to Turkey in 1965 in Figure 2. Figure 3 also shows that there was a rapid deterioration in the relations between the United States and Turkey around this time where positive (negative) values indicate deterioration (improvement) in relations between the two countries from 1 year to the next.
31
Soviet aid to Turkey. Source: Bach (2003). Deterioration in the relations between U.S. and Turkey. Source: Bailey, Strezhnev and Voeten (2017).
There are two aspects that merit elaboration regarding the two cases described above. First, the protégé state of the rival alliance obviously has agency in influencing the sequence of events. In the case of Yugoslavia, Tito’s inclination to pursue an expansionist foreign policy was an important source of the rift between Belgrade and Moscow (Perović 2007). 32 Similarly, in the case of Turkey, İnönü disclosed his discontent with Washington’s policy to the wider international community. 33 Second, the cause of the dispute that gave rise to the opportunity for the divider to exploit may have deep roots that the divider may not be able to ex ante manipulate. In the case of Yugoslavia, the rift between Belgrade and Moscow was wholly unexpected for Washington (Perović 2007). Likewise, it would be appropriate to describe Moscow’s behavior in the wake of the Cyprus crisis as reactive rather than proactive. Indeed, Figure 2 seems to confirm the account that the Russia’s attempt to drive a wedge occurred as a response to the deterioration in the relations between the United States and Turkey, which in turn, were functions of the political calculations of both Ankara and Washington.
Washington-Tbilisi-Moscow, 1993–2008
Washington’s wedge strategy with respect to Tbilisi suggests that wedge strategies are not mere relics of the Cold War era. Georgia was a member of two defensive alliances led by Russia, namely the Collective Security Treaty Organization (CSTO) and the Commonwealth of Independent States (CIS). 34 Although Tbilisi and Moscow were formal allies, their relations were marked with serious disputes since Georgia’s independence in 1991 as Russia was involved in Georgia’s wars in South Ossetia and Abkhazia (Nichol 2008; NYT 1993). Not surprisingly, Russia was blamed for Georgia’s loss of control over these regions (Benjamin 2005).
Tbilisi’s motivations behind Georgia’s membership for NATO were clear given the history of conflicts with Moscow. Shevardnadze declared at the 1992 NATO summit in Prague that Georgia was willing to eventually join NATO (Nichol 2009; Shevardnadze 2002). Tbilisi under Shevardnadze sent costly signals that it sought to form closer relations with Washington, sending troops to Iraq in 2003 (Liklikadze 2012). The Saakashvili administration continued to seek closer ties with Washington by sending troops to Afghanistan starting in 2004 (Georgia 2013) and committing to deploy more troops in Iraq (Kramer 2007). Although such commitments did not represent a quid pro quo for Georgia’s formal NATO membership, many Georgians implicitly understood this to be the case (Kramer 2007).
The Bucharest NATO summit in April 2008 marked an important step in Georgia’s aspiration for NATO membership. President Bush declared that “NATO should welcome Georgia and Ukraine into the Membership Action Plan” (Bush 2008). Although Washington alone could not dictate Georgia’s accession to NATO and there was opposition from Berlin and Paris (Erlanger and Myers 2008), Russia initiated its invasion of Georgia in August 2008.
35
In response, Washington committed to increase its aid to Georgia, as shown by the spike in Figure 4. The Bush administration responded by announcing in September that Washington will increase aid to Georgia (Office 2008) and subsequently confirmed this commitment in Brussels in October (Henrietta 2008). Consistent with the logic behind my strategic opportunity argument, the relations between Russia and Georgia were already severely strained by the time the United States offered Georgia substantial aid and membership in NATO. US aid to Georgia.
Conclusion
Why and when do major power dividers attempt wedge strategies? I have argued in this paper that major power dividers attempt wedge strategies when there is a rift in the relations between the rival patron and its protégé. I have tested my argument using a mixed-method of descriptive case studies and statistical analysis and have found some empirical evidence to support my argument. Operationalizing economic aid as a carrot-based wedge strategy, the statistical analysis shows that a rapid deterioration in the relations between the rival protégé and its patron is positively associated with the divider offering more economic aid to the protégé.
There are two main contributions of this study. First, I have sought to address when major powers are likely to attempt wedge strategies. Although wedge strategy is an important component to understand in alliance politics, scholars have devoted less attention to the topic compared to other questions on alliance management. However, a complete understanding of alliance politics such as the question of abandonment would only be better understood when we consider the role of the divider (Crawford 2021).
This paper makes a contribution to the wedge strategy literature by proposing a theory on the conditions under which major power dividers are likely to use carrot-based wedge strategy. While much of the conventional view provided in the wedge strategy literature expounds on the benefits of carrot-based wedge strategy in countervailing situations (Crawford 2011; Izumikawa 2013), this paper proposes an alternative view by arguing that it is likely to be attempted in a reinforcing fashion as well. In addition, the argument I propose suggests that dividers who usually lack reward power may still be able to adopt carrot-based approaches by exploiting a rift in the rival alliance.
Second, this study contributes to our understanding of the conditions under which major powers give economic aid. Although scholars have certainly noted that major donors give economic aid out of strategic motivations (e.g., Bearce and Tirone 2010; Carter and Stone 2015), there has been a lack of attention in the International Political Economy literature on how economic aid features in major power competition. I argue and show that the patron-protégé relations were an important factor for the endowment of aids for certain countries.
There are important limitations to this study. Although the use of foreign aid has been the focus of this study, wedge strategy is a broad concept that encompasses many tools. The nature of the target would affect the type of tool a major power divider would use. For example, scholars have noted that Russia sought to drive a wedge among EU members with its energy supply (Wigell and Vihma 2016). 36 For future research, scholars may wish to consider more carefully the types of tools a divider would be likely to employ when seeking to exploit a rift in the rival alliance.
There are other important questions on wedge strategies that could be addressed for future research. The obvious question to investigate is the effectiveness of wedge strategies. There are various contexts through which such a question may be answered. For example, political analysts have noted that China is attempting to gain influence in various parts of the world at the expense of the United States. Under what conditions are such attempts likely to be effective? Answering such questions requires deep theoretical analysis and presents significant challenges. However, this remains a topic that should be further studied, given its theoretical and practical significance.
Supplemental Material
Supplemental Material - Major Power Rivalry and Wedge Strategy of Concessions
Supplemental Material for Major Power Rivalry and Wedge Strategy of Concessions by Hankyeul Yang in Journal of Conflict Resolution.
Footnotes
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
The replication package can be found at https://doi.org/10.7910/DVN/7HO970 (Yang, 2026).
Supplemental Material
Supplemental material for this article is available online.
Notes
References
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.
