I explain the persuasive core of the logic underlying the theory of liberal democratic peace logic in three places. The two-part essay “Kant, Liberal Legacies and Foreign Affairs” published in Philosophy and Public Affairs (1983) showed how Immanuel Kant's (1970) 1795 essay, “Perpetual Peace,” could be constructed as a coherent explanation of two important regularities in world politics—the tendencies of liberal states simultaneously to be peace-prone in their relations with each other and war-prone in their relations with nonliberal states. Republican representation, an ideological commitment to fundamental human rights, and transnational interdependence are the three causal mechanisms of the explanation. These are Kant's three “definitive articles”—the constitutional, international and cosmopolitan laws—of the hypothetical peace treaty he asks states to sign. The first part of the two-part essay focuses on the liberal peace and its Kantian sources. The second part of the two-part essay focuses on exposing the dangers of liberal imperialism, liberal aggression and liberal appeasement (Rosato 1996 cites the reprints of the two articles in Debating the Democratic Peace). I also addressed these themes in the American Political Science Review in December 1986 and distinguished Kantian “liberal internationalism” from “liberal pacifism” and “liberal imperialism.” In 1997, in Ways of War and Peace, I distinguished liberalism from the two other major traditions of international thought, Realism and Marxism.

All three have one consistent and key argument: “No one of these constitutional, international or cosmopolitan sources is alone sufficient, but together (and only where together) they plausibly connect the characteristics of liberal polities and economies with sustained liberal peace” ([1983a, 1983b] 1996, 27). I repeat the same sentence as the summary of the argument—“No single constitutional, international … ” in the American Political Science Review (Doyle 1986, 1162); and the identical sentence (this time in italics for emphasis) in Ways of War and Peace (1997, 284). I explicitly say that the three causes explained liberal peace and liberal war when, and only when, combined. Rosato's critique of the work, nonetheless, rests on treating each of these factors—“normative,” “institutional,” etc.—in isolation as if they were sufficient.

This is important because, in my view, no one of the factors alone is a sufficient explanation of the liberal peace or liberal war. First, as Rosato correctly suggests, there is no reason for all direct or indirectly majoritarian governments to be peaceful toward other majoritarian governments. Clearly, a democracy of xenophobes or hyper-nationalists would externalize their preferences. Anticipating Rosato's critique, I (1997, chaps. 4 and 9) pointed out in Ways of War and Peace—in chapters that discuss Rousseau and Marx—that democratic institutions are completely compatible with Realist foreign policy when preferences are integrally and exclusively nationalist and with Socialist solidarity and international class warfare when strictly egalitarian (and societies lack individual liberties and private property). Jean Jacques Rousseau's classic account of democratic theory, for example, anticipates that democracies will be locked, as any Realist would agree, in a generalized “state of war” with all other states, whether democratic or not (Rousseau 1756/1917). If information flows across borders are limited, subject to manipulation and nationalist myth-making, and each democracy cultivates a normative commitment to complete autonomy and self-help, democracies will be likely to clash (Mearsheimer 1990; VanEvera 1990).

Second, there should be no expectation that a population widely sharing liberal values associated with human rights norm will shape policy unless they have democratic representation with the transparency and accountability that can shape public decision-making.¹

And third, there is no guarantee that commercial and other forms of interdependence will alone provide material foundations for cooperation among societies, rather than for sources of imperial rivalry and fuel to balance of power competition, unless trade and investment are part of a relationship of trust and respect.²

Rosato (2003) is right to criticize the logic of each of these strands of the “democratic peace” standing alone.

Once combined, however, the three sources do help explain why liberal states maintain peace with each other and are, nonetheless, and for explicable reasons prone to war and imperialism with nonliberal states. Given the absence of this explanation in Mr. Rosato's (2003) critique, it is worth briefly summarizing the three hypotheses here. Each emphasizes one aspect of what characterizes a liberal republic.

First, republican representative democratic governments tend to create an accountable relationship between the state and the voters, particularly median voters. They preclude monarchs or dictators turning their potentially aggressive interests into public policy while assuming that the costs will be borne by a subordinate public. Democratic representation introduces republican caution, Kant's (1970) “hesitation,” in place of autocratic caprice. Representative government allows for a rotation of elites. This encourages a reversal of disastrous policies as electorates punish the party in power with electoral defeat. Legislatures and public opinion further restrain executives from policies that clearly violate the obvious and fundamental interests of the public, as the public perceives those interests.

As importantly, representation together with transparency (what Kant [1970] called “publicity”) may provide for effective signaling, assuring foreign decision makers that democratic commitments are credible because rash acts and exposed bluffs will lead to electoral defeat. Able to make more credible commitments than regimes with more narrow selectorates, democracies would thus be less likely to stumble into wars.³

We should not, however, overemphasize rational signaling. The division of powers and rotation of elites characteristic of republican regimes can permit mixed signals, allowing foreign powers to suspect that executive policies might be overturned by legislatures, courts, or the next election. On the other hand, the shared powers of republics should encourage better chances for deliberation. Most importantly, the combination of representative institutions and purely rational material interests do not control for the possibility that powerful states can have rational incentives to conquer wealthy and exploit wealthy, weak democracies. If reputations are short and differentiable and supposedly pacifying long-run interests are indeterminate, as they often are, something more than rational material interest will be needed to explain liberal peace.

Representation should, however, ensure that liberal wars are only fought for popular, liberal purposes. This does not produce peace. The historical liberal legacy is laden with popular wars fought to promote freedom, protect private property or support liberal allies against nonliberal enemies.⁴

In order to see how the pacific union removes the occasion of wars among liberal states and not wars between liberal and nonliberal states, we need to shift our attention from liberal representation to liberal principles and liberal interests, the other two elements in the liberal explanation of peace and war. These latter two elements account for the purposes that representative processes promote and what credible signaling needs to signal.

Second, liberal principles add the prospect of international respect. Liberal principles, or norms, involve an appreciation of the legitimate rights of all individuals. Connecting these principles to public policy requires publicity. Domestically, publicity helps ensure that the officials of republics act according to the principles they profess to be just and according to the interests of the electors they claim to represent. Internationally, free speech and the effective communication of accurate conceptions of the political life of foreign peoples are essential to establish and preserve the understanding on which the guarantee of respect depends.

These principles begin the differentiation of policy toward liberal and nonliberal states, requiring trust of and accommodation toward fellow liberals and producing distrust of and opposition toward nonliberals. Domestically just republics, which rest on the consent of free individuals, presume foreign republics to be also consensual, just, and therefore deserving of the accommodation that the individuals that compose them deserve. The experience of cooperation helps engender further cooperative behavior when the consequences of state policy are unclear but (potentially) mutually beneficial. At the same time, liberal states assume that nonliberal states, which do not rest on free consent, are not just. Because nonliberal governments are perceived to be in a state of aggression with their own people, their foreign relations become, for liberal governments, deeply suspect. In short, fellow liberals benefit from a presumption of amity; nonliberals suffer from a presumption of enmity. Both presumptions may be accurate. Each, however, may also be self-fulfilling.

Democratic liberals do not need to assume either that public opinion rules foreign policy or that the entire governmental elite is liberal. They can assume that the elite typically manages public affairs but that potentially nonliberal members of the elite have reason to doubt that antiliberal policies would be electorally sustained and endorsed by the majority of the democratic public.

Third and last, material incentives sustain interliberal normative commitments. The “spirit of commerce” spreads widely and creates incentives for states to promote peace and to try to avert war. Liberal economic theory holds that these cosmopolitan ties derive from a cooperative international division of labor and free trade according to comparative advantage when the parties can expect to be governed by a rule of law that respects property and that enforces legitimate exchanges. Each economy is said to be better off than it would have been under autarky; each thus acquires an incentive to avoid policies that would lead the other to break these economic ties. But, because keeping open markets rests on an assumption that the next set of transactions will also be determined by prices rather than coercion, a sense of mutual security is vital to avoid security-motivated searches for economic autarky. Thus, avoiding a challenge to another liberal state's security or even enhancing each other's security by means of alliance naturally follows economic interdependence.

In this same regard, a further cosmopolitan source of liberal peace is that the international market removes difficult decisions of production and distribution from the direct sphere of state policy. A foreign state thus does not appear directly responsible for these outcomes; states can stand aside from, and to some degree above, these contentious market rivalries and be ready to step in to resolve crises. The interdependence of commerce and the international contacts of state officials help create cross-cutting transnational ties that serve as lobbies for mutual accommodation. According to modern liberal scholars, international financiers and transnational and transgovernmental organizations create interests in favor of accommodation. Moreover, their variety has ensured no single conflict sours an entire relationship by setting off a spiral of reciprocated retaliation.

Conversely, the suspicion that characterizes relations between liberal and nonliberal governments can lead to restrictions on the range of contacts between societies. And this can increase the prospect that a single conflict will determine an entire relationship. As importantly, in relations with weak societies, “protecting “native rights” from native oppressors, and protecting universal rights of property and settlement from local transgressions, introduced especially liberal motives for imperial rule” (Doyle [1983a, 1983b] 1996, p. 37). When property lacks clear title and exchanges are subject to manipulation and uncertain legal enforcement—the typical environment of non-liberal states—then economic contact generates strife.

No single constitutional, international, or cosmopolitan source alone is sufficient. This variant of liberal theory is neither solely institutional, nor solely ideological, nor solely economic. But together (and only together) the three specific strands of liberal institutions, liberal ideas, and transnational ties plausibly connect the characteristics of liberal polities and economies with sustained liberal peace. But in their relations with nonliberal states, liberal states have not escaped from the insecurity caused by anarchy in the world political system considered as a whole. Moreover, the very constitutional restraint, international respect for individual rights, and shared commercial interests that establish grounds for peace among liberal states establish grounds for additional conflict in relations between liberal and nonliberal societies.

Thus, when Rosato (2003, 588, 593) criticizes the “norm externalization” argument or the “institutional logic” explanations, he is in each case missing two thirds of the liberal argument. When he argues that economic interests and local strategic interests shaped liberal imperial policy in the 19th century, he is not refuting—he is confirming—the logic of the liberal peace. In these cases, principled liberal motives joined material interests in liberal imperialism. Campaigns against the slave trade destabilized commercial oligarchies, making them prone to collapse. The mission civilatrice and the “dual mandate” imperial ideologies both included liberal principles, albeit ones that allowed for liberal imperial paternalism of the sort J. S. Mill (1859/1973) endorsed for societies he and his fellow liberals saw as incapable of governing themselves. But commercial and property interests, which lacked institutionalization in much of Africa and Asia, were even more important. Lacking both legal recognition and the context of interliberal respect, commercial and property claims fueled imperialism. (Doyle [1983a, 1983b] 1996, 37–9). Liberals were all too ready to enforce those property claims both as a matter of material interest and principled defense of rights. Interliberal peace rests on the combined effect of the three pillars. Absent one of them, pacific policy is underdetermined and undermined.

During the Cold War, the United States did intervene against or take measures to undermine covertly numerous popular regimes in the Third World. In many cases the U.S. administration in office was convinced that the regimes in question (Mossadegh in Iran, Arbenz in Guatemala, Jagan in Guyana, Allende in Chile, and the Sandinistas in Nicaragua) were threats both to property and to the rule of law. The fact that these regimes were more progressive and popular than any previous regime in those countries (and, in some cases, since) did not make them well-established liberal democracies. Many U.S. officials doubted their stability as democracies. They were also seen as influenced by and allied with communist regimes. President Kennedy articulated the logic clearly, referring to the assassination of Trujillo in the Democratic Republic: “There are three possibilities in descending order of preference, a decent democratic regime, a continuation of the Trujillo regime or a Castro regime. We ought to aim at the first, but we cannot really renounce the second until we are sure that we can avoid the third” (Schlesinger 1965, 769, quoted in Doyle [1983] 1996, 41). As importantly, all of these interventions were covert; they lacked the mechanisms of publicity on which the liberal peace rests. The explanation underlying the liberal peace makes no assumption that every official, always and everywhere, is motivated by liberal principle and interest—just that over the normal political cycle nonliberal principles and interests will not become the norm in the formation of liberal foreign policy.

A much more logical explanation comes with methodological costs. Data sets on the liberal peace do not adequately code for these three pillars together and separately. My own coding (1983a, 1983b, 1986, 1997) was approximate. The most thorough recent empirical test of Kantian propositions (Russett and O'Neal 2001) shows the separate positive effects of democratic institutions and trade (and membership in international organizations), but it doesn't separately code for liberal norms. The substantial statistical confirmation that inter-democratic peace, (coding for democratic institutions), does receive is thus probably a reflection of the tendency for principles of liberal individualism and democratic institutions to evolve together.⁵

But we cannot be sure of this. Compared to other testable international theories of similar scope, the empirical confirmation of the liberal peace is exceptionally strong, but that does not mean that the theory does not need additional testing.