The Institutional Origins of Miscalculation in China's International Crises

When is China prone to miscalculate in international crises? Scholars have long noted that miscalculation is an important pathway to international conflict.¹ When decision-makers inaccurately assess the balance of power, the strength of adversary resolve, or the feasibility of their policy options, they are prone to choose strategies ill-suited to achieving their goals. Inaccurate assessments can thus lead decision-makers into international conflicts at the end of which they are no better off, and often much worse off, than before the confrontation began. As China considers potential strategies to advance its interests in the coming years—such as a military invasion of Taiwan, the enforcement of its territorial claims in the South China Sea, and economic coercion against its neighbors—its choices will have tremendous consequences. Susceptibility to miscalculation while making these decisions could mean the difference between war and peace.

China's record of miscalculation in international crises has varied considerably. From 1949 to 2012, China was party to twenty-six international crises involving the potential or actual use of military force. In many of these crises, Chinese decision-makers accurately assessed the situations that they confronted and the feasibility of available policy options. In over one-third of cases, however, China chose to initiate or escalate violent and costly international crises on the basis of inaccurate assessments. Although existing studies on China's use of force² often recognize that Chinese leaders have miscalculated in the past, they have not systematically studied when and why they are likely to do so.³

I argue that one important explanation for China's miscalculations centers on what I term its national security institutions: the rules that shape the flow of information between leaders and their diplomatic, defense, and intelligence bureaucracies. These institutions can be thought of as bridges between leaders and their advisers. How these bridges are designed shapes the roles, constraints, and expectations of key bureaucracies to relay quality information to the leader during decision-making. Thus, the information available to China's leaders as they decide between peace and conflict often hinges on their institutions at the time.

I argue that variation in China's national security institutions helps to explain when and why China's leaders are more likely to miscalculate as they choose whether to initiate or escalate an international crisis. I develop a theoretical framework differentiating between three types of institutions: integrated, fragmented, and siloed. Integrated institutions reduce the risk of miscalculation by building capacity for bureaucrats to relay information and by promoting information sharing between bureaucracies. Increased capacity to search for information allows leaders to make decisions with access to more information than they could collect on their own. Competitive dialogue between bureaucrats motivates them to provide information more candidly and with less organizational bias. Although other factors can still cause leaders to miscalculate, integrated institutions tend to increase the availability of quality (complete and accurate) information on which leaders may base their decisions.

In contrast, two pathological types of institutions raise the risk of miscalculation because information that decision-makers need to assess their crisis prospects gets trapped or distorted at lower levels of the state. Each type creates a distinct pathway to miscalculation. Fragmentated institutions increase the risk of miscalculation by restricting bureaucratic advisers' access to decision-making. Exclusion raises the costs of relaying bureaucratic information and discourages advisers from speaking their minds. Siloed institutions increase the risk of miscalculation by restricting information sharing between bureaucracies. Lack of information from outside a bureaucratic adviser's silo impairs their ability to judge the value of their own data and to point out when peers are relaying inaccurate information. As leaders judge whether initiating or escalating an international crisis will advance their goals, those sitting atop fragmented or siloed institutions thus have access to lower-quality information than those sitting atop integrated institutions.

To probe the plausibility of the theoretical framework, this article offers the most comprehensive analysis to date of when and why China is likely to miscalculate in an international crisis.⁴ I first present a medium-N analysis, which systematically codes China's performance across its twenty-six international crises from 1949 to 2012. I find that inaccurate assessments occurred more frequently when China possessed fragmented or siloed institutions than when it possessed integrated ones.⁵ I also find that China was less likely to achieve its crisis objectives under fragmented and siloed designs than under integrated ones, suggesting that assessment quality may have shaped Chinese leaders' ability to prospectively steer away from conflicts that were unlikely to advance their goals. I then use case studies examining the 1962 Nationalist invasion scare, the 1969 Sino-Soviet border conflict, and the 2001 EP-3 reconnaissance aircraft incident to illustrate the theoretical mechanisms underlying integrated, fragmented, and siloed institutions. The analysis draws on a wealth of newly available primary sources, Chinese-language memoirs, records from foreign ministry and provincial archives, interviews with Chinese interlocutors, and an original dataset measuring bureaucratic participation across four decades of national security decision-making.⁶

This article makes two contributions to the study of Chinese foreign policy and international relations theory. First, the theory and findings illuminate how institutional miscalculations can lead China to initiate or escalate costly conflicts because of low quality information about its prospects for success. This institutional pathway to international conflict complements existing accounts that emphasize shifts in China's bargaining position and the balance of power.⁷ Yet the findings also suggest that fragmented and siloed institutions can lead China into confrontation in part because leaders miscalculate how the balance of power shapes the ability to achieve their goals through the threat or use of military violence.

Second, the article contributes to the broader study of how bureaucracies shape foreign policy in authoritarian regimes by making an important distinction between national security institutions and personalist regimes.⁸ The logic of personalism suggests that states are more likely to miscalculate when leaders amass unchecked power such that other political elites cannot hold them accountable for poor foreign policy outcomes.⁹ Leaders are thus prone to initiate conflicts without carefully evaluating the costs and benefits because their political survival is assured even if the conflict goes poorly. Some apply this logic to China, citing Mao Zedong as a prototypical example of how personalist rulers choose misguided foreign policies and pointing to China's shift to collective rule after his death as the progenitor of wiser choices.¹⁰ Yet existing theories of personalism are incomplete, in part because they tend to view leader accountability as synonymous with the quality of information that leaders receive. While many personalist regimes feature pathological information provision from the bureaucracy, the case of China illustrates that even personalist regimes can cultivate comparatively effective institutions. Conversely, non-personalist regimes can still feature informational pathologies if they have suboptimal institutions.

This article proceeds in seven sections. The first section introduces key definitions and the typology of national security institutions. The second section measures China's performance in twenty-six international crises from 1949 to 2012, presenting a medium-N analysis that provides evidence consistent with the typology's predictions. The subsequent three sections trace the evolution of China's national security institutions and illustrate the theoretical mechanisms through three case studies of China's crisis decision-making. The sixth section briefly considers alternative explanations. Finally, the conclusion summarizes the findings and discusses policy implications.

Following existing definitions, I use the term “international crisis” to describe situations in which one state takes an action that creates a perceived, time-sensitive threat to another state's interests and that raises the possibility of military hostility.¹¹ I assume that leaders (presidents, prime ministers, and dictators) have agency in how crises unfold. Initiating or escalating crises can be an attractive strategy for decision-makers to advance their goals. Because crises can also impart material and reputational costs, however, leaders generally prefer to avoid crises that fail to advance those goals beyond the status quo.

Leaders face systematic challenges in prospectively identifying whether crises are likely to further their objectives. If decision-makers could perfectly predict how crises would turn out, they could steer away from confrontations in which they fail to achieve their goals.¹² Uncertainty, however, opens up the possibility that leaders initiate or escalate crises that do not turn out as expected. I use the term “crisis miscalculation” to denote a situation in which an inaccurate assessment causes a leader to initiate or escalate an international crisis that ultimately fails to advance their goals. In short, leaders choose international confrontation because they judge that the crisis will deliver more than it ultimately does. Decision-makers may sometimes stumble into success despite inaccurate assessments—or fail to achieve objectives despite accurate assessments—but I show in the next section that these alternative combinations are comparatively rare in China.

Quality information is critical to avoiding crisis miscalculation. While the international system sets limits on what any state actor can know, leaders can make the best of their uncertain environment by building state capacity to search for information available in the international system and by setting procedures for how that information is processed within the state's organizational system.

Leaders can manage the tasks of searching for and processing information needed for crisis decision-making in systematically different ways. The formal and informal rules that establish roles, constraints, and expectations for bureaucracies performing these functions, which I refer to as national security institutions, can either enhance or degrade the flow of quality information between bureaucracies and from bureaucracies to leaders.¹³ Broadly, national security institutions can be divided into three stylized types: integrated, fragmented, and siloed.¹⁴

Integrated institutions leverage formal and informal mechanisms to encourage high vertical information flow between the leader and their bureaucracies, as well as high horizontal information flow between the bureaucracies. First, they feature high levels of participation from multiple types of bureaucracies on committees, councils, and bodies that advise the leader. While each crisis is different, the functional demands of international crises tend to require participation from bureaucracies that specialize in diplomatic, defense, and intelligence affairs. Second, integrated institutions utilize coordinators and coordination bodies to monitor and enforce information sharing between bureaucracies. In integrated institutions, both advisory and coordination bodies follow predictable routines that reduce the transaction costs of acquiring, processing, and distributing information within the state.

I identify three reasons why the patterns of information flow established by integrated institutions tend to provide quality (i.e., complete and accurate) information on which leaders can base their choices. First, integrated institutions improve capacity to search for information in the international environment. Leaders face cognitive and resource limitations that affect their ability to personally perform the time-intensive tasks of information search and processing.¹⁵ Divisions of labor within the government allow bureaucrats to specialize and develop domain-specific expertise. Second, by elevating the status of bureaucratic advisers through appointments to key advisory bodies, integrated institutions reassure bureaucrats that leaders seek accurate, rather than self-confirming, information. Third, facilitating inter-bureaucratic information sharing allows bureaucrats to police one another. Access across organizational divides allows bureaucrats to question each other's information. Knowing that others will scrutinize their analysis motivates bureaucracies to expend effort to acquire and report information capable of withstanding the competitive dialogue.¹⁶ Leaders with integrated institutions thus tend to enjoy access to more complete and less biased information.

In contrast, fragmented institutions limit the participation of diplomatic, defense, and intelligence bureaucracies in decision-making, which degrades the quality of information available to leaders as they choose between peace and conflict.¹⁷ First, fragmented institutions decrease capacity to search for information. Disbanding or removing bureaucratic advisers from a decision-making committee, for example, makes it comparatively difficult for them to relay information that they collect to the leader. Leaders sitting atop designs that exclude diplomats may not receive information on adversary resolve collected from foreign embassies, whereas those that exclude defense advisers may not receive information on the balance of capabilities. Second, by limiting access to decision-making and lowering the status of bureaucratic advisers, fragmented institutions confirm fears that leaders might punish those who are not pliant. In response, advisers who are able to gain access to leaders pursue sycophantic strategies in which they withhold or distort the information they provide to conform to the leader's prior beliefs. Such information degradation is common to many regime types, but it can be particularly pernicious in dictatorships, where leaders can more severely punish bureaucrats.¹⁸ In sum, by creating low-vertical, low-horizontal information flows within the state, fragmented institutions tend to deliver information that is incomplete and biased toward what leaders already believe.

Siloed institutions permit bureaucracies to advise leaders, but they limit the capacity and authority of coordinators and coordination bodies to facilitate information sharing between bureaucracies. Restricting horizontal information sharing in this way has two second-order effects on the quality of information available to leaders. First, siloed institutions impede a bureaucracy's ability to identify whether certain data are worth relaying to a leader when the data's value depends on contextualizing information that only other bureaucracies possess. For example, two bureaucracies may each possess a bit of information suggesting that an adversary's intentions have become more malign, but no single piece of information provides enough evidence to merit reporting it to the leader. If the system is siloed, both bureaucracies may withhold such information because each underestimates its value.¹⁹ Second, siloed institutions undermine competitive dialogue between bureaucrats. Because information is not shared horizontally, bureaucrats know that others are not able to question the information that they provide, which reduces incentives to expend effort to find higher-quality information and increases the risk that bureaucracies distort information to suit their parochial interests. In sum, by creating high-vertical, low-horizontal information flows within the state, siloed institutions tend to deliver information that skews toward the interests of the bureaucracy providing it.

I argue that leaders sitting atop fragmented or siloed institutions are more prone to crisis miscalculation than those sitting atop integrated ones. As a leader contemplates initiating or escalating a crisis, information readily observable in the international system either never reaches the leader or is distorted when it does. This missing or distorted information impairs the leader's ability to accurately judge whether they can achieve their goals through international confrontation. Leaders may thus initiate or escalate international crises on the basis of low quality information about the potential to achieve their goals. In contrast, leaders with integrated institutions can base their decisions on more complete and accurate information about their prospects for success, pushing them to deescalate international tensions based on clear-eyed estimates of what conflict would yield. Thus, leaders with fragmented or siloed institutions tend to more frequently find themselves in international conflicts that fail to reach their objectives because leaders base their initial choices on less complete and more biased information.

It is worth emphasizing that fragmented and siloed institutions are not the only potential sources of miscalculation.²⁰ Decision-makers can draw inaccurate assessments owing to numerous other reasons, ranging from psychological biases to bad luck. Yet my typology suggests that decision-makers may cope with these constraints in more and less effective ways depending on their institutions. Institutional design shapes whether leaders are able to make the best of the imperfect decision-making environment that they confront. If the theory is correct, over time integrated institutions should exhibit better patterns of performance relative to fragmented and siloed ones.

In addition, fragmented and siloed institutions should produce different types of miscalculations. Siloed systems tend to provide leaders with more information than fragmented systems, which might reduce the magnitude of leaders' assessment errors. Although siloed institutions tend to have biased information that may cause leaders to initiate crises, leaders may play bureaucrats off one another to uncover inaccuracies in the information that bureaucracies provide as the crisis unfolds. Thus, I expect miscalculation to be more likely under both fragmented and siloed institutions than integrated ones—but miscalculations to be less costly under siloed institutions.

Finally, while the origins of national security institutions remain outside the scope of this article, two possible factors are briefly worth noting. First, fragmentation may make sense politically, particularly during periods in which leaders fear a coalition between bureaucrats and political opponents. Second, some leaders may lack the political strength to compel the bureaucracy to accept an integrated system. Siloed institutions tend to appeal to bureaucrats because they afford unilateral access to the leader while also maintaining the bureaucracy's autonomy by minimizing coordination with other bureaucratic organizations. Thus, weak leaders may prefer integrated institutions but settle for siloed ones.

China was involved in twenty-six international crises from 1949 to 2012. Chinese decision-makers made inaccurate assessments in only fifteen of these crises and failed to achieve their primary goal in only twelve of them.²¹ Changes in national security institutions help to explain this variation.

Table 1 summarizes China's assessment quality, primary goal, primary outcome, and battlefield costs for each crisis. I compiled the list of crises using secondary sources on China's use of force and crisis management, as well as the International Crisis Behavior and Militarized Interstate Disputes datasets. First, I recorded China's goal(s) before the crisis began. I identified at least one goal for each crisis, but for crises that featured multiple goals, I noted what the existing literature suggests was the primary one. This approach allows me to analyze crisis performance in two ways: first by examining the share of goals that decision-makers achieved, and then by examining whether they achieved their most important one. Second, for each crisis I coded whether China achieved its goals and the number of China's battlefield casualties. Finally, I documented whether there is evidence of an inaccurate assessment during decision-making. Potential types of erroneous assessments include inaccurate projections of battlefield costs, adversary responses, and the trends underlying China's motivation for action. Section 3 of the online appendix provides narrative summaries of each coding, details regarding sample construction, and coding rules.²²

I based my coding of China's national security institutions on the observable indicators listed in table 2. As described in subsequent sections, I find that China's institutions were generally integrated from 1949 to 1962, fragmented from 1963 to 1981, and siloed from 1982 to 2012 (see also section 2 in the online appendix). I used several types of data to develop these codings. First, I drew on primary and secondary sources that describe vertical and horizontal information flow. Second, I leveraged records of the daily activities of Chinese Communist Party (CCP) elites to build an original dataset, entitled the PRC National Security Decision-Making Dataset (see the online supplementary materials), which systematically measures characteristics of meetings in which party elites considered issues pertaining to foreign and defense affairs. For each of the 1,953 meetings in the dataset, I coded the forum (e.g., Politburo, Central Military Commission [CMC], informal), as well as whether diplomatic and defense bureaucracies participated.²³ I analyzed these data to measure trends in de facto participation in decision-making and coordination. Third, I examined CCP organizational records at both the advisory level (leader-bureaucracy interactions) and the coordination level (bureaucracy-bureaucracy interactions). At the advisory level, I coded bureaucratic representation on key bodies (such as the Politburo), which can facilitate vertical flow of information between leaders and the bureaucracy. Bureaucratic advisers who are appointed to these bodies can channel information to the leader and contribute to policy deliberations when these bodies meet. Appointments also confer higher status to these bureaucratic advisers, which may shape their willingness to speak their minds during deliberations. These dynamics may further shape incentives for advisers to provide quality information during informal meetings or through reports outside face-to-face interactions. At the coordination level, I coded bureaucratic appointments on key coordination bodies, which facilitate horizontal information flow between bureaucracies.²⁴ Such appointments also afford bureaucratic advisers with access to key reports and cables that are circulated among the institutionally affiliated party elites. I also coded the presence of coordinators who possess formal or informal authority to monitor and facilitate inter-bureaucratic information.

These data suggest three patterns consistent with the typology's predictions.²⁵ First, the data support the theory's contention that integrated institutions are less likely to produce inaccurate assessments than fragmented or siloed ones. Inaccurate assessments occurred in only two of nine crises during China's integrated period (1949–1962) compared with all eight crises during its fragmented period (1963–1981) and five of nine crises during its siloed period (1982–2012).²⁶

Second, China's pattern of crisis outcomes aligns with the framework's predictions. The framework expects that because integrated institutions produce higher quality information, they will exhibit better patterns of crisis performance over time relative to fragmented and siloed ones: leaders are better informed as to whether initiating or escalating a crisis will advance their goals, allowing them to steer away from confrontations unlikely to do so. Consistent with this intuition, China achieved its primary goal in eight of nine crises during its integrated period. In contrast, it achieved only two of eight primary goals in crises under fragmented institutions and four of nine in crises under siloed ones.²⁷ Accounting for the fact that decision-makers typically had more than one crisis goal, China on average achieved 89 percent of its crisis goals when its institutions were integrated, but only 36 percent and 41 percent of its goals when they were fragmented and siloed, respectively. Crisis outcomes appear to be related to assessment accuracy. On average, China achieved 37 percent of its goals during crises in which there was evidence of an inaccurate assessment, but 85 percent of its goals when there was no such evidence of inaccuracy.

Third, congruent with the typology's predictions about the costs of crisis miscalculation, inaccurate assessments were more costly to China under fragmented institutions than under siloed institutions. During its fragmented period, China paid battlefield costs in seven of eight crises in which there was an inaccurate assessment. By comparison, during its siloed period, China was able to avoid escalation to violent conflict in all but one of the crises in which there was an inaccurate assessment.²⁸

Taken together, the typology correctly predicts the assessment accuracy, outcome, and level of cost in seventeen out of China's twenty-six international crises. While national security institutions are not deterministic, they help to explain broad patterns in China's crisis behavior.²⁹

While medium-N analysis is useful in identifying general patterns, there are multiple ways to interpret why performance varied across China's three institutional periods. Thus, the remaining sections in this article illustrate theorized mechanisms using three cases: the 1962 Nationalist invasion scare, the 1969 Sino-Soviet border conflict, and the 2001 EP-3 reconnaissance aircraft incident. There are particularly rich data available on China's decision-making for each of these episodes, which allows for in-depth and vivid illustrations of how institutions shaped crisis decision-making. In each case, I first describe China's institutions during the period as a whole, establishing the organizational design and pattern of information flow. I then describe the background of the specific crisis, China's crisis goals and outcome, and the accuracy of its assessments. Finally, I analyze whether assessment quality can be traced back to China's national security institutions, providing evidence of the impact that institutions had on searching for and processing information.

As summarized in table 2, the theoretical framework would expect China's leaders to receive higher quality (more complete and more accurate) information from the bureaucracy under integrated institutions than under fragmented and siloed institutions. Furthermore, the reasons underpinning informational pathologies should be different. Under fragmented institutions, the theoretical framework expects that China's bureaucracies will fail to deliver information because of either degraded capacity or fear of political retaliation. Under siloed institutions, the framework expects that bureaucracies will submit information that suits their parochial interests—and that other bureaucracies will fail to challenge these biases because they either lack access to information or have few opportunities to debate across bureaucratic silos.

In addition to illustrating the theoretical mechanisms, this set of cases is helpful for several reasons. First, these cases demonstrate the full range of variation in the explanatory variable, including integrated, fragmented, and siloed institutions. Second, these cases help to consider the effects of national security institutions alongside alternative explanations, such as leader personality and the balance of power. Differences in institutional performance during the 1962 invasion scare and the 1969 Sino-Soviet border conflict illustrate how institutions can shape crisis decision-making under the same leader. Similarly, in all three cases, China faced a materially superior adversary or alliance—and the first two cases occurred after the Sino-Soviet split, when China was no longer backed by a materially stronger ally. These similarities are helpful because the balance of power might prompt China to be cautious, choosing to avoid confrontations with a stronger opponent. Alternatively, the balance of power might prompt China to adopt more belligerent strategies to demonstrate resolve.³⁰ Finally, as discussed in more detail below, two of these cases are puzzling from the perspective of theories emphasizing how higher levels of leader accountability by elite audiences might improve crisis decision-making. That is, China exhibited comparatively effective decision-making under Mao Zedong's strongman leadership during the 1962 invasion scare but dysfunctional decision-making under collective leadership during the 2001 EP-3 incident.

The People's Republic of China's national security institutions were generally integrated for more than a decade under its first leader, Mao Zedong. In its institutional design, bureaucratic participation in decision-making was high. The CCP Politburo included both senior defense and senior diplomatic officials.³¹ Coordination bodies were similarly inclusive. The CMC included China's early foreign ministers, Zhou Enlai and Chen Yi. Zhou also facilitated inter-ministerial coordination through the State Council, on which China's senior diplomatic and military leadership both sat at the time. Beginning in the mid-1950s, the CCP Secretariat, chaired by Deng Xiaoping, coordinated and oversaw defense, international, and intelligence affairs.³² The Secretariat included senior figures from the military and diplomatic bureaucracies, as well as party leaders managing China's intelligence organization, the Central Investigation Department (CID).³³ A series of so-called leading small groups—inter-ministerial bodies focused on managing specific issues, such as foreign affairs and Taiwan—further bolstered information sharing capacity.³⁴

China's integrated institutions featured high vertical information flow. Figure 1 reports the annual average number of formal and informal meetings attended by Mao Zedong in which the agenda touched on defense or foreign policy.³⁵ The top panel demonstrates that from 1950 to 1962, Mao convened the Politburo over ten times per year on average to discuss these issues. The bottom panel illustrates that both military and diplomatic officials participated. Party records suggest that Mao's advisers provided information and reports during at least 40 percent of these meetings. In May 1953, for instance, Mao instructed Defense Minister Peng Dehuai and People's Liberation Army (PLA) Chief of General Staff Huang Kecheng to provide routine reports on CMC activities to the Politburo “once or twice every month.” Mao asked that the CMC write a report, add it to the Politburo's agenda, and have the CMC leadership “give a short explanation” during a meeting. Mao instructed that other departments, presumably including the foreign ministry, should follow the same procedure.³⁶ These data also show that part of Mao's national security decision-making during this period occurred in informal settings. Critically, however, even informal meetings featured bureaucratic participation from both diplomatic and defense advisers. The status and authority established by formal appointments and integration into formal decision-making likely spilled over into informal interactions as well.³⁷

Integrated institutions featured generally high horizontal information sharing. Some inter-ministerial coordination occurred through formal bodies, such as the CMC, State Council, and Secretariat. Even after resigning from the CMC near the end of the Korean War, Zhou Enlai continued to periodically chair CMC meetings throughout the 1950s and 1960s.³⁸ From 1953 to 1962, formal meetings of the State Council or Secretariat touching on foreign and defense affairs occurred about once per month on average. Informal interactions, afforded by Deng Xiaoping's and Zhou Enlai's status and authority within the party, frequently facilitated horizontal coordination as well. Both military and diplomatic representatives attended the majority of these meetings.

In total, the available evidence suggests that the early Mao period featured quality information provision. Bureaucratic advisers offered differing advice and deliberated across bureaucratic lines before Mao Zedong reached his decisions.³⁹ Senior officials were more than simply Mao's “yes-men.”⁴⁰ They could gain “trust and access” with Mao and gave “important counsel” on international affairs.⁴¹ Diplomatic and defense bureaucracies offered information and intelligence on important matters that “affected decision-making” at the highest levels.⁴²

Formal and informal meetings among Politburo members played a role during most of China's international crises in this period: China's entry into the Korean War in 1950, the 1954 and 1958 Taiwan Strait Crises, the 1959 Sino-Indian border clashes, the 1962 Nationalist invasion scare, and the 1962 Sino-Indian War. Senior military leader Lin Biao argued against China's involvement in the Korean War and offered potential ways to mitigate risk of escalation during the 1958 Taiwan Strait Crisis.⁴³ Foreign Minister Zhou Enlai proved critical in nudging Mao Zedong toward a less confrontational policy regarding Vietnam, Taiwan, and the United States in the mid-1950s.⁴⁴ Mao consulted with both military and diplomatic personnel inside and outside Politburo meetings during confrontations with India in 1959 and 1962.⁴⁵ The CMC similarly played a role in coordinating among various bureaucracies during most of these crises. Prior to the Korean War, for example, party leaders encouraged bureaucrats to review intelligence from other departments. One Chinese source notes that military leaders maintained “close contact” with senior diplomatic and intelligence officials through “sharing intelligence and exchanging views.”⁴⁶

Several additional points are worth noting. First, although the early Mao period featured a comparatively high level of horizontal coordination, siloes existed at lower levels of the bureaucracy. Second, the density of social ties among party members from the revolutionary generation likely helped to integrate China's institutions. These social ties may have both shaped Mao Zedong's ability to adopt more integrated institutions and facilitated information flow directly. Finally, one of the most catastrophic economic policy failures in Chinese history—the Great Leap Forward, launched in 1958—occurred when China's national security institutions were integrated. The roots of this failure are beyond the scope of my theory, but this point emphasizes why it is important to distinguish between national security institutions and those charged with handling other aspects of government decision-making.

China's decision-making during the 1962 Nationalist (Kuomintang, KMT) invasion scare illustrates how integrated institutions reduced the risk of crisis miscalculation by providing quality information to the leader.⁴⁷ Since 1949, China has maintained a territorial claim over Taiwan, which remained controlled by KMT forces after the Chinese civil war subsided.⁴⁸ The KMT similarly maintained plans to “return to the mainland” through military attack, subversion, or regime change. The economic disaster of the Great Leap Forward during the late 1950s led KMT leader Chiang Kai-shek to believe that a major military operation against the mainland might succeed. In his diary, Chiang commented that there was “clear proof” that popular support for the Communist regime was waning, and the people hoped to be “rescued” by the KMT.⁴⁹ He felt this was his “only favorable opportunity” for offensive action and ordered preparations for a series of attacks against the mainland.⁵⁰ In early 1961, Chiang and his senior advisers drafted plans for a surprise mainland attack.⁵¹ From February to May 1962, Chiang oversaw numerous military planning meetings and told Vice Chief of General Staff Lai Ming-tang that the first stage of the operation would involve a large-scale airdrop of approximately 200–300 KMT soldiers, followed by an amphibious landing.⁵²

china's strategy, goals, and crisis outcome. China's response to the threat of KMT military action had two principal components. First, in early June, China deployed roughly 100,000 elite troops to southeast China and placed them on alert. Second, later that month, Chinese diplomats probed U.S. intentions and signaled China's concern about the threat. In so doing, China sought to both deter a KMT invasion and limit U.S. support to KMT military actions against the mainland.

China achieved both crisis goals at minimal cost. In July, one National Security Council (NSC) staff member noted that the White House had “made a mistake” in supporting Nationalist proposals.⁵³ The U.S. ambassador in Taiwan wrote that the United States hoped to “put on brakes in matters of large scale drops of 200 or more.”⁵⁴ By November 1962, the NSC Policy Planning Council noted that the United States would avoid any large-scale military operations against the mainland, and would insist that the Nationalists do the same. The United States would attempt to convince Taiwan that “the only circumstance in which we envisage the possibility of their return to the mainland would be one wherein they had” been invited to do so.⁵⁵ As one historian notes, U.S. pressure on Taiwan “played a considerably important role” in the “sudden halt” to Nationalist military plans in the summer of 1962.⁵⁶

effects of integrated institutions. I identify two ways that integrated institutions shaped Mao Zedong's decision-making during the 1962 Nationalist invasion scare by providing quality information from bureaucrats. Without the accurate information that integrated institutions supplied, Mao might have made erroneous judgments that would have supported escalating the confrontation with the United States, perhaps by attacking the offshore islands and exercising the “noose” strategy that Mao had adopted during the 1958 Taiwan Strait Crisis.⁵⁷ Instead, China was able to “avoid formal misjudgment” and sidestep what would have likely been a more costly outcome.⁵⁸

First, integrated institutions provided capacity to search for information about U.S. intentions, which were critical to the KMT's ability to attack because of the Mutual Defense Treaty between the United States and Taiwan. Although President John F. Kennedy was initially open-minded about the prospects of military action, the unfolding crisis quickly turned the administration against KMT plans.⁵⁹ In late June, the United States approved a diplomatic message to be delivered to China in Warsaw to ensure that it did not “act under a misunderstanding of [U.S.] intentions.”⁶⁰ At the diplomatic exchange, the U.S. ambassador stated that the United States “had no intention of committing or supporting aggression against [China] anywhere.”⁶¹ Further, the U.S. ambassador emphasized that the KMT invasion would require U.S. consent, which it would not provide.⁶² In his report to Beijing, Chinese envoy Wang Bingnan concluded that there “were in fact contradictions between Washington and Taipei” on the issue of attacking the mainland, such that the United States would not “risk releasing Chiang for fear of disaster.”⁶³ According to Wang, China's military maneuvers made the United States anxious, even though it perceived them as defensive.⁶⁴ In his memoirs, Wang reflected that this information directly affected Mao Zedong's decisions:

Second, the information that the bureaucracy provided showed no signs of systematic bias toward Mao Zedong's prior beliefs. In fact, initially, Mao seems to have been concerned that the United States would support a much larger Nationalist offensive than the Kennedy administration was considering. In May, Mao estimated that the size of the Nationalist invasion force could be as large as 400,000, and he discussed the need to prevent both a Nationalist assault along the southeastern coast and airborne operations into central China.⁶⁶ China's large-scale military mobilization further suggests that Mao took the threat seriously.⁶⁷

Integrated institutions provided information that seems to have helped Mao Zedong revise his beliefs. After receiving reports of the KMT's preparations, Mao asked PLA Chief of General Staff Luo Ruiqing to convene the CMC to discuss the matter.⁶⁸ Inter-ministerial assessment and coordination began immediately.⁶⁹ On May 30 and 31, both senior military and diplomatic advisers provided input during the CMC meeting to discuss the possibility of a KMT attack.⁷⁰ On the basis of Foreign Minister Chen Yi's personal opinion shared with the Soviet ambassador earlier that month, it seems that Chen believed that U.S. support would not be forthcoming.⁷¹ Chen assessed that the United States might not have approved all the KMT's proposals. “It is more likely,” Chen explained, “that Chiang Kai-shek has ignored U.S. restraints,” and that the United States would “dare not take the risk” of a major war with China.⁷² The issue was subsequently discussed at a special meeting of the senior party leadership on June 1. Senior military officials briefed the party leadership at Politburo meetings on June 2 and June 6—and again in conjunction with Chen on June 20.⁷³ By the end of the assessment process and “after careful deliberation,” the party leadership reached the conclusion that the KMT's choices hinged on whether the United States would support Chiang's plan, and that the United States was not prepared for even a limited conflict with China.⁷⁴

Beginning in late 1962, China shifted toward fragmented institutions that lasted until the early 1980s. Bureaucratic participation in national security decision-making decreased. Mao Zedong first changed the locus of decision-making from the more inclusive Politburo to its more restrictive Standing Committee. Party records suggest that, from 1950 to 1962, over two-thirds of Politburo meetings touching on either foreign or defense affairs were regular or expanded sessions that included a range of bureaucratic representatives. From 1963 to the eve of the Cultural Revolution in 1965, however, participation in over three-quarters of Politburo meetings was restricted to the Standing Committee, which excluded China's diplomats. Bureaucratic participation further decreased after the launch of the Cultural Revolution in 1966. The top panel of Figure 1 shows that Mao set aside the Politburo almost completely.⁷⁵ Even after Mao's death in 1976, the Politburo did not routinely consider national security issues and lacked an appointed representative from the foreign ministry when it did.

Particularly during the Cultural Revolution, China dramatically decreased its bureaucratic capacity for information collection. Nearly three-quarters of diplomatic personnel were recalled from abroad and more than 70 percent of diplomatic personnel in Beijing were sent to the countryside to learn from the example of the peasants, and by 1967 China had only one ambassador stationed abroad.⁷⁶ One of China's principal coordinators, Deng Xiaoping, was removed from his leadership positions. Even Zhou Enlai, who retained his position, lost much of his administrative support to oversee inter-ministerial coordination.⁷⁷ As Mao Zedong told Zhou in June 1968, Chen Yi “has already been brought down and you are nearly there [too].”⁷⁸ Months later, Zhou commented that he could not “find a single minister” to help him to manage international affairs.⁷⁹ Senior intelligence organizations, such as the CID, were similarly targeted—and the CID's think tank was shut down.⁸⁰ Although the CMC was at first isolated from the chaos, Mao downgraded its organizational status in March 1968 to an “administrative body” (banshizu) and ordered the CMC Standing Committee to stop convening.⁸¹ The Foreign Affairs Leading Small Group was not reestablished until 1981 and did not include representatives from the defense or intelligence bureaucracies until the mid-1980s.⁸²

China's fragmented system featured comparatively poor vertical information flow. For one, there were simply fewer chances for bureaucratic advisers to relay information to leaders. The bottom panel of Figure 1 depicts the dramatic decrease in diplomatic participation in meetings with Mao Zedong after 1962. Diplomatic reporting outside these meetings also declined.⁸³ After Mao's death, the Politburo and CMC also appear to have convened less routinely than during Mao's early years. Diplomatic participation in decision-making remained low in both formal and informal settings.

These dynamics degraded the quality of information that Mao and his successor, Hua Guofeng, received from the bureaucracy. One Chinese historian notes that after the disciplining of senior diplomats in the fall of 1962, “Mao's subordinates did not dare to suggest changes in Chinese foreign policy.”⁸⁴ According to firsthand accounts, China's diplomatic corps became filled with “sycophants.”⁸⁵ Mao Zedong noted that he did not “trust the reports” from the foreign ministry.⁸⁶ In 1975, one senior official commented that the military was failing to properly perform its advisory function.⁸⁷ Military leaders did not “dare to speak their minds” out of fear.⁸⁸ Even after Mao's death, the information provided by the bureaucracy continued to exhibit systematic biases. Senior military officers recollected that strategic assessment was “a forbidden area” that only the leader was qualified to consider.⁸⁹ These dynamics were present among China's diplomats as well, who sought to “cater” to the leader by playing a type of “intelligence game,” in which bureaucrats tried to determine the leadership's beliefs in order to adjust their reports to match them.⁹⁰

China's decision-making during the Sino-Soviet border conflict illustrates how fragmented institutions raised the risk of high-cost crisis miscalculations by providing information that was incomplete and skewed toward the leader's prior beliefs. In the 1960s, the Soviet Union and China were locked in a broad competition for international leadership within the Communist bloc.⁹¹ In addition, the two sides had unresolved territorial disputes regarding features in the Amur and Ussuri rivers. In the late 1960s, clashes erupted along the border, which sources suggest were instigated by both sides.⁹² In the beginning of 1967, border violations by Chinese authorities increased, including forward patrolling on the islands.⁹³ A series of border clashes and standoffs occurred from January 1968 to February 1969.⁹⁴

china's strategy, goals, and crisis outcome. China's strategy in response to growing tensions with the Soviet Union was to initiate a violent interstate conflict that directly targeted its militarily superior adversary. On March 2, 1969, China ambushed Soviet forces on Zhenbao Island—a contested territorial feature along China's northeastern border with the Soviet Union—resulting in ninety-one Chinese and about two hundred Soviet casualties.⁹⁵ In initiating the crisis, China had three main goals: to deter a potential Soviet attack against China; to deter the Soviet military build-up along China's border, guarding against a long-term security challenge; and to secure its territorial control over Zhenbao Island.⁹⁶

China's strategy backfired. Rather than reducing the risk of Soviet aggression, the Chinese ambush at Zhenbao increased it. The Soviets retaliated against the Chinese positions at Zhenbao and, after diplomacy failed, escalated the conflict by initiating fighting along the western border. In fact, only after the Chinese ambush did Moscow begin to contemplate serious offensive action.⁹⁷ The clashes prompted Moscow to sound out Washington's reaction to a surgical strike inside China.⁹⁸ Even after the crisis abated, the Soviet Union expanded the number of Soviet infantry divisions stationed on China's border by approximately 50 percent from 1969 to 1973.⁹⁹ These trends continued through the 1970s, as the Soviet Union held military exercises, including the largest ever proximate to the Chinese border. While China's tactical hold ver Zhenbao Island improved as a result of the conflict, border talks with the Soviet Union from October 1969 to June 1978 resulted in few concessions from Moscow's negotiators. China continued to protest Soviet military border violations after the dispute.¹⁰⁰ Moreover, the Soviet Union had offered to cede control of the island in 1964 as part of a border settlement, which was nearly identical to the one that the two countries eventually signed in 1991.¹⁰¹

inaccurate assessments. The assessments that led Mao Zedong to approve the ambush were inaccurate in three ways. First, Mao overestimated the threat of Soviet aggression. Mao believed that the United States and Soviet Union were colluding to oppose China, which might lead to an attack by one or both. In October 1964, Mao suggested that the Soviet Union might use military force to occupy Xinjiang, Heilongjiang, or Inner Mongolia.¹⁰² The Soviet invasion of Czechoslovakia in August 1968 exacerbated Mao's fears.¹⁰³ On November 28, 1968, Mao contended that both the United States and the Soviet Union had “the capacity to start a war” and feared that one might begin.¹⁰⁴ In early March 1969, Mao told his inner circle that the Soviets might seek to occupy “Europe and Asia, including inside China.”¹⁰⁵ According to Mao, the United States and the Soviet Union had colluded in the past, expanded their arms, and hoped to divide the world anew, and thus the northern areas of China should be prepared for a Soviet and U.S. invasion.¹⁰⁶

Yet there is little evidence to suggest that the Soviet Union intended to launch a military strike against China before the border clashes in March—and certainly not one in collaboration with the United States. Entries in Soviet leader Leonid Brezhnev's diary from late 1968 to early 1969 contain no hint of plans for an offensive. Instead, Brezhnev wrote that he preferred to downplay the border skirmishes if possible.¹⁰⁷ On the contrary, Soviet leaders feared that China's numerical superiority might lead China to attack the Soviet Union.¹⁰⁸ Indeed, Chinese military officers later concluded that Beijing had “overestimated the threat” of Soviet offensive action.¹⁰⁹

The second inaccurate assessment was that the Zhenbao ambush would reduce the chance of a Soviet invasion by demonstrating China's resolve.¹¹⁰ Chinese military historian Xu Yan notes that Mao Zedong anticipated that the Chinese attack at Zhenbao would not escalate into a wider conflict.¹¹¹ On the contrary, Mao seems to have believed that the Soviet Union would perceive China's ambush as a defensive measure. At a meeting with his inner circle, Mao noted that the Soviet Union knew that China would not invade “as it is so cold there.”¹¹² Chinese historian Yang Kuisong notes that the border conflict eventually deteriorated to levels “beyond Mao's worst expectations.”¹¹³

The third inaccurate assessment was that there were few immediate prospects for the United States to help China address the Soviet threat. Even after Richard Nixon assumed office in January 1969, Mao Zedong's comments in private suggest that he did not foresee a direct path to rapprochement. On February 7, weeks before the Zhenbao ambush, Mao told his inner circle that “there might be a few changes, but for the time being there will not be any major changes” in U.S.-China relations.¹¹⁴ Even before entering office, however, Nixon was keenly interested in improving bilateral relations with China. Although National Security Advisor Henry Kissinger later claimed that the border clashes helped shape his own views on improving relations with Beijing, his writing at the time suggests that the White House was already “implicitly moving in this direction” before the crisis began.¹¹⁵ As early as January 11, the U.S. NSC was considering options for adjusting its policy toward China.¹¹⁶ On February 1, 1969—one month before the Sino-Soviet border conflict began—Nixon commented that the United States “should give every encouragement” that the administration was “exploring possibilities of rapprochement with the Chinese.”¹¹⁷

effects of fragmented institutions. Fragmented institutions help to explain why Mao Zedong based his strategic choices on low-quality information. Similar to the 1962 Nationalist invasion scare, information was available in the international system that contradicted Mao's initial beliefs. But fragmented institutions impeded the effective aggregation of this information before hostilities in two ways.

First, fragmented institutions degraded China's capacity for signal reception.¹¹⁸ Beginning in 1966, for instance, the United States had sent signals that it was open to warmer relations. Chinese Ambassador Wang Guoquan noted at U.S.-China ambassadorial talks in Warsaw that he picked up on hints that “the United States was moving towards recognizing” the Communist regime and that “implied a clear change in direction of U.S. domestic and foreign policy.”¹¹⁹ After the PRC ambassador reported the change in the U.S. side back to Beijing, however, Wang's report “did not receive sufficient emphasis or timely discussion.”¹²⁰ As a result, China “missed a favorable opportunity to move U.S.-China relations forward in a timely fashion.”¹²¹ This missed opportunity was directly tied to China's fragmentated institutions. Wang was recalled from Warsaw in the summer of 1967 because of the Cultural Revolution.¹²² Wang suggests that if China had not missed this opportunity, its foreign policy toward the United States might have adjusted sooner, such that normalization of U.S.-China relations “would perhaps not [have] been delayed until 1972.”¹²³

There were similar failures to effectively receive and process signals indicating that Soviet intentions were less malign than Mao Zedong feared. Shortly after the border clashes began, according to Chinese diplomats the Soviet leadership “tried to directly approach the Chinese leadership in order to defuse the conflict and seek out the Chinese attitude before making a decision.”¹²⁴ But calls from Brezhnev and Soviet Premier Alexei Kosygin requesting to speak with the Chinese leadership were refused by Chinese bureaucrats.¹²⁵ During one attempted call, the Chinese operator cursed the Soviet Union, saying, “Revisionist! Who do you think you are to talk with our great leader?”¹²⁶ In the words of one Chinese historian, “a potential dialogue at the highest levels between China and the Soviet Union had so come to an untimely end.”¹²⁷

Indeed, China's diplomats quietly assessed that there were limits to the Soviet threat. In December 1968, shortly before China initiated the border clashes, Chen Yi assessed that the U.S.-Soviet collusion was exaggerated.¹²⁸ Others in the foreign ministry noted that “the situation inside and outside the Soviet Union was not advantageous” for conflict and that Soviet leaders “would not dare to keep fanning the flames of war.”¹²⁹ China's diplomats also inferred that the Soviet Union was desperately attempting to “understand China's attitude,” implying the need for prudence and engagement.¹³⁰

Yet fragmented institutions restricted the foreign ministry's access to decision-making. According to Mao Zedong, there was no discussion within the Politburo prior to the Zhenbao ambush.¹³¹ Chinese diplomats recognized that Chen Yi retained his post in name only.¹³² The PRC ambassador in Moscow had been recalled, and the diplomatic activities of those remaining behind were “reduced to a minimum.”¹³³ The foreign minister did not attend the more important meetings during the 9th Party Congress in April 1969, nor did he or other foreign ministry leaders participate in many of the key diplomatic engagements in the months before the crisis.¹³⁴

The second way that fragmented institutions contributed to inaccurate assessments was through providing bureaucratic information that was systematically biased toward Mao Zedong's prior beliefs. The political climate of the Cultural Revolution “blocked free discussion,” such that China's diplomats “did not dare express genuine ideas” for fear of making an “ideological mistake.”¹³⁵ Even the famous “Four Marshals group”—comprised of senior officials whom Mao instructed to assess the international situation in 1969—quietly questioned its ability to offer new ideas, noting that it would be “easier said than done to put forward a different way of thinking.”¹³⁶ Specifically, the group worried that any suggestions for alternative thinking would be viewed as dissenting with Mao's policies promulgated at the 9th Party Congress.¹³⁷

The Chinese military may have catered their information to Mao Zedong's beliefs as well. Over the summer of 1969, the military leadership echoed the conclusions that Mao had drawn in the spring. The Chinese military continued to “overestimate the possibility of war” and assessed that “the possibility of a large-scale invasion by the Soviet Union” was high.¹³⁸ In addition, the military likely lacked information from other government bureaucracies that might have attenuated these conclusions.¹³⁹ The military and the Four Marshals group reportedly had little interaction.¹⁴⁰ Similarly, according to one senior officer, the Chinese military lacked intelligence indicating that the Soviet leadership was divided and that many in Moscow were advocating for diplomatic dialogue with China—information the foreign ministry seems to have known.¹⁴¹

In sum, fragmented institutions help explain why Mao Zedong miscalculated during the 1969 border conflict with the Soviet Union. Mao initiated the crisis on the basis of low-quality information. With different institutions, Mao might have had more complete and accurate information regarding the severity of the Soviet threat, the urgency with which action was required, and how the Soviet Union would react to China's provocations. Such information might have allowed Mao either to steer away from costly confrontation with the Soviet Union or to choose an alternative strategy that better advanced his goals.

In the 1980s, China's national security institutions shifted toward a siloed design. By the 12th Party Congress in September 1982, the decision-making of China's new leader, Deng Xiaoping, exhibited balance between the participation of defense and diplomatic advisers.¹⁴² Decision-making authority returned to the Politburo, which began convening routinely beginning in the late 1980s.¹⁴³ The Politburo's agenda frequently included international topics ranging from Taiwan policy to relations with Vietnam.¹⁴⁴ From the mid-1980s to the early 2000s, the Politburo included representatives from both the military and diplomatic bureaucracies. By the late 1990s, China's leaders also sat atop bureaucratically inclusive leading small groups as well.¹⁴⁵

These institutional features facilitated high vertical flow of information between leaders and the military, diplomatic, and intelligence bureaucracies.¹⁴⁶ The Politburo provided a forum for routine briefings by the bureaucracy.¹⁴⁷ The position of bureaucracies, particularly the foreign ministry, became “respected, assured, and protected.”¹⁴⁸ Hence, different views were expressed in a more “systematic and institutionalized” manner.¹⁴⁹

Despite the increased access of China's bureaucracies to the political leadership, several aspects of their new institutional design created obstacles to effective information sharing between them. China's siloed design featured “no formal communications mechanism” between the military and key civilian bureaucracies, such as the foreign ministry.¹⁵⁰ China's top leaders—Deng Xiaoping, Jiang Zemin, Hu Jintao—oversaw defense decision-making primarily through formal and informal interactions with the CMC leadership. There was also no formal or informal coordinator between the two silos. Unlike in the Mao era, when his subordinates Zhou Enlai and Deng Xiaoping informally coordinated between the defense and diplomatic sectors, China's contemporary leaders kept this role for themselves. Only the senior-most party leaders were “allowed to touch military affairs.”¹⁵¹ While the Foreign Affairs Leading Small Group may have facilitated coordination across civilian and defense silos, it lacked authority to oversee the CMC.¹⁵² For most of this period, the military representative on the Foreign Affairs Leading Small Group was the defense minister, whose role did not include military operations or intelligence.¹⁵³ The CMC vice chairmen who managed military operations were not included.

Restricted horizontal information flow, particularly between the military and the foreign ministry, degraded the quality of information that leaders received.¹⁵⁴ Civilian and military reporting channels were separate and autonomous, allowing the CMC to directly transmit intelligence and policy recommendations to the leader. The siloed design made it difficult for other bureaucracies to vet or contest the military's information.¹⁵⁵ Consequently, subordinates regularly submitted inaccurate reports to superiors that reflected their parochial interests and perspectives.¹⁵⁶

China's decision-making during the 2001 EP-3 incident illustrates how siloed institutions raised the risk of low-cost crisis miscalculations by providing leaders with information biased toward bureaucratic interests.¹⁵⁷ On April 1, 2001, a Chinese fighter jet and a U.S. military reconnaissance aircraft collided near Hainan Island, killing the Chinese pilot. The U.S. crew made an unapproved emergency landing in Hainan. A crisis began when the Chinese government denied U.S. requests to promptly return the U.S. crew and plane, instead holding the crew and the plane until April 12 and July 3, respectively.

china's strategy, goals, and crisis outcome. Guidelines promulgated by the Chinese government, as well as accounts of private diplomacy between U.S. and Chinese diplomats, suggest that China had four objectives for denying U.S. requests to immediately release the crew and the plane. The first was to convince the United States to cease “further hostile reconnaissance” flights along China's border.¹⁵⁸ CCP General Secretary Jiang Zemin's first public statement on April 3 demanded that the United States “stop such [reconnaissance] flights in China's coastal airspace.”¹⁵⁹ Foreign Minister Tang Jiaxuan similarly told U.S. Ambassador Joseph Prueher on April 4 to “stop reconnaissance activities,” noting that the flights were “the root cause of the collision.”¹⁶⁰ Second, China sought U.S. admission of responsibility for causing the collision. On April 2, Assistant Foreign Minister Zhou Wenzhong demanded that the United States bear “full responsibility” for the collision.¹⁶¹ Jiang reiterated these demands publicly on April 3.¹⁶²

Third, China sought to extract an apology from the United States, although this goal is not specifically mentioned in Chinese sources describing the central government's guidance. On April 4, Tang Jiaxuan told Ambassador Prueher that China would consider allowing the crew to depart if the United States “admitted its mistake in clear-cut terms and apologized.”¹⁶³ Finally, China sought to balance the first three goals with consideration for “the overall situation of Sino-American relations.”¹⁶⁴ At the time of the crisis, Chinese leaders were attempting to shape attitudes within the George W. Bush administration, which had entered office with more pessimistic views of China than those of the Bill Clinton administration.¹⁶⁵

China's strategy during the crisis did not advance its first two goals. U.S. reconnaissance missions resumed in early May 2001, and U.S. negotiators maintained that China bore responsibility for the collision.¹⁶⁶ Although an official letter from U.S. Secretary of State Colin Powell to Vice Premier Qian Qichen acknowledged that the United States was “sorry” for the death of the pilot and that clearance to enter Chinese airspace had not been granted, the document's language fell short of Beijing's demands.¹⁶⁷ After the release of the letter, Powell clarified that the United States had done nothing wrong and “therefore, it was not possible to apologize.”¹⁶⁸ Finally, the incident reinforced U.S. apprehensions concerning China's trustworthiness and confirmed the need for surveillance against China.¹⁶⁹ According to one U.S. official, the crisis was “central” in “hardening views on China” within the George W. Bush administration.¹⁷⁰

inaccurate assessments. Chinese sources make clear that Jiang Zemin based his initial decisions on inaccurate information. Specifically, the Chinese military reported that a “sudden change in direction” by the U.S. plane caused the crash.¹⁷¹ Presumably based on this information, China's officials repeatedly claimed that the collision occurred because the U.S. aircraft had “rammed” the Chinese fighter.¹⁷² While the technical features of the two planes make this assertion implausible, the CMC reported this account to Jiang without allowing other bureaucracies to vet or contextualize it.¹⁷³ This lack of vetting likely led Jiang to believe that China had more leverage over the United States than it actually did. After all, the United States had apologized to China when a U.S. missile accidentally bombed the Chinese embassy in Belgrade in 1999. If the United States was again in the wrong, an apology might easily follow.¹⁷⁴

effects of siloed institutions. Siloed institutions help to explain why the information on which Jiang based his decision was inaccurate in ways that suited the military's organizational interests. Chinese interlocutors suggest that inaccurate information stemmed from China's “stovepiped” intelligence system during the incident.¹⁷⁵ Jiang Zemin himself chided the military afterward, questioning its “strategic awareness” and ordering it to “reflect” on the incident.¹⁷⁶ A subsequent review of the EP-3 incident by a think tank affiliated with Chinese intelligence noted that one of the key functions that the U.S. NSC provided—and, implicitly, that the Chinese system lacked—was a mechanism for “comprehensive processing” of intelligence to ensure that “senior decision-makers do not respond improperly” in the early stages of a crisis.¹⁷⁷

Indeed, the central government had information at the time suggesting that the military's reports were not credible. In December 2000, the U.S. military attaché in Beijing had protested that routine Chinese aerial intercepts had become increasingly risky—that is, they were flying dangerously close to U.S. aircraft.¹⁷⁸ But this information does not seem to have been raised by diplomatic or intelligence bureaucracies during deliberations before or during the crisis. It is likely either that this information stayed in the military's silo, or that civilian bureaucracies believed that they lacked authority to comment on such defense matters.¹⁷⁹

In sum, siloed institutions help to explain Jiang's miscalculation during the EP-3 crisis. On the one hand, Chinese sources suggest that that the information provided to Jiang might have been different under a different set of institutions, which might have attenuated the incident's fallout.¹⁸⁰ In the words of one Chinese source, the “negative impact of the crisis on bilateral relations might have been less serious and enduring” if China had expeditiously released the crew.¹⁸¹ On the other hand, siloed institutions also help to explain why the crisis was not even more costly. By mid-April, senior Chinese decision-makers reportedly recognized that the Chinese pilot caused the incident—and that the United States was unlikely to end its reconnaissance program.¹⁸² In addition, China's bargaining position changed “after it became clear” that it would not be able to compel the United States to take full responsibility and end its surveillance flights.¹⁸³ One plausible explanation for this change to China's bargaining demands is that Chinese decision-makers eventually benefited from a variety of bureaucratic perspectives that siloed institutions afforded. Failure to ultimately arrive at these assessments might have led China to prolong the detention of the U.S. crew, which likely would have imposed an even greater cost on China's bilateral relations with the United States.

This section highlights three alternative explanations for the observed patterns in China's crisis miscalculations. One alternative explanation argues that the threat of removal motivates leaders to make better decisions, as policy failures may jeopardize their political careers. When personalist dictators amass absolute power, they are prone to assessment errors because they know that they cannot be removed from office, which decreases their incentives to search for better information.¹⁸⁴ This logic offers an incomplete explanation for the case of China. Using Mao Zedong's death as the divide between personalist and non-personalist rule, inaccurate assessments occurred in seven of fourteen crises under personalist rule but in nine of twelve crises under non-personalist rule.¹⁸⁵ Thus, while personalism might shape a leader's propensity for interstate violence generally, it does not appear to explain China's miscalculation patterns.

Another variant of this line of reasoning is that fragmented and siloed designs are simply personalism in another form. Yet national security institutions are conceptually distinct from personalism in two ways. First, the traditional focus of personalism is on the leader's political power. Barbara Geddes, Joseph Wright, and Erica Frantz define personalist dictatorship as an authoritarian regime “in which the leader has concentrated power at the expense of his closest supporters.”¹⁸⁶ In contrast, national security institutions are concerned with capacity to channel quality information from the bureaucracy to the leader. Not only might personalist regimes have effective institutions that search for and process information—but non-personalist regimes might not. Second, personalism tends to be concerned with the relationships among political leaders, whereas national security institutions are concerned with relationships with the bureaucracy. While party elites outside the bureaucracy might weigh in on policy debates, their influence is often limited because their expertise and responsibility rest outside national security affairs.

A second alternative explanation might be derived from existing literature on how the balance of power shapes China's crisis behavior. Military superiority, for instance, could cause China to become more risk acceptant, selecting into crises that have a lower probability of success because it is confident of victory on the battlefield even if the confrontation escalates. Yet the frequency of miscalculation was highest from the mid-1960s to the early 1980s, well before China's contemporary military modernization began. Moreover, several of China's crisis miscalculations featured a more powerful adversary (e.g., the Soviet Union in 1969; the United States in 1954, 1964, and 2001) and, in other cases, the balance of power was simply unclear (e.g., India in 1967 and 1986; Vietnam in 1979).¹⁸⁷

A related interpretation might instead emphasize negative shifts in power or bargaining position.¹⁸⁸ In such frameworks, Chinese decision-makers might initiate a crisis knowing that it was unlikely to succeed because they expected that the prospects of achieving their goals would decline in the future. Two points militate against this interpretation. First, Chinese decision-makers perceived negative power shifts in most of China's international crises—both those in which it did and those in which it did not miscalculate.¹⁸⁹ Second, while such a theory might explain unfavorable crisis outcomes, it cannot explain why assessments were inaccurate. This explanation would have more explanatory power if decision-makers assessed that they were unlikely to succeed, but neither the medium-N analysis nor the case studies support this conclusion.

A third alternative explanation is that crises that fail to achieve national security objectives might still be attractive to leaders if they can mobilize domestic audiences in the process. Yet domestic mobilization offers an incomplete explanation. Mobilization does not appear to have been a factor in many of China's crisis miscalculations (e.g., Sino-Indian border clashes in 1967 and the Sino-Indian border standoff in 1986; the 2001 EP-3 reconnaissance aircraft incident). For those cases in which mobilization might have been a goal, national security institutions still help to explain why Chinese decision-makers held inaccurate assessments about their prospects for achieving their domestic political objectives. Similarly, institutional type might also shape China's ability to gauge whether decision-makers can effectively exploit international crises for domestic purposes. For instance, domestic mobilization appears to have been more successful during the 1958 Taiwan Strait Crisis than in the 1969 confrontation with the Soviet Union.¹⁹⁰

Changes in national security institutions help to explain when and why China has been prone to miscalculate in international crisis since 1949. Beijing's leaders are less likely to miscalculate when they have access to quality information that integrated institutions enable and encourage diplomatic, defense, and intelligence bureaucracies to provide. Chinese leaders are more likely to err during periods in which fragmented or siloed institutions undermine the flow of quality information from these bureaucracies. While both fragmented and siloed institutions tend to raise the risk of error, however, the costs of miscalculation under fragmented institutions tend to be higher than those under siloed ones.¹⁹¹

National security institutions will be an important factor shaping both whether China is likely to miscalculate in the future and how costly such miscalculations might be. Particularly since the 20th Party Congress in October 2022, Xi Jinping has amassed considerable political power. Yet the theory and findings suggest that the key question is not whether personalist rule has returned to China, but rather how Xi's uncontested power might shape patterns in information flow. While it is difficult to draw definitive conclusions about China's contemporary institutions, several trends are worth discussing.

On the one hand, Xi's consolidation of power may have helped to push through institutional reforms to address problems with horizontal information sharing between bureaucrats. Xi's most significant reform was to create the Central National Security Commission in 2013, which was likely intended to curb institutional silos. Since 2017, Xi has also appointed China's State Councilors for Foreign Affairs (first Yang Jiechi and then Wang Yi) to the Politburo. These promotions are noteworthy, as neither Yang's predecessor nor recent foreign ministers have been Politburo members, although all have been appointed to the key advisory bodies for foreign affairs.

On the other hand, it is unclear how much Xi Jinping's reforms have positively shaped information sharing between bureaucracies. Xi seems to have reenforced his direct authority over and communication with the CMC, which indicates that information flow between the military and the foreign ministry may remain as siloed below the party apex as before Xi's reign. In addition, an incident in 2023 in which a Chinese espionage balloon traversed U.S. territory raises new questions about whether the information that might have led Xi to approve such a gambit stemmed from the same type of siloed decision-making that China exhibited during the 2001 EP-3 reconnaissance aircraft incident. The problems that these pathologies pose, however, might not be especially severe. My typology suggests that siloed institutions would make China prone to low-cost miscalculations akin to the 2001 EP-3 incident rather than high-cost miscalculations akin to the 1969 Sino-Soviet border conflict.

Yet there are some signs that Xi Jinping's consolidation of power may be more fundamentally reshaping his relationship with the national security bureaucracy. According to some accounts, Xi's decision-making has been comparatively insular. For example, Xi has reportedly utilized the Central National Security Commission primarily to manage domestic issues rather than to oversee international ones. In addition, Xi's anti-corruption campaign has included some high-profile targets inside China's military, which could make the surviving bureaucrats fear similar punishment if they displease the leader. Similarly, Xi's promulgated writings now cover a broad range of issues, including foreign and defense affairs, which may curtail the space in which China's bureaucrats feel comfortable offering new and candid information.

These are significant changes that could undermine the quality of information that Xi Jinping receives. In and of themselves, however, they may not be as destructive as China's fragmented institutions during the 1960s and 1970s. Changes that would signal a more complete shift toward fragmentated institutions may include if Xi were either to suspend key advisory bodies (e.g., the Foreign Affairs Commission) or to remove senior diplomatic and senior defense officials from these bodies. Other possibilities would be if Xi were to appoint senior diplomatic and senior defense advisers with minimal professional credentials—or to remove these same officials for offering candid policy advice. Absent such changes, it is unclear whether contemporary China will be prone to the types of miscalculations that characterized its worst period of national security decision-making.

For their helpful feedback and comments, the author wishes to thank Risa Brooks, Hyun-Binn Cho, Thomas Christensen, Zack Cooper, Fiona Cunningham, Alexandre Debs, M. Taylor Fravel, Avery Goldstein, Kelly Greenhill, Mariya Grinberg, Michael Horowitz, Alastair Iain Johnston, Sean Lynn-Jones, Nuno Montiero, Aroop Mukharji, Reid Pauly, Stephen Rosen, Austin Strange, Jessica Weeks, Alex Weisiger, and Jiakun Jack Zhang, as well as workshop participants at Yale University, Columbia University, the University of Pennsylvania, and the Fairbank Center for Chinese Studies at Harvard University. Caroline Deitch provided excellent research assistance translating Russian-language materials. This project was determined exempt by the Harvard University Committee on the Use of Human Subjects (Protocol No. IRB15-2010). Online supplementary materials are available at https://doi.org/10.7910/DVN/VF9X6A.