Liberalization and Foreign Policy

Liberalization and Foreign Policy, by Miles Kahler, editor

1. Democratic States and Commitment in International Relations

Kurt Taylor Gaubatz

The Four Hundred . . . departed widely from the democratic system of government. . . . They also sent to Agis, the Laced�monian king, at Decelea, to say that they desired to make peace, and that he might reasonably be more disposed to treat now that he had them to deal with instead of the inconstant commons.

--Thucydides ₁

Confederations are dissolved for the sake of some advantage, and in this republics abide by their agreements far better than do princes. Instances might be cited of treaties broken by princes for a very small advantage, and of treaties which have not been broken by a republic for a very great advantage.

--Machiavelli ₂

The traditional view of popular government as shifting and unreliable, which Thucydides attributes to the Athenian oligarchs, has a long and distinguished history. Machiavelli, who takes issue with this view, attributes it to "all writers" and "all historians." ₃ The significant, if still somewhat tenuous, worldwide trend toward democratization of the past decade has renewed interest in the implications of democratic governance for the international behavior of states. ₄ Most of that interest has focused on the relationship between democracy and conflict. But, as is made clear in the work of both Thucydides and Machiavelli, there are long-standing debates about other important dimensions of the international behavior of popular governments. I return here to the basic question suggested by Thucydides and Machiavelli as to the ability of democratic states to make commitments in their international relations. I argue that there is both a theoretical and an empirical basis for rejecting the traditional view of the "inconstant commons."

The ability of states to make commitments is a critical dimension of the international system. Between two states, commitments run the gamut from formal defense treaties to casual assurances between diplomats. For liberal institutionalists, the ability to make commitments is central to the process of international institutionalization. ₅ But commitments do not have to reflect only cooperative behavior. Even for realists, the ability to make commitments is critical to international interactions. The efficacy of deterrence threats and the functioning of alliance politics clearly hinge on the ability of actors to make credible commitments. ₆

The dominant assumption in the study of international relations has been that the ability, or the lack of ability, to make commitments is a function of the anarchic international system. ₇ Recent work in conflict studies, combined with the apparent trend of democratization, has raised anew the question of whether there are theoretically worthwhile distinctions to be made in analyzing the international behavior of different regime types. ₈ Given the importance of commitment and the traditional concern about the inconstancy of popular rule, the possibility that liberal and democratic domestic political and economic arrangements may have distinct effects on the ability of states to make credible international commitments would seem well worth investigating.

On the face of it, the challenge of signaling and maintaining commitment in political systems that require public deliberation and approval for major international actions would seem formidable. But the relationship between international commitments and domestic politics is more complex than might be assumed from a narrow focus on the idea of "the inconstant commons." In this essay I set out a working definition of liberal democracy and draw out of that definition several implications for the ability of states to make international commitments. As against the common perspective of democratic inconstancy, I argue that there are both normative and structural characteristics of liberal democratic states that can significantly enhance the strength of their international commitments. I then turn to a consideration of democratic alliance behavior as a preliminary empirical indicator for the distinctive nature of democratic commitments in the international system. In particular, I bring forward strong empirical evidence to show that alliances between liberal democratic states have proven more durable than either alliances between nondemocratic states or alliances between democratic and nondemocratic states.

Democracy and commitment are both complex phenomena. Many books have been written on both subjects. For the purpose of this analysis, I offer working definitions that, while inadequate as complete philosophical statements, can serve as the basis for a discussion of these phenomena within the context of international affairs.

A state makes a commitment to a course of action when it creates a subjective belief on the part of others that it will carry through with that course of action. There are trivial commitments that involve doing things that are clearly in one's interest to do. The more interesting commitments are those that bind the state to take some set of actions that do not look to be in its narrow self-interest as an international actor. ₉ Thus, the commitment problem for the United States in using nuclear deterrence to defend Europe against a Soviet attack was how to make it clear to both the Europeans and the Soviets that in the event of a war American leaders would be prepared to sacrifice New York in order to save Berlin or Paris. ₁₀ In this essay I will deal in particular with alliance commitments. Alliances, at their core, are a reaction to the problem of nontrivial commitment. ₁₁ If the narrow self-interest of one alliance partner would be served by defending the other, the two would not need to formalize their commitment on paper, beyond some minimal efforts to coordinate defense policies and practices. The creation of a formal alliance is an attempt to signal both to the alliance partners and to other states that a genuine commitment to some level of mutual defense exists.

The definition of democracy is even more problematic. I focus in this paper on the notion of "liberal democracy." Scholars, of course, continue to debate the relationship between these two terms, but my argument proceeds analytically from both concepts. "Liberalism" refers to a conception of the state that faces juridical limits on its powers and functions. _{12 "} Democracy" refers to a form of government in which power rests with the majority. Democracy requires govern ments to be able to garner majority approval of their performance in order to stay in power. At the same time, liberalism will require that minority opinions can be expressed and that rivals for power will be able to exercise their rights to try to form alternative majorities. The demands that power be limited and that it rest with the majority can be in tension. ₁₃ In the modern world, however, liberalism and democracy have become strongly, though not perfectly, interconnected. Indeed, a number of scholars argue that modern democracy in its juridical or institutional sense is a natural extension of liberalism. ₁₄ For the purposes of this analysis, then, liberal democracies are states that are limited in their conduct of international affairs by constitutionally defined institutions of popular will and of juridical constraint. ₁₅ At the domestic level, the survival of liberal democracy and the ability of governments to make credible commitments are inherently intertwined. The existence of liberal democracy ultimately rests on the ability of the majority to convince minorities that it will not remake institutions when its narrow self-interests might be better served by abandoning the notion of limited government. A central question of liberal democratic theory, then, is how it is that the majority commits to accept limits on its power. ₁₆ Similarly, scholars have long debated the implications of limited government and majority rule for external commitments. Before moving to the analytic portion of this inquiry, it is worth a brief detour to summarize some of these existing perspectives on the ability of liberal democratic states to make commitments in their international relations.

Three Perspectives on Democratic Commitments

The traditional views on the ability of democratic states to make international commitments can be grouped into three perspectives. The first perspective emerges from the dictate of structural realism that internal organization will be irrelevant to the external behavior of states. ₁₇ In this view the ability of states to make commitments will be based on the demands of the distribution of power in the anarchic international system. There is little room, then, for different behaviors to arise systematically from variations in domestic regimes. In the words of Kenneth Waltz, "International politics consists of like units duplicating one another's activities." ₁₈ All states will have trouble making commitments because the system is anarchic, and the incentives for keeping or breaking commitments will be no different for democratic or nondemocratic regimes. To date, the vast majority of the literature on the nature of commitments in international relations has treated regime type as irrelevant.

Those who have addressed domestic dynamics and the impact of regime type have tended to take a second perspective, which views democratic states as distinctively less capable of making strong commitments. As Machiavelli asserts, there is a long tradition of skepticism regarding the efficacy of internal democracy for external relations in general, and regarding the ability of democratic states to make external commitments in particular. Democratic foreign policy, in this view, is dependent on the vagaries and passions of public opinion. Thucydides' reference to the possibility that other states might be more willing to enter into agreements with the oligarchy of the Four Hundred than with the "inconstant commons" expresses this basic concern. Tocqueville's oft-quoted observation that "in the control of society's foreign affairs democratic governments do appear decidedly inferior to others" is bolstered by his claim that a democratic government tends "to obey its feelings rather than its calculations and to abandon a long-matured plan to satisfy a momentary passion." ₁₉ The nineteenth-century British prime minister Lord Salisbury points to the regular changes of leadership demanded by democratic publics as a significant limitation on the ability of any given leader to commit the state to a course of action. "For this reason, if no other," he argues, "Britain could not make military alliances on the continental pattern." ₂₀

The third perspective sees democracies as well able to enter into long-term commitments. Some who hold this view make a positive argument about the characteristics of democracy that will enhance the strength of international commitments, while others attribute the strength of democratic commitments to an inability to change course rapidly. Machiavelli typifies the more negative view that the cumbersome machinery of democratic foreign policy making will increase democratic reliability even after objective interests have changed. Immanuel Kant exemplifies the positive view, holding that states with "republican" forms of government will be united by bonds of trade and shared norms. In Kant's regime of "asocial sociability," the demo cratic norms of nonviolent problem solving will be operative between as well as within democratic states. ₂₁ It is for this third perspective that I will argue here: distinctive institutions and preferences should enhance the ability of democratic states to make credible international commitments.

The Theoretical Bases forDemocratic Distinctiveness

I make the argument for a distinctive democratic capability to make lasting international commitments in three parts. First, I look at several arguments about the basic stability of democratic foreign policy. I then argue that there are particular and distinctive values and foreign policy preferences in democratic states that can contribute to stable international commitments. Finally, I suggest that some characteristics of the internal institutions of democratic states are critical in enhancing the credibility of external commitments.

The Stability of Foreign Policy in LiberalDemocratic StatesThe central argument of those who question the ability of democratic states to make credible commitments in the international system focuses on the putative instability of democratic policy choices. I will therefore begin at that point in setting out the case for strong democratic commitments. The stability of policy making in liberal democratic states is, of course, an enormous and significant subject. I briefly assess foreign policy stability here in terms of the stability of public preferences, the stability of democratic leadership, and the stability of foreign policy institutions. In each case I begin with a look at the traditional view of democratic instability and then turn to a positive argument for the stability of the international commitments of democratic states.

The Stability of Public Preferences Gabriel Almond sets the tone for the view of fickle democratic foreign policy making in his classic analysis of the American public and foreign policy:

An overtly interventionist and "responsible" United States hides a covertly isolationist longing, . . . an overtly tolerant America is at the same time barely stifling intolerance reactions, . . . an idealistic America is muttering soto voce cynicisms, . . . a surface optimism in America conceals a dread of the future. ₂₂

This image has been further bolstered by the public opinion work that emphasizes the weakness of political conceptions in the general public. ₂₃ If democratic publics are fickle, and if democratic foreign policies are especially sensitive to public preferences, ₂₄ then we might expect democratic foreign policies to be highly unpredictable.

While the image of changeability is a strong one, it is not one we should accept too hastily. The most significant of the recent work in this area has argued that democratic states are actually quite stable in their domestic preference orderings. ₂₅ In assessing the stability of democratic policy, it is well to remember Waltz's warning that when evaluating the abilities of democratic states in the foreign policy arena it is important to consider those abilities relative to the abilities of nondemocratic states. ₂₆ That democratic states flip and flop between isolationism and interventionism may be true, but this does not mean that other states have stable preferences simply because they are headed by a single despot. ₂₇ Machiavelli makes such a comparative argument in rejecting the view of the masses as fickle--a view that he ascribes to Titus Livy and "all other historians":

I claim, then, that for the failing for which writers blame the masses, any body of men one cares to select may be blamed, and especially princes. . . . The nature of the masses, then, is no more reprehensible than is the nature of princes, for all do wrong and to the same extent when there is nothing to prevent them doing wrong. Of this there are plenty of examples besides those given, both among the Roman emperors and among other tyrants and princes; and in them we find a degree of inconstancy and changeability in behaviour such as is never found in the masses. ₂₈

In the more contemporary setting, we can consider the frequent criticisms of the response of democratic states to the rise of Nazi Germany. ₂₉ If analysts wish to draw strong lessons from the vacillation of the democracies in the interwar years, then it is only fair to point to the dramatic shifts in Soviet-German relations in that period as well. The democratic states were uncertain about how to interpret their obligations to Czechoslovakia. They did, however, finally pursue their treaty obligations with Poland in quite certain terms. Meanwhile, the Germans and Soviets were experimenting with dramatic shifts in their positions toward one another. Ultimately, of course, the Nazi-Soviet Pact proved worthless. The democratic states, on the other hand, maintained the basic shape of their commitments to one another despite very high international and domestic costs.

Contrary to the pessimism of many analysts, foreign policy issues do seem to have played an important role in American electoral politics. ₃₀ This role has not led to the extremes of either chaos or paralysis, as the critics of democratic foreign policy have predicted. The policy views of the public in aggregate have been reasonably stable and well connected to the exigencies of external events. ₃₁ When we look at the issue of policy stability from an empirical angle, the reality seems to be that democracies can maintain stable equilibrium policies. ₃₂

Social Choice Problems in Democratic States A final point of concern about the stability of democratic foreign policy preferences needs to be addressed. A recent set of arguments posits that the limitation of government power by institutions of popular will may bring daunting problems of social choice mechanisms into the foreign policy arena. ₃₃ When more than two decision makers have a voice in a decision that involves more than one dimension, there is a high probability that there will be cycling between different social choices. The social choice mechanisms of democracy should be particularly prone to this problem. ₃₄ To the degree that cycling is a problem, democratic states will find that without robust equilibrium choices it will be difficult to sustain commitments. Today's policy could shift tomorrow if the agenda is rephrased or if coalitions shift even very slightly.

While important, the problem of cycling should not be overstated. Indeed, one of the preeminent analysts of this phenomenon, William Riker, argues that the danger of shifting majorities can actually enhance policy stability in liberal democracies. In his definition of liberalism, voting is viewed as a tool for restraining officials and limiting government rather than a tool for revealing some "true" construct of the general will. He suggests that the prospect of electoral accountability will force leaders to choose policies with sufficient appeal to avoid offending a variety of potential majorities. ₃₅

Even if we had doubts about Riker's defense of liberal democracy, we could return to the comparative perspective. Nondemocracies should also face cycling to the degree that there are multiple dimensions to a given foreign policy problem. ₃₆ As was evident in the twists and turns of policy in the disintegrating Soviet Union, factionalism can undermine the consistency of policy making in a wide variety of political regimes. The more concentrated the decision making, the less likely that this will be a problem, but even a single decision maker could succumb to incoherence and cycling when making judgments about issues that involve multiple dimensions of value that are difficult to compare. ₃₇

The Stability of Democratic Leadership A central fact of the constraints on government power in the modern liberal democracies has been limitations on the tenure of government leaders, in the form of either formal limits on the length of public office or informal limitations imposed by the application of changeable public preferences in the process of assessing and ratifying government leaders. The fact of regular leadership change is an important element in thinking about the relationship between democracy and commitment. Henry Bienen and Nicholas Van de Walle have shown that the leaders of democratic states do tend to have shorter tenures than the leaders of nondemocratic states. ₃₈ Those who would enter into commitments with democratic states must face the possibility that a new leader will be less inclined to honor previous commitments. The United States faces the prospect of major leadership change every four years. In parliamentary systems, the government could fall at any time. Some kinds of agreements surely will survive across governments, but it is plausible that the myriad small understandings that condition relations between states might be threatened by a new administration with its new team of top foreign policy makers and ambassadors. In the late nineteenth century, Otto von Bismarck had strong preferences for the Conservative Party in England, headed by Salisbury, over the Liberal Party, headed by William Gladstone. Bismarck's desire to avoid dealing with Gladstone spilled over into the Anglo-German relationship even when Salisbury was in power. C. J. Lowe describes this dynamic as important to several issues, including Anglo-German cooperation to guarantee the integrity of Persia:

Although Bismarck had considerable regard for Salisbury personally there was no guarantee that Gladstone would not return to office (as in fact happened in January 1886). Though Salisbury had thought there would be "continuity of policy in this matter" Bismarck did not, and, to his mind, dealing with Gladstone was an impossibility. ₃₉

The simple fact that leadership change is more frequent is not, however, necessarily a negative factor for commitment. Again, a com parative perspective is important. Democratic leadership changes are regularized as well as being regular. The ability of democratic states to make smooth leadership transitions can help improve the stability of commitments. Indeed, Riker argues that rapid elite circulation can itself stabilize policies. ₄₀ Nondemocratic states that do not have effective means for making leadership transitions may have fewer leadership changes, but those changes may be accompanied by greater shifts in preferences and policies. Thus, such states may present an even greater risk for other states that demand a reasonable probability of continuity for entering into a commitment. The transition from President Carter to President Reagan pales in comparison to the change from the Shah of Iran to Ayatollah Khomeini, from Mao Tse-tung to Deng Xiaoping, from Joseph Stalin to Nikita Khrushchev, or from Leonid Brezhnev to Mikhail Gorbachev.

Finally, it is important to remember that the juridical nature of liberal democracy gives current leaders the power to commit future leaders. Political power in liberal democracies rests abstractly with the office and is limited by juridical principles, rather than resting with specific individuals or being unlimited. Thus, future leaders are bound by the domestic legal environment to honor the treaty commitments of their predecessors. In this way international commitments are strengthened by the ability of the liberal democracies to make internal commitments.

The Stability of Democratic Institutions While the political life of individual leaders may be relatively short and unpredictable in liberal democracies, domestic political institutions themselves are considerably more stable. As I have argued above, liberal democracy requires that majorities be able to commit to stable institutional arrangements that codify minority rights and constrain majority powers. To the degree that democratic states possess institutional stability despite regular and regularized leadership change, it should be easier for them to enter into commitments. Stable civil service bureaucracies that handle foreign affairs, for example, help ensure some degree of policy continuity and thus make democratic states more likely to maintain commitments. The measurement of the stability of institutional structures is more difficult than the simple observation of leadership turnover or political violence. The United Kingdom and the United States have had very stable political institutions. Whether there has been stability in the foreign policy institutions or personnel of countries such as France or India is an empirical question that warrants future consideration.

The Distinctive Preferences of Liberal Democracies

A demonstration that democratic foreign policy preferences do not exhibit the extremes of changeability that some have predicted is important but is not itself a sufficient answer to the basic question about the ability of democratic states to make international commitments. In responding to the traditional critique of democratic foreign policy making, we also need to look at the kinds of values democratic states bring to bear in thinking about international commitments. It is common for analysts of the liberal democratic states to focus on their political culture. In this line of argument there is something distinctive about the ideas and values that are held by democratic publics. Whether such distinctive ideas are a result of democratization or a cause of it is an important question, but it does not need to concern us here. For my purposes it is sufficient to ask whether there are distinctive preferences that would make democratic states either more or less able to make commitments than are other kinds of states.

Tocqueville made a number of assertions about the distinctive preferences that would emerge in democratic political culture. It should come as no surprise to those familiar with his generally pessimistic assessment of the foreign policy prowess of democratic states that he viewed these preferences as largely inimical to effective foreign policy commitments and sustained international involvement in general. ₄₁ Isolationism is a characteristic frequently attributed to democratic states. To the degree that democratic states turn inward, they will pay less attention to their international obligations and may thus prove less reliable. But this logic is not definitive. There are at least two other possible connections between isolationism and international commitments. First, following Machiavelli's argument, an isolationist turn may make states take less account of the need to abandon a commitment that begins to conflict with their interests. ₄₂ Second, the isolationist state may be inclined to make only those commitments that involve truly vital national interests and thus are more likely to be honored. ₄₃

The Role of Law in Liberal Democracy Tocqueville also suggests that respect for law is a critical component of democratic political culture. ₄₄ As I suggested above, the internal practice of liberal democracy requires a basic respect for legal commitments. More recently, some have argued that these internal norms are also reflected in preferences over external policies. ₄₅ While it is arguable whether law is an overwhelming force in democratic foreign policy making, there is some evidence for a connection between international commitments and domestic legal commitments. For example, international law has long been expressly incorporated into the domestic legal order in the Anglo-American legal tradition and has spread to most of the other major liberal democracies as well. ₄₆ In relations between states, legalism and the reputation of a state for reliability do seem to have at least significant rhetorical appeal in democratic polities. Whether the respect for law emerges from practice, from ideology, or from some other fundamental inclination, if democratic peoples hold legal norms to be of some overarching legitimacy, then this will increase their sense of the binding nature of international commitments. ₄₇

Democratic Interdependence Tocqueville identifies a third source of distinctive preferences in liberal democratic states pointing to the effects of "interdependence." The combination of openness and wider political participation that characterizes democratic states may allow for more of what Robert Keohane and Joseph Nye call "multiple channels of communication" than we would expect to see in relationships between relatively more closed nondemocratic societies. Liberal economic orders that lead to increased trade and other associations between their citizens will naturally make them more interdependent. This logic follows closely Kant's argument about the "pacific union" of democratic states, based on the free flow of people and goods. ₄₈ Tocqueville suggests interdependence as a basis for the lack of war between democratic states:

As the spread of equality, taking place in several countries at once, simultaneously draws the inhabitants into trade and industry, not only do their tastes come to be alike, but their interests become so mixed and entangled that no nation can inflict on others ills which will not fall back on its own head. So that in the end all come to think of war as a calamity almost as severe for the conqueror as for the conquered. ₄₉

Tocqueville focuses on conflicts between two interdependent liberal states. But suppose that there are two interdependent liberal democracies and another state that chooses to attack one of them. We might paraphrase Tocqueville to suggest that a third-party attack on an ally might be almost as severe for the interdependent ally as it is for the attacked state. Thus, interdependence can increase the credibility of commitments between states faced with an outside threat.

More recently, an important line of research in international relations theory argues that in an interdependent world, repeated interactions increase the likelihood that states will be able to make commitments because of the iterative nature of their relationships. ₅₀ The importance of interdependence remains a contentious issue, ₅₁ but to the degree that we hold it to be important, it seems likely that democratic states will be particularly prone to its dynamics. Interdependence will generate groups within each state that have a vested interest in the commitment relationship between the states.

The Institutional Resources for Democratic Commitments

Liberal democracy makes it more likely that interdependent interest groups will be able to push the larger society to take their interests into consideration. The role of interest groups with vested interests in international commitments not only reflects on the distinctive preferences of liberal states but also points to the role of their internal institutions in strengthening commitments.

The Multiple Levels of Democratic Domestic Politics The notion of liberal democracy as a system of majoritarian and juridical limits on government action is suggestive of Robert Putnam's argument that two-level games are a useful analogue for many aspects of international politics. ₅₂ In his model, state leaders must negotiate in the international arena and then return home to sell commitments in the domestic arena. Wide participation in the political process makes this a particularly acute issue for democratic leaders. If foreign policy is dependent on public approval, and if public preferences are either distinct from leader preferences or are constantly and dramatically changing, then the state will have difficulty making the credible commitments it would otherwise choose. In this regard, Putnam makes a particularly interesting distinction between voluntary and involuntary defection from cooperative schemes. As with Woodrow Wilson and the League of Nations, or Jimmy Carter and the SALT II Treaty, democratic leaders can enter into international agreements in good faith but then find themselves unable to implement the agreement because of democratic constraints on their power at home.

This, however, is not a sufficient consideration of the role of domestic constraints. Walter Lippmann worried in Essays in the Public Philosophy that democratic states would be frozen into undesirable policies by the inability to mobilize public support for change. ₅₃ This is also the basis of Machiavelli's assertion that democratic states are less likely to break treaties, even when they have strong incentives to do so. ₅₄ By this logic, the same factors that make it difficult for democratic states to enter into commitments also make it harder to get out of them. For example, the stronger role of domestic actors makes it more difficult to break or renegotiate commitments once the difficult barrier of ratification has been passed.

Domestic politics will be particularly effective at increasing the ability of democratic leaders to make commitments that accord with the interests of a strong domestic constituency. In this regard, the "liberal" component of liberal democracy is particularly important. The assurance that minority opinions will not be suppressed creates a plethora of interest groups that may have a concentrated interest in a particular international commitment. The United States can make effective commitments to Israel even without a formal alliance because it has a substantial domestic audience that will monitor and enforce that commitment in the domestic arena. Leaders with differing perspectives on the Middle East may come and go (though there are some obvious constraints on that process as well), but Israel and the Arab states can reasonably predict that there will be strong incentives for a pro-Israeli policy in the American domestic political process. The United States is a particularly dramatic set of these kinds of incentives because of its diverse ethnic makeup. But similar dynamics can certainly be at work in other states as well. One example here would be Germany's somewhat reticent acquiescence to the 1994 round of the Basel Convention banning all exports of hazardous wastes. Germany's compliance with the agreement will be closely monitored, not only by the other parties but also by Germany's own environmental activists. Thus the combination of interdependence and a strong voice for domestic actors has the potential to substan tially increase the ability of democratic states to make commitments when the interests of other states are shared by significant domestic groups.

The Transparency of Democratic Domestic Politics The multiple levels of democratic policy making take on particular significance because democratic political systems are relatively transparent. Without the ability to observe what the government is doing and the freedom to express and organize alternative political views, the liberal notions of limited government and political competition would be meaningless. It is very difficult, however, to discriminate against external actors in providing transparency to internal actors. Outsiders will be able to observe with relative ease what is going on inside democratic states. Any embassy can subscribe to the major newspapers that provide day-to-day investigative services on the policy-making activities of the democratic state. Of course, foreign analysts may have just as much trouble making sense of the wealth of information that is available as do the legions of confused analysts within democratic states. Nevertheless, the difference between this and the process of discerning the policy process in closed societies with small and tight-knit leadership groups is vast. This transparency has an important implication for the ability of democratic states to make international commitments, because outsiders can observe linkages between commitments made to them and commitments made to the domestic audience. When a democratic leader makes a public commitment to a specific course of action, deviation from that course might bring domestic as well as international repercussions. When President Bush vowed to remove Iraqi troops from Kuwait, the Iraqis should have known that that vow would bear on the ensuing election as well as on the internationalsituation.

Recent work at the interstices of economics and political science has shed new light on the relationship between social organization and the ability of states to make commitments to domestic audiences. Two particularly interesting examples of this literature are Douglas North and Barry Weingast's interpretation of the Glorious Revolution as an exercise in recasting a constitution in order to increase the ability of the state to make commitments, and François Velde and Thomas Sargent's similar interpretation of the French Revolution. ₅₅ In these pieces the respective authors argue that democratic institu tions can increase the ability of the state to make commitments to large numbers of domestic actors. This literature does not reflect directly on the ability of the state to make commitments in the international arena, where the number of actors is small and repeated interaction already is likely. My argument here is that in the international arena, the ability to link external commitments transparently with internal commitments will allow democratic states to draw on domestic audiences to aid their international credibility.

Thomas Schelling points to the importance of political costs for enhancing the credibility of international commitments. ₅₆ He focuses on incurring political costs within the international system itself. But similar benefits can be derived from incurring these costs at home if they can be adequately observed from outside. The linkage between external commitments and internal political costs is represented formally in James Fearon's work on the role of audience costs in international interactions. ₅₇ When democratic leaders send signals in the international arena that bear domestic costs at home, those signals will have more credibility than would similar signals that bore no significant domestic costs. All states face some domestic costs for their international actions, but democratic states may be distinctive in the degree of domestic accountability. The critical aspects of democracies in this regard are (1) that there are a number of institutions that can be easily used for linking internal and external commitments--such as elections--and (2) that internal commitments and domestic costs can be more easily observed from the outside because of the nature of information institutions within democratic states.

Domestic costs can enhance commitments in two distinct ways. First, taking actions that bear immediate domestic costs can send a signal about the genuine importance of an issue to an actor. ₅₈ Second, some actions can create domestic costs for a failure to maintain a commitment in the future. A trade agreement, for example, may lead to the development of a domestic constituency with a vested financial interest in maintaining the trade relationship. To renege on such an agreement would incur domestic costs proportional to the strength of that constituency. More generally, statements and actions may create domestic expectations that will lead to later audience costs or electoral punishment if a leader fails to carry out an international commitment.

Making credible international commitments is difficult at best for all states. I have argued here that, contrary to the traditional image of unreliability, democratic states should be relatively effective at making international commitments. The task now is to turn to some empirical attempts to assess the overall ability of democratic states to make commitments and to abide by them.

Empirical Soundings: Democratic Alliance Behavior

Alliances are the most salient form of commitment behavior in the current international system. States join formal alliances in order to indicate both to their alliance partners and to other states that there is a level of commitment that is greater than the level of commitment that would be expected based simply on observed international interests. My research focus here is on the duration of alliance commitments. If democratic states are unreliable because of shifting majority preferences, we would expect to see this reflected in the length of time that they are able to maintain alliances.

Alliance duration is not a perfect indicator of the ability of democratic states to make effective international commitments. Conceptually, there are many reasons why alliance relationships might endure or be cut short. But in the aggregate, if those other reasons are independent of the democratic status of a state, we should still be able to pick up some meaningful variation in looking at alliance duration. Similarly, we might prefer to measure more directly the effectiveness of alliance commitments under stress. When states are attacked, do their allies back them up? Do allies help out in wars? We know from other studies that allies rarely join in when wars break out. ₅₉ But this indicator is almost surely biased by the deterrence effect of allies who are expected to be reliable. Unreliable alliances are more likely to be challenged than reliable alliances. Thus, while future work could well benefit from attempts to carefully measure, code, and analyze these more direct concepts, for this analysis I have started with the more straightforward consideration of alliance durability as an appropriate first cut at this issue.

The analysis of alliance commitments is also appropriate to the degree that they are an indicator of international community. Drawing on Kant's essay "On Perpetual Peace," Michael Doyle's explanation for the liberal peace turns on a natural community of liberal states:

Since morally autonomous citizens hold rights to liberty, the states that democratically represent them have the right to exercise political independence. Mutual respect for these rights then becomes the touchstone of international liberal theory. When states respect each other's rights, individuals are free to establish private international ties without state interference. Profitable exchanges between merchants and educational exchanges among scholars then create a web of mutual advantages and commitments that bolsters sentiments of public respect. These conventions of mutual respect have formed a cooperative foundation for relations among liberal democracies of a remarkably effective kind. ₆₀

Ultimately, it would also be useful to examine other indicators of "the web of commitments" between liberal states. Alliance behavior is a good starting point, since it is a manifestation of the commitments between states for which some data are readily available. ₆₁ Alliances have been used as indicators of a number of important international phenomena, including the basic interests of states, their attitudes toward risk, and the effectiveness of international norms. ₆₂ Military alliances are also a good starting point for this line of research in that this is an area in which we would expect to find the least distinctiveness to democratic behavior. As outlined in the liberal logic, it is in the economic sphere that we would more expect to see the effects of interdependence. ₆₃ It would be a strong result, then, to find evidence of a distinctive character to democratic commitments in this area.

There has been some empirical work on the question of democratic alliance behavior. Ole Holsti, Terrence Hopmann, and John Sullivan included a polity variable in their 1973 analysis of alliance politics. ₆₄ Their conclusions about democratic alliance behavior are mixed. In their survey of all alliances between 1815 and 1939, they find that ideological similarity disposes states to ally with each other and to effect some increase in the length of alliances, ₆₅ though their bottom line is that after alliances are formed, the impact of ideological differences is minimal. ₆₆ They also find some areas of democratic distinctiveness in their case study work. For example, looking at the differences between Chinese and French defection from their respective alliance systems in the 1950s, they argue that in pluralistic polities intra-alliance disputes tend to stay confined to a narrow range of issues, while in nonpluralistic polities intra-alliance disputes tend to spill over into all issue areas. ₆₇ In an argument that echoes the Kantian hypothesis, they posit that the source for this effect is the complex interdependence that creates a large number of nongovernmental ties between pluralistic states.

Randolph Siverson and Juliann Emmons, in an analysis that focuses specifically on democratic states, confirm with more rigorous statistics the observation of Holsti, Hopmann, and Sullivan that ideologically similar states are more likely to form high-commitment alliances. ₆₈ They show that at the dyadic level there is a strong tendency for democratic states to form alliances with each other at a greater rate than would be expected from the null model assumption that alliance formation should be independent of ideological orientation.

My goal here is to expand on these existing results with an attempt to assess the relative durability of democratic and nondemocratic alliances. The statistical analysis of Holsti, Hopmann, and Sullivan is largely limited to contingency table analysis. I focus on the case of democratic states to confirm the rather tentative relationship they describe for the relationship between alliance duration and ideological affinity. By using more sophisticated techniques for analyzing duration data, I am able to provide a more nuanced assessment of the effect of shared democratic norms on alliance duration.

The Data: Measuring Democracy and Measuring Alliances

Two kinds of data are required for this analysis: data about polities and data about alliances. Both of these present difficulties and require certain judgment calls. The danger in this kind of endeavor is that the specific research objectives can unduly influence the making of these judgments. For this reason, and not incidentally as a matter of efficiency, I have relied on the data collection efforts of others. I have used Michael Doyle's coding of liberal regimes ₆₉ and the coding of alliances from the Correlates of War Project. ₇₀ In so doing, of course, I have inherited the controversies that surround both of these efforts.

For my purposes, the democracy measure is reasonably straightforward. It is not necessary to resolve the significant debates about the meaning of these terms in political philosophy and comparative politics in order to advance propositions about the implications of liberal democracy for foreign policy and international relations. Even the problematic distinction between "liberal" and "democratic" retreats in importance in the face of the empirical reality that the two phenomena have been highly coincident in modern history. There is a relatively clear set of states that have been regularly labeled "democratic" or "liberal." While one might disagree about some cases on the edges, the results I report here are relatively insensitive to small definitional changes. Still, it should be acknowledged that by adopting a binary measure of democracy that aggregates across a number of important features of democracy, we lose the ability to assess the relative importance of different aspects of democratic governance. I will not be able to evaluate here the degree to which differences in the ability to enter into durable alliances result from regular elections, constitutionalism, or the distinctive preferences of democratic citizens. Even more serious is the possibility that the causality may flow from other factors that this class of states largely have in common. This becomes a problem to the degree that these factors, though historically correlated with democracy, are not an essential part of it. Of course, all analytic categories are constructs of lesser and correlated concepts. Democracy, too, could be broken down into its constituent parts and correlates. But such a move would come at the expense of the ability to do any kind of aggregate analysis. In the end, then, the analysis that follows is a description of the behavior of a specific group of states, and it remains to be shown that it is democracy, and not something else, that is the determinative characteristic of this group.

The conceptual problems surrounding the measurement of alliances are more immediately serious. In thinking about these problems it is important to start by emphasizing that the things that we can easily observe about alliances are not necessarily the phenomena in which we are most interested. ₇₁ Theoretically, we are most interested in the quality of international interactions. Is some set of countries more or less amicable in its relations? Would its component states defend one another if attacked? ₇₂ Will the states refrain from action that might be injurious to an alliance partner? Consider, for example, the fact that there is no formal defense pact between the United States and Israel. Israel, the United States, and most other interested states know that there is a substantial defense commitment between the United States and Israel (though the question of its exact extent may be a matter of considerable consequence and concern for Israel and its enemies). In using alliances as an indicator of defense relationships we will be limited to the degree either that good relationships are not formalized or that bad relationships are. The case of Israel and the United States is fairly clear-cut, but states will often have difficulty assessing the quality of relationships. Indeed, alliances can be seen as a mechanism for trying to deal with these uncertainties in the international environment. Nonetheless, there are some particular kinds of uncertainties that are probably more severe for the analyst trying to consider a large number of alliances than for the state trying to assess the quality of a given relationship at a given point in time.

One particularly vexing conceptual issue is whether alliance behavior should be analyzed with the alliance as the unit of measurement or with the dyad as the unit. The Correlates of War data are organized by dyads. Holsti, Hopmann, and Sullivan, as well as Bruce Russett, have worked with alliances as the unit of observation. ₇₃ Siverson and Emmons base their study on dyadic relationships. ₇₄ There are conceptual arguments in both directions. A focus on the formal treaties would lead us to concentrate on the alliance as the observation: how long treaties are in force would be the most relevant question. If, however, we are interested conceptually in the underlying relations between individual countries, we will need to turn to the analysis of dyads. A focus on the alliance as the unit of observation also runs into problems when multiple treaties reflect the same relationship. For example, while a single treaty unites the NATO countries, the Warsaw Pact countries cemented their relationship with a large number of bilateral treaties. The use of treaties as the unit of observation would bias the data toward this kind of multilateral relationship. The use of dyads as the unit of observation would give extra weight to multilateral treaties. Both biases present serious problems. In both cases multilateral alliances lead to problems in assessing the relationship between individual states when formal relationships end because of a falling-out between other alliance members. Future work might benefit from trying to assess more subjectively the true extent of multilateral relationships. In this analysis, my approach is to run the statistical tests with both kinds of data. The fact that the findings are reasonably robust with both kinds of data increases our confidence in the results.

Alliance dyads are coded from the Singer-Small set, and then a democracy score is added from Doyle's listing. Again, the central drawback to this coding scheme is that some multilateral alliance relationships get very high weight because of the large number of individual countries involved. The worst offender in this regard is the Organization of African Unity, which is coded as an entente and counts for 362 of the 1,321 dyad observations.

Translating the Singer-Small data to the alliance level, rather than to the dyadic level, is more complex than it might appear at first blush. ₇₅ The decisions I have made in this regard are not always transparent, and thus they bear some discussion. Should we count the West European Union as a different treaty than NATO? Is the Rio Pact with Cuba a different alliance than the Rio Pact without Cuba? I have used two different kinds of decision rules, and the results seem reasonably insensitive to these coding variations. First, I tried to identify the individual treaties and gave them their longest life, regardless of new members coming and going (Reduced Model I). Second, I identified starting and ending dates in the dyadic data set and collapsed the data around these values (Reduced Model II). The first method tends to overcount multilateral alliances that use bilateral treaties, such as the Warsaw Pact. The second method overcounts multilateral alliances that have more changes over time, such as NATO or the Arab League.

Multilateral treaties are also problematic for coding the democracy variable when they include states with different political systems. My focus is on relationships between democratic states, so I have chosen in both of these reduced data sets to decompose treaties that have mixed democratic and nondemocratic members. Thus, for example, I code NATO as three observations: a relationship between democracies, a relationship between democracies and nondemocracies, and a relationship between nondemocracies. Interestingly, this affected only six alliances, including three nineteenth-century alliances involving Britain, France, or Italy in their democratic periods, NATO, the Rio Pact, and the Arab League (which included Lebanon when it was coded as liberal).

International Alliance Behavior and Democratic States

Figure 1.1 tracks the average number of alliance relationships for democratic and nondemocratic states for each decade between 1816 and 1965. Several salient points are noteworthy in this plot. Before 1870 there are very few democratic states, and those that there are have decidedly fewer alliance relationships of any kind than do the nondemocratic states. After 1870, the curves for the democratic and nondemocratic states track each other very closely. From 1870 until 1920 alliance relationships are at a fairly low level for both democratic and nondemocratic states. Finally, in 1920 a strong trend began toward an increasing number of alliance relationships. One small but noteworthy point that I do not show on this chart is that from 1930 on, the democratic states have tended to be in more defense pact relationships--the highest form of defense commitment in the Correlates of War coding--than the nondemocratic states have. ₇₆ Figure 1.1 is, of course, a simple presentation of the relationship between alliances and democracy with no controls for confounding factors. On its face, this pattern would give the most support to the expectation that domestic regime type should not make much difference in international behavior in general and in the ability to make commitments in particular. ₇₇ These results do not support the idea that democratic states should be more alliance-prone, but neither do they support the more often expressed concern that democratic states cannot make credible commitments. Democratic states find just as many alliance partners as nondemocratic states do. More recently they have been able to find more partners willing to enter into high-level defense pacts. Either Salisbury was wrong or something has changed since he suggested that democratic states cannot keep their promises and thus will have trouble entering into alliances. At a minimum, democratic states are finding other states that are at least willing to sign the papers.

The broad review of international alliance behavior represented in figure 1.1 does not suggest significant differences between democratic and nondemocratic states in terms of their propensity to form alliance relationships. ₇₈ But the question in which we are most interested is not simply how many alliance relationships democratic states enter but what level of commitment those relationships represent. We can move one analytic step closer to this more fundamental issue by considering the length of time that democratic and nondemocratic alliances tend to last.

The Duration of Alliances

Statistical analysis of duration data is made treacherous by several factors. Briefly, the two primary problems are nonlinear relationships and the censoring of data. ₇₉ Duration data are said to be right censored when some of the events are still ongoing at the end of the observation period. For example, a seemingly robust alliance that starts just two years before the end of the observation period should not be coded as having ended after just two years. If we did not take censoring into account, we would bias our analysis for all the cases of alliances that were still in effect at the end of the period of observation. This bias is nontrivial because it would tend to be the longest-lasting alliances that would be censored. This is of particular importance in the study of alliances because a large number of alliances are still ongoing.

The most common method for examining survival data, given these problems, is the use of Kaplan-Meier or product-limit estimates of the survival function. This is a nonparametric technique for estimating, in our case, the probability that an alliance will survive a given number of years. The Kaplan-Meier estimate of the probability that an alliance will last k years is the product of the estimate of the probability that the alliance will last k-1 years and the observed survival rate in year k. Thus, censored observations will provide information as to the number of alliances that last k-1 years, while the uncensored observations will provide the observed survival rate in any given year. ₈₀ Figure 1.2 displays the Kaplan-Meier estimates of survival times for the first reduced data set based on treaties. The three lines show the estimated survival function for democratic alliances, nondemocratic alliances, and mixed alliances. The distinctiveness of democratic alliances is clearly visible in this figure. Reading across the chart at the 50 percent survival mark, we can see that the median survival time for both mixed and nondemocratic alliances is about seven years, while democratic alliances have a median survival time of about seventeen years. ₈₁

The central limitation of the Kaplan-Meier estimates is that they are nonparametric. While they provide an effective visual indicator of survival patterns, it is difficult to control for important covariates and to specify more exactly the independent effect of democracy on alliance duration. A next step, then, is to turn to a parametric survival model. The model I use here to assess the effects of democracy on alliance duration is an accelerated failure time model. ₈₂ In this procedure the following model is estimated:

Where T ₀is a time value from a baseline Weibull distribution with s as a scale term. ₈₃ X is a vector of covariates and B is a vector of coefficients, so that exp(x'B) is a proportional scale on the baseline distribution. The nature of the model is more easily seen with the natural log transformation of each side:

In this transformation it can be seen that a specific coefficient Bk will represent the change in the natural log of the duration for a given change in a covariate x_k . There is also a more straightforward interpretation of the coefficients in this model since the coefficient Bk will approximate the percentage change in duration for a unit change in x_k .

We deal with censoring in these estimations by calculating the maximum likelihood estimators of the parameters with the probability density function for the uncensored durations and the survivor function for the censored durations. In other words, the estimating procedure maximizes the likelihood for the parameter estimates using the known survival times of the uncensored durations and the predicted survival times based on the predicated underlying distribution for the censored observations. ₈₄

The Covariates The dependent variable in this model is the duration of the alliance. The independent variables are the year of entry, the type of alliance, the major power status of the states, the number of states in the alliance, and whether the alliance members are democratic. Discrete variables are transformed into dummy variables for each of their values. In order for the model to be full rank, coefficients can only be provided for n-1 of n discrete levels. The specification and definitions of the independent variables are shown in table 1.1.

There are no variables DEM0, MAJOR0, or TYPE3 because of the rank requirement. If both DEM2 and DEM1 are equal to zero, then the equation is being estimated for the case of no democracies. When MAJOR2 and MAJOR1 are both zero, then the model is estimated for the case of no major powers. The expected duration for an entente will be estimated when both TYPE1 and TYPE2 are zero.

Tables 1.2, 1.3, and 1.4 show the results of fitting the model to six different subsets of the data. Table 1.2 examines dyads. Table 1.3 examines the data set with the dyadic observations collapsed around treaties as the unit of observation. Table 1.4 collapses the data set around alliance starting and ending dates, as described above. For each of the tables, the analysis is run with all three kinds of alliance relationships and with just defense pacts. Seven extreme cases, in which the alliance lasted longer than thirty years, have been removed from the data set. ₈₅

As should be apparent from each table, this model does show a significant effect for the duration of alliances between liberal democracies. These effects are consistent in direction across all of the aggregations of the data and are statistically significant for the dual democracy coefficient in all of the models that use all alliances and for one of the dual democracy coefficients in the defense pact models. The magnitude of the democracy effect varies from 0.47 to 1.3, though all but one of the coefficients fall between 0.47 and 0.74. The impact of these coefficients can be seen more concretely in the examples given in table 1.5, in which I have presented the expected durations of a defense pact between two major powers that starts in 1925. ₈₆ With all of the other independent variables held constant, the effect of the democracy variables is clearly visible in each row of the table. Most of the models predict fairly similar results.

As in the Siverson and Emmons work and in the work on democracies and wars, it is again the dyadic effects of democracy that are the most notable. We can make a distinction between the case of two democracies and that of either one or no democracies. But there is no statistically significant separation between the cases of one democracy and no democracies. Democracies are no different from nondemocracies when it comes to relationships with nondemocracies. It is only alliances between democracies that appear to be more durable. If alliance duration is an indicator of the ability to make commitments, then democracy by itself does not appear either to increase or to decrease the ability of a state to make commitments to nondemocracies.

That democracies would be no worse at making commitments than nondemocracies are is itself interesting in light of the frequent concerns about the instability of democratic decision making. The dyadic finding, however, suggests that the most important explanations do not lie within the stability of democratic institutions themselves. Rather, the most promising source of explanation for these findings is likely to be either in the distinctive preferences that democratic states may hold for maintaining their relationships with each other or in the institutional elements that develop in the relationships between democratic states.

Technical Limitations of the Analysis As so often, there are a number of limitations in the data that should temper our confidence in the results I have reported here. There is some multicollinearity between the covariates. Democratic dyads tend more often to involve major powers and are more often related by defense pacts. More important, until very recently, the number of democracies in the world has been small. Though there are 1,322 dyads, only 121 of them are coded as involving two democracies. In the reduced set, of the 185 alliances, there are only 17 that are coded as democratic. Equally problematic for the analysis is the fact that many of these democratic alliances are of more recent vintage. Most of them are still in effect and are thus censored (only 15 of the 121 democratic dyads and 8 of the 17 democratic alliances are uncensored). Of course, it is not theoretically trivial in a study of alliance durability that most democratic relationships are still in effect. The coefficients for the democracy variables are reasonably consistent between a pre-1945 and a post-1945 partition of the data, though they tend to get both smaller and less statistically significant. ₈₇

The other major factor that tempers these results is the problem of unobserved heterogeneity. There are, no doubt, all manner of difficult-to-measure factors that may be driving these results. We may be comfortable with leaving some of them under the umbrella label of "liberal democratic states." If, for example, alliance durability is a function of intense trading relationships, and if we believe that democracy and trading relationships tend to go together, then it is interesting, but not problematic, for the model to leave out trade. Trade, in this situation, would just be one of the component characteristics of democratic states that drive the relationships between them. If, however, we believe that levels of trade are independent of democratic status, then this is a more serious omission.

Conclusions

The central characteristic of liberal democracies is juridically limited majority rule. For foreign policy decision making, this has meant that decision makers are limited in their ability to commit the state both because of the limits in their power at any given time--for example, the requirement that the president of the United States submit treaties to the Senate for ratification--and because of the possibility that public preferences will change. Drawing on these characteristics, the most traditional argument about the relationship between democratic states and commitment in the international system focuses on the "inconstant commons" and the expectation that democratic governance will be particularly ill-suited to long-lasting commitments. The relationship between polity type and the ability to make commitments is more complex than this traditional argument would allow. As Riker has argued, there is a theoretical basis for policy stability in liberal democratic regimes, a consistent between a pre-1945 and a post-1945 partition of the dand this position has been supported in several studies of foreign policy stability. Moreover, at the theoretical level the creation of links between external commitments and internal commitments and the development of shared preferences through interdependence should also enhance the ability of liberal democracies to forge effective international commitments.

Ultimately, these factors will have to be disentangled and their individual importance assessed empirically to discern the net effect of the factors that push for and against democratic commitments. I have offered here a start on that empirical task with a broad analysis of the duration of democratic alliances. Consistent with the conjectures of Doyle and Kant, there are distinctive elements in the alliance behavior of democratic states. As Siverson and Emmons have shown, democracies tend to ally with other democracies. I have shown here that there is also a tendency for these alliances to last longer than either the relationships between nondemocracies or the relationships that mix democracies and nondemocracies. Democratic alliances do appear to be distinctively durable when measured against the background of the constantly shifting international environment. Considerably more work will be required before we will want to endorse a robust version of the "pacific union" of democratic states. We can be more emphatic in the assertion that contrary to the pessimistic views of the likes of Tocqueville or Salisbury, democratic states have not demonstrated an inability to make lasting commitments.

I have benefited greatly in this project from the comments of the other participants in the Social Science Research Council Liberalization and Foreign Policy project and from the comments of three anonymous reviewers. John Ferejohn, Jeffry Frieden, Joanne Gowa, Miles Kahler, Lisa Martin, and Barry Weingast have been particularly helpful. I am indebted to Kenneth Schultz and Marissa Myers for their able research assistance. Funding was generously provided by the Center for International Security and Arms Control and by the Institute for International Studies, both at Stanford University. Doug Rivers was of considerable help in thinking about the statistical dimensions of this paper. For much of the data used in this project I am grateful to the Correlates of War Project and the Inter-university Consortium for Political and Social Research (ICPSR).

This chapter was published previously as "Democratic States and Commitment in International Relations," International Organization 50 (1) (Winter 1996):109-139. ©1996 by the IO Foundation and the Massachusetts Institute of Technology.

Note 1: Thucydides, The Peloponnesian War, book VIII, ch. 25, para. 70, p. 491. Back.

Note 2: Machiavelli, The Discourses, book I, discourse 59, p. 259. Back.

Note 3: Ibid., book I, discourse 58, p. 252. Back.

Note 4: See Huntington, The Third Wave. Back.

Note 5: Keohane, After Hegemony. Back.

Note 6: Schelling, The Strategy of Conflict; Arms and Influence. Back.

Note 7: Grieco, "Anarchy and the Limits of Cooperation." Back.

Note 8: Babst, "A Force for Peace"; Small and Singer, "The War Proneness of Demo-cratic Regimes"; Doyle, "Kant, Liberal Legacies, and Foreign Affairs," parts 1 and 2; Chan, "Mirror, Mirror on the Wall . . ."; Weede, "Democracy and War Involvement"; Doyle, "Liberalism and World Politics"; Maoz and Abdolali, "Regime Types and International Conflict"; Russett, Grasping the Democratic Peace. Back.

Note 9: The way in which international commitments bind states to future actions is a matter of considerable interest but is beyond the scope of this analysis. Snyder and Diesing (Conflict Among Nations) argue that explicit commitments engage important values such as honor, resolve, and reputation. Fearon ("Domestic Political Audiences and the Escalation of International Disputes") makes a similar argument pointing to the role of "audience costs" in the calculus of state actions. Back.

Note 10: See Schelling, Arms and Influence, ch. 3. Back.

Note 11: This theme underlies the discussion of Kegley and Raymond (When Trust Breaks Down) in which they examine the evolution of alliance norms over time. Back.

Note 12: Manning, Liberalism, p. 15; Bobbio, Liberalism and Democracy, p. 1. One might also stress liberalism's focus on the standing of individuals, but for the purposes of this analysis the notion of legal limits on government power is sufficient. Back.

Note 13: Bobbio, Liberalism and Democracy, p. 2. Back.

Note 14: Ibid., p. 31. The strong connection between liberalism and democracy also can be seen clearly in Rawls's Political Liberalism. On the other hand, the recent trend toward democratization has brought about some weakening of the connection between liberalism and democracy. See Anderson, "Democratization and Foreign Policy in the Arab World," ch. 4 in this volume. Karl and Schmitter capture a part of this phenomenon in their distinction between "consolidated" and "unconsolidated" democracy; see "Democratization Around the Globe," pp. 54-55. Back.

Note 15: This definition accords well with the notion of democracy that has been used implicitly by a number of authors who have recently written in this area. See, for example, Bueno de Mesquita and Lalman, War and Reason. Back.

Note 16: Mill and Tocqueville are exemplars of the importance of this tension in the rise of nineteenth-century liberalism. For some recent treatments of this vexing issue, see Hochschild, What's Fair, and Riker, Liberalism Against Populism. Back.

Note 17: On some of the limitations of the realist approach in this area, see Barnett and Levy, "Domestic Sources of Alliances." Back.

Note 18: Waltz, Theory of International Politics, p. 97. Interestingly, in the 1960s Waltz wrote a book on the impact of democratic politics on foreign policy (Foreign Policy and Democratic Politics). An important conclusion in that book, however, is that democracies were probably no worse than other states in their foreign policy endeavors. This conclusion stands against the prevailing wisdom of that time that democracy could be a serious barrier to effective and timely foreign policy decision making. Back.

Note 19: Tocqueville, Democracy in America, vol. I, part II, ch. 5, sec. 13, pp. 228-229. Back.

Note 20: Lowe, The Reluctant Imperialists, p. 10. Back.

Note 21: Kant, "Perpetual Peace." A more recent proponent of this position is Dixon, "Democracy and the Management of International Conflict." See also Maoz and Russett, "Normative and Structural Causes of Democratic Peace." Back.

Note 22: Almond, The American People and Foreign Policy, p. 67. Back.

Note 23: Converse, "The Nature of Belief Systems in Mass Publics." Back.

Note 24: See Monroe, "Consistency Between Public Preferences and National Policy Decisions," and Page and Shapiro, "Effects of Public Opinion on Policy." Back.

Note 25: Shapiro and Page, "Foreign Policy and the Rational Public"; Russett, Controlling the Sword, pp. 92-95. Back.

Note 26: Waltz, Foreign Policy and Democratic Politics, p. 17. Back.

Note 27: Hartz, The Liberal Tradition in America; Klingberg, "The Historical Alternation of Moods in American Foreign Policy." Back.

Note 28: Machiavelli, The Discourses, book I, discourse 58, p. 254. Back.

Note 29: See, for example, Taylor, The Origins of the Second World War, p. xi. Back.

Note 30: Aldrich, Sullivan, and Borgida, "Foreign Affairs and Issue Voting." Back.

Note 31: Page and Shapiro, The Rational Public; Holsti, "Public Opinion and Foreign Policy"; Nincic, Democracy and Foreign Policy. Back.

Note 32: Russett, Controlling the Sword; Page and Shapiro, The Rational Public. Back.

Note 33: For a useful introduction to this issue, see McLean, Public Choice. For an application to the role of public opinion in foreign policy making, see Gaubatz, "Intervention and Intransitivity." Back.

Note 34: Riker, Liberalism Against Populism; Jervis, "Rational Deterrence," p. 205. Back.

Note 35: Riker, Liberalism Against Populism. Back.

Note 36: Waltz, Foreign Policy and Democratic Politics, p. 309. Back.

Note 37: Jervis, "Rational Deterrence," p. 206. Back.

Note 38: Bienen and Van de Walle, Of Time and Power. Back.

Note 39: Lowe, The Reluctant Imperialists, p. 90. Back.

Note 40: Riker, Liberalism Against Populism. Back.

Note 41: Tocqueville, Democracy in America, vol. I, part II, ch. 5, sec. 13, p. 223. Back.

Note 42: Machiavelli, The Discourses, book I, discourse 59, p. 259. Back.

Note 43: I am indebted to an anonymous reviewer for this second point. Back.

Note 44: Tocqueville, Democracy in America, vol. I, part II, ch. 6, sec. 4, p. 240. Back.

Note 45: Doyle, "Kant, Liberal Legacies, and Foreign Affairs," p. 230. See also Dixon, "Democracy and the Peaceful Settlement of International Conflict," and Maoz and Russett, "Normative and Structural Causes of Democratic Peace." Back.

Note 46: von Glahn, Law Among Nations, ch. 3. Back.

Note 47: Burley ("International Law and International Relations Theory") argues that liberal states operate within a "zone of law" in which transnational legal arrangements strengthen ties between the liberal democratic states. Dixon ("Democracy and the Management of International Conflict") argues that domestic dispute resolution procedures within democracies carry over to their international relations. Back.

Note 48: Kant, "Idea for a Universal History," p. 50. For a recent review of the notion of a cosmopolitan international economic order, see Neff, Friends But No Allies. Back.

Note 49: Tocqueville, Democracy in America, vol. II, part III, ch. 26, p. 660. Back.

Note 50: Keohane and Nye, Power and Interdependence; Axelrod, The Evolution of Coopera-tion. These two works suggest two mechanisms for this dynamic. In Axelrod, the potential for repeated interactions increases the shadow of the future that makes cooperative behavior more likely. For Keohane and Nye, the growth of intertwined interests increases the mutual vulnerability and sensitivity of state policies. Back.

Note 51: Grieco, "Anarchy and the Limits of Cooperation." Back.

Note 52: Putnam, "Diplomacy and Domestic Politics." Back.

Note 53: Lippmann, Essays in the Public Philosophy, pp. 18-19. Back.

Note 54: Machiavelli, The Discourses, book I, discourse 59, p. 259. Back.

Note 55: North and Weingast, "Constitutions and Commitment"; Velde and Sargent, "The Macro-economic Causes and Consequences of the French Revolution." Back.

Note 56: Schelling, Arms and Influence, p. 49. Back.

Note 57: Fearon, "Domestic Political Audiences and the Escalation of International Disputes." Back.

Note 58: The foundational work in this area is Spence, Market Signaling. Back.

Note 59: Sabrosky estimates that states joined a war on the side of an ally in only 27 percent of the cases in which there was an opportunity to do so. If, however, we include the act of abstaining or remaining neutral, which is all that many alliances call for, then alliances were "honored" in 88 percent of the cases (Sabrosky, "Interstate Alliances," p. 177). Back.

Note 60: Doyle, "Kant, Liberal Legacies, and Foreign Affairs," p. 213. Back.

Note 61: The use of alliances as an indicator of community is, of course, a nontrivial assumption. Bueno de Mesquita (The War Trap) has argued, for example, that allies are more likely to fight than non-allies. But in Bueno de Mesquita's formulation this is because of the closeness rather than the conflict of their interests. Furthermore, this dynamic has not proved to be operative for democratic alliance dyads. See Siverson and Emmons, "Birds of a Feather." Back.

Note 62: See, for example, Bueno de Mesquita, The War Trap; Lalman, "Conflict Resolution and Peace"; Kegley and Raymond, When Trust Breaks Down. Back.

Note 63: Keohane and Nye, Power and Interdependence. Back.

Note 64: Holsti, Hopmann, and Sullivan, Unity and Disintegration in International Alliances. Back.

Note 65: Ibid., pp. 62-68. Back.

Note 66: Ibid., pp. 61--64. Back.

Note 67: Ibid., pp. 160-161. Back.

Note 68: By "high-commitment alliance" Siverson and Emmons mean defense pacts in the Correlates of War coding, as opposed to the lower commitment level of ententes and neutrality pacts; see Siverson and Emmons, "Birds of a Feather." Back.

Note 69: Doyle, "Kant, Liberal Legacies, and Foreign Affairs." Back.

Note 70: Singer and Small, Annual Alliance Membership Data. Back.

Note 71: The formal definition of alliances used by Singer and Small, while more tractable for empirical work with a large number of observations, is particularly limited in this way. See Walt, The Origin of Alliances. Back.

Note 72: Sabrosky ("Interstate Alliances") notes that in roughly two-thirds of the cases states fail to act when there is an opportunity to join an alliance partner in a military conflict. My initial review of these data comparing democracies and nondemocracies suggests that the two kinds of states are roughly similar on this dimension. These data are, however, very sensitive to case selection. If we include the Korean War (in which a number of democracies made at least token contributions to the U.S. effort), we find democracies to be much more likely to participate in the disputes of their democratic allies. If we leave Korea out, the two classes of regimes are roughly similar, with a slight advantage to the democracies. One of Sabrosky's measures involves the propensity of allies to attack each other. Here, following the literature on the lack of wars between democratic states, there is a significant advantage for the democracies. Back.

Note 73: Holsti, Hopmann, and Sullivan, Unity and Disintegration in International Alliances; Russett, "An Empirical Typology of International Military Alliances." Back.

Note 74: Siverson and Emmons, "Birds of a Feather." Back.

Note 75: Two errors in the distributed Correlates of War alliance data have been corrected. Yugoslavia, Greece, and Turkey entered into an alliance in 1953, rather than Yugoslavia, Greece, and Libya as the Correlates of War codes it. China and Russia entered into an alliance with North (rather than South) Korea in 1961. Back.

Note 76: There were an average of eight defense pact relationships for the democratic states as against five for the nondemocratic states. Using a t-test with Satterthwaite's adjustment for unequal variances, this difference is significant at the .01 level. Back.

Note 77: This picture of systemwide trends would support Kegley and Raymond's contention that alliance norms have evolved over time (When Trust Breaks Down). Back.

Note 78: If we had a theory of alliance formation sufficiently robust to suggest a set of measurable independent variables for which we could control, we might yet find an underlying difference in the alliance propensities of democratic and nondemocratic states. Back.

Note 79: See Kiefer, "Economic Duration Data and Hazard Functions," for an extensive review. For a discussion that is more oriented to political science, see Bienen and van de Walle, Of Time and Power, ch. 3. Other basic sources would include Kalbfleisch and Prentice, The Statistical Analysis of Failure Time Data; Lawless, Statistical Models and Methods for Lifetime Data; Allison, Event History Analysis; Lancaster, The Econometric Analysis of Transition Data; and Lee, Statistical Methods for Survival Data Analysis. Back.

Note 80: See Lee, Statistical Methods for Survival Data Analysis, chs. 4 and 5. Back.

Note 81: A generalized Wilcoxon rank test shows the difference between the survival functions for democratic alliances and for nondemocratic and mixed alliances to be significant at above the .005 level. Back.

Note 82: The accelerated failure time model is equivalent to the more common proportional hazards model for the special case of the Weibull distribution, as I use here. The estimation is done using the LIFEREG procedure in SAS. Back.

Note 83: The data were also run with exponential, gamma, lognormal, and loglogistic models. For all of the data sets, either the Weibull provided the best fit or it provided a fit that was statistically indistinguishable from the gamma. I have used the Weibull throughout because it is one parameter more straightforward than the gamma and because of the benefits of consistency in having to display and explain the results from only one distribution. Back.

Note 84: See Kalbfleisch and Prentice, The Statistical Analysis of Failure Time Data, ch. 3. Back.

Note 85: These seven cases involve 38 dyads. The inclusion of these dyads weakens the fit of the model but does not substantively or significantly change the results. Back.

Note 86: These values are chosen simply for illustration. Durations are predicted to get longer by between 0.7 and 1.5 percent per year. The effects of differences in power between alliance partners are not particularly stable across the models, so there is no simple description of the difference between two major powers and the other configurations. Back.

Note 87: Partitioning the data set too finely is difficult because of the small number of democratic cases. The results change very little if we drop the nineteenth-century cases. The direction of the coefficients is the same if we restrict the cases to those alliances starting before 1930 (thus eliminating the NATO partners from the analysis), but the significance and stability of the coefficients are greatly limited. Back.

Bibliography

Aldrich, John, John Sullivan, and Eugene Borgida. 1989. "Foreign Affairs and Issue Voting." American Political Science Review 83(1): 123-141.

Allison, Paul D. 1984. Event History Analysis: Regression for Longitudinal Event Data. Beverly Hills: Sage.

Almond, Gabriel. 1950. The American People and Foreign Policy. New York: Praeger.

Anderson, Lisa. 1997. "Politics Unleashed: Liberalization and Radical Politics in the Arab World," ch. 4 in this volume.

Axelrod, Robert. 1984. The Evolution of Cooperation. New York: Basic Books.

Babst, Dean. 1972. "A Force for Peace." Industrial Research (April): 55-58.

Barnett, Michael N., and Jack S. Levy. 1991. "Domestic Sources of Alliances and Alignments: The Case of Egypt, 1962-73." International Organization 45(3): 369-395.

Bienen, Henry, and Nicholas Van de Walle. 1991. Of Time and Power. Stanford: Stanford University Press.

Bobbio, Norberto. 1990. Liberalism and Democracy. Translated by Martin Ryle and Kate Soper. London: Verso.

Bueno de Mesquita, Bruce. 1981. The War Trap. New Haven: Yale University Press.

Bueno de Mesquita, Bruce, and David Lalman. 1986. "Reason and War." American Political Science Review 80(4): 1113-1130.

------ 1992. War and Reason: Domestic and International Imperatives. New Haven: Yale University Press.

Burley, Anne-Marie. 1993. "International Law and International Relations Theory: A Dual Agenda." American Journal of International Law 87:205-238.

Chan, Steve. 1984. "Mirror, Mirror on the Wall . . . Are the Freer Countries More Pacific?" Journal of Conflict Resolution 28:617-648.

Converse, Philip. 1964. "The Nature of Belief Systems in Mass Publics." In David Apter, ed., Ideology and Discontent, pp. 206-261. New York: Free Press.

Degenhardt, Henry. 1986. Treaties and Alliances of the World. 4th ed. Essex: Longman Group.

Dixon, William. 1993. "Democracy and the Management of International Conflict." Journal of Conflict Resolution 37(1): 42-68.

----. 1994. "Democracy and the Peaceful Settlement of International Conflict." American Political Science Review 88(1): 14-32.

Doyle, Michael. 1983a. "Kant, Liberal Legacies, and Foreign Affairs," part 1. Philosophy and Public Affairs 12(3): 205-235.

----. 1983b. "Kant, Liberal Legacies, and Foreign Affairs," part 2. Philosophy and Public Affairs 12(4): 323-353.

----. 1986. "Liberalism and World Politics." American Political Science Review 80(4): 1151-1169.

Fearon, James. 1994. "Domestic Political Audiences and the Escalation of International Disputes." American Political Science Review 88(3): 577-592.

Gaubatz, Kurt Taylor. 1995. "Intervention and Intransitivity: Public Opinion, Social Choice, and the Use of Military Force Abroad." World Politics 47(4): 534-554.

Grenville, J. A. S. 1974. The Major International Treaties: 1914-1973. New York: Stein and Day.

Grenville, J. A. S., and Bernard Wasserstein. 1987. The Major International Treaties Since 1945. London: Methuen.

Grieco, Joseph. 1988. "Anarchy and the Limits of Cooperation: A Realist Critique of the Newest Liberal Institutionalism." International Organization 42(3): 485-507.

Hartz, Louis. 1955. The Liberal Tradition in America. New York: Harcourt, Brace, Jovanovich.

Hochschild, Jennifer. 1981. What's Fair. Princeton: Princeton University Press.

Holsti, Ole. 1992. "Public Opinion and Foreign Policy: Challenges to the Almond-Lippmann Consensus." International Studies Quarterly 36(4): 439-466.

Holsti, Ole, Terrence Hopmann, and John Sullivan. 1973. Unity and Disintegration in International Alliances: Comparative Studies. New York: Wiley.

Huntington, Samuel P. 1991. The Third Wave: Democratization in the Late Twentieth Century. Norman: University of Oklahoma Press.

Hurst, Michael. 1972. Key Treaties for the Great Powers: 1814-1914. 2 vols. New York: St. Martin's.

Jervis, Robert. 1989. "Rational Deterrence: Theory and Evidence." World Politics 41(2): 183-207.

Kalbfleisch, John D., and Ross L. Prentice. 1980. The Statistical Analysis of Failure Time Data. New York: Wiley.

Kant, Immanuel. [1794, 1970] 1991. "Idea for a Universal History with a Cosmopolitan Purpose." In Kant: Political Writings. Edited by Hans Reiss. Translated by H. B. Nisbet. 2d ed., pp. 41-53. Cambridge: Cambridge University Press.

----. [1795, 1970] 1991. "Perpetual Peace: A Philosophical Sketch." In Kant: Political Writings. Edited by Hans Reiss. Translated by H. B. Nisbet. 2d ed., pp. 93-130. Cambridge: Cambridge University Press.

Karl, Terry, and Philippe Schmitter. 1994. "Democratization Around the Globe: Opportunities and Risks." In Michael T. Klare and Daniel C. Thomas, eds., World Security: Challenges for a New Century, pp. 43-62. New York: St. Martin's.

Kegley, Charles, and Gregory Raymond. 1990. When Trust Breaks Down: Alliance Norms and World Politics. Columbia: University of South Carolina Press.

Keohane, Robert. 1984. After Hegemony: Cooperation and Discord in the World Political Economy. Princeton: Princeton University Press.

Keohane, Robert, and Joseph Nye. 1977. Power and Interdependence: World Politics in Transition. Boston: Little, Brown.

Kiefer, Nicholas. 1988. "Economic Duration Data and Hazard Functions." Journal of Economic Literature 26: 646-679.

Klingberg, Frank. 1952. "The Historical Alternation of Moods in American Foreign Policy." World Politics 4(2): 239-273.

Lalman, David. 1988. "Conflict Resolution and Peace." American Journal of Political Science 32(3): 590-615.

Lancaster, Tony. 1990. The Econometric Analysis of Transition Data. Cambridge: Cambridge University Press.

Lawless, Jerald. 1982. Statistical Models and Methods for Lifetime Data. New York: Wiley.

Lee, Elisa. 1992. Statistical Methods for Survival Data Analysis. 2d ed. New York: Wiley.

Lippmann, Walter. 1955. Essays in the Public Philosophy. Boston: Little, Brown.

Lowe, C. J. 1967. The Reluctant Imperialists: British Foreign Policy, 1878-1902. Vol. 1. London: Routledge and Kegan Paul.

Machiavelli, Niccoló. 1970 (1530). The Discourses. Translated by Leslie Walker. Edited by Bernard Crick. Middlesex, Eng.: Penguin.

McLean, Iain. 1987. Public Choice: An Introduction. Oxford: Basil Blackwell.

Manning, D. J. 1976. Liberalism. New York: St. Martin's.

Maoz, Zeev, and Nasrin Abdolali. 1989. "Regime Types and International Conflict, 1816-1976." Journal of Conflict Resolution 33(1): 3-35.

Maoz, Zeev, and Bruce Russett. 1993. "Normative and Structural Causes of Democratic Peace, 1946-1986." American Political Science Review 87(3): 624-638.

Monroe, Alan. 1979. "Consistency Between Public Preferences and National Policy Decisions." American Politics Quarterly 7(1): 3-19.

Neff, Stephen C. 1990. Friends But No Allies: Economic Liberalism and the Law of Nations. New York: Columbia University Press.

Nincic, Miroslav. 1992. Democracy and Foreign Policy: The Fallacy of Political Realism. New York: Columbia University Press.

North, Douglas, and Barry Weingast. 1989. "Constitutions and Commitment: The Evolution of Institutions Governing Public Choice in Seventeenth-Century England." Journal of Economic History 49(4): 803-832.

Page, Benjamin, and Robert Shapiro. 1983. "Effects of Public Opinion on Policy." American Political Science Review 77(1): 175-190.

------ 1991. The Rational Public: Fifty Years of Trends in Americans' Policy Preferences. Chicago: University of Chicago Press.

Putnam, Robert. 1988. "Diplomacy and Domestic Politics: The Logic of Two-Level Games." International Organization 42(3): 427-460.

Rawls, John. 1993. Political Liberalism. New York: Columbia University Press.

Riker, William. 1982. Liberalism Against Populism. San Francisco: W. H. Freeman.

Russett, Bruce. 1971. "An Empirical Typology of International Military Alliances." Midwest Journal of Political Science 15(2): 262-289.

------ 1990. Controlling the Sword: The Democratic Governance of National Security. Cambridge: Harvard University Press.

------ 1993. Grasping the Democratic Peace: Principles for a Post-Cold War World. Princeton: Princeton University Press.

Sabrosky, Alan. 1980. "Interstate Alliances: Their Reliability and the Expansion of War." In J. David Singer, ed., The Correlates of War II. New York: Free Press.

Schelling, Thomas. 1960. The Strategy of Conflict. Cambridge: Harvard University Press.

------ 1966. Arms and Influence. New Haven: Yale University Press.

Shapiro, Robert, and Benjamin Page. 1988. "Foreign Policy and the Rational Public." Journal of Conflict Resolution 32(2): 211-247.

Singer, J. David, and Melvin Small. 1966. Annual Alliance Membership Data: 1815-1965. ICPSR #5602. Ann Arbor: ICPSR.

------ 1984. Wages of War, 1816-1980: Augmented with Disputes and Civil War Data. ICPSR #9044. Ann Arbor: ICPSR.

Siverson, Randolph, and Juliann Emmons. 1991. "Birds of a Feather: Democratic Political Systems and Alliance Choices." Journal of Conflict Resolution 35(2): 285-306.

Small, Melvin, and David J. Singer. 1976. "The War-proneness of Democratic Regimes, 1816-1965." Jerusalem Journal of International Relations 1(4): 50-69.

Snyder, Glenn, and Paul Diesing. 1977. Conflict Among Nations: Bargaining, Decision Making, and System Structure in International Crises. Princeton: Princeton University Press.

Spence, Michael. 1974. Market Signaling: Informational Transfer in Hiring and Related Screening Processes. Cambridge: Harvard University Press.

Taylor, A. J. P. 1961. The Origins of the Second World War. 2d ed. New York: Fawcett.

Thucydides. 1951 (400 b.c.). <i>The Peloponnesian War. Translated by Richard Crawley. New York: Random House.

Tocqueville, Alexis de. 1969 (1835). Democracy in America. Translated by George Lawrence. Garden City, N.Y.: Doubleday.

Velde, François, and Thomas Sargent. 1995. "The Macro-economic Causes and Consequences of the French Revolution." Journal of Political Economy 103:474-518.

von Glahn, Gerhard. [1965] 1992. Law Among Nations: An Introduction to Public International Law. 4th ed. New York: Collier.

Weede, Erich. 1984. "Democracy and War Involvement." Journal of Conflict Resolution 28(4): 649-664.

Walt, Stephen. 1987. The Origin of Alliances. Ithaca: Cornell University Press.

Waltz, Kenneth. 1967. Foreign Policy and Democratic Politics: The American and British Experience. Boston: Little, Brown.

------ 1979. Theory of International Politics. Reading, Mass.: Addison-Wesley.

Liberalization and Foreign Policy