War and Constitutional Design, T Ginsburg, F de Vitoria

Tags: constitutions, Cambridge University Press, democracies, New York, legislative involvement, Princeton, peace literature, John A. Vasquez, legislature, democratic peace, results, Rowman & Littlefield, J. David Singer, War and Peace, military capabilities, constitutional design, Comparative Constitutions Project, American Political Science Review, International Security, The Yale Law Journal, International Organization, causes of war, conflicts, legislatures, Oxford University Press, Princeton University Press, Democracies at War, International, Eric Posner, democratic institutions, Congress 200, legislative power, declaration of war, the War Powers Act, Julius Caesar, President Barack Obama, mature democracies, University of Chicago Law School, Tom Ginsburg, academic literature, United States Armed Forces, David Cameron, Harvard University, Brookings Institution Press, Michael B. MacKuen, University Press, War Powers, John Sherman Cooper, Thomas Jefferson, dictatorships, war power, Stanford University Press, Ryan Schildkraut, Michigan Journal of International Law 27:1239-59, Constitution of the United States, President Clinton, the war powers resolution, power to declare war, Congress, James Lee, John Yoo, Committee on Foreign Relations, Foreign Force, constitutional rules
Content: WAR AND CONSTITUTIONAL DESIGN Tom Ginsburg* version 2.0 *** " a king is not by himself capable of examining into the causes of war and the possibility of a mistake on his part is not unlieklye and such a mistake would bring great evil and ruin to multitudes. Therefore war ought not to be made on the sole judgment of the king, nor, indeed, on the judgement of a few, but in that of many, and they wise and upright men." -Francisco de Vitoria (1532) *** In late August 2013, U.S. President Barack Obama, along with his close allies David Cameron of Britain and Francois Hollande of France, considered a military response to Syrian government chemical attacks on civilians. Prime Minister Cameron went to parliament to ask for support, but on August 30 he lost the vote 285-272, and announced that he would respect parliament's decision. The French parliament also debated the issue, although it was not * Leo Spitz Professor of International Law, University of Chicago Law School. Thanks to Adam Chilton, Dhammika Dharmapala, Willliam Hubbard, Yonatan Lupu, Richard McAdams, Monika Nalepa, Eric Posner, Mila Versteeg, and Matt Waxman, along with audiences at the PIPES and Comparative Politics Workshops at the University of Chicago, the University of Chicago Law School Conference on the Law and Economics of International Law, Tel Aviv University School of Law, for helpful comments. Albert Chang, Emily Heasley, David King and Jeyshree Ramachandran provided helpful research assistance. Robert Schub provided great help with data. 1
constitutionally required to hold a vote. And, a few days later, Obama surprised observers by announcing that he would also consult with the U.S. Congress in considering his decision, though he carefully refrained from saying he would abide by the resolution. As Members of Congress remained aloof on the issue and it appeared that Obama would lose the vote, Vladimir Putin emerged as an unlikely savior, seizing on an offhand comment of Secretary of State John Kerry and proposing a peaceful resolution under which Syria would dismantle its chemical weapons in exchange for the U.S. foregoing an attack. The Congressional vote never happened, and neither did the strike. Would Obama have attacked Syria without Congressional authorization? We can only speculate. But what is not in doubt is that the decisions, by both Cameron and Obama, to approach reluctant legislative bodies for approval raised the costs of military action, and contributed to the executive backing down from earlier threats. And it was precisely this effect that the framers of the U.S. Constitution had in mind when they designed our constitutional scheme to chain the proverbial "dogs of war."1 The overall scheme of the Constitution, wrote John Jay, would be such that "not only fewer just causes of war will be given by the national government, but it will also be more in their power to accommodate and settle them amicably."2 The first two rows of Figure 1 attempt to capture this intuition in the framework of the academic literature on war: although conflict is typically modeled as involving two unitary states, the American founders recognized that internal features of state decision-making would affect outcomes on the international plane. More recent constitutions tend to provide for 1 See Boyd 1958, 397; and Wormouth and Firmage 1989. The Dogs of War phrase comes from Shakespeare's Julius Caesar. Shakespeare, Julius Caesar, act 3. scene 1, line 274. 2 Jay 1987a. 2
legislative approval of actions of the commander in chief, which occurs even earlier in the sequence of international interaction. But the basic intuition of the framers that legislative constraints can affect international policy seems intact. FIGURE 1 HERE The framers, of course, lived in a very different era, in which assumptions about military capabilities, threats to national security, and the functioning of the executive and legislature were very different from the situation today. But in light of the lingering hagiographic views of the framers' wisdom, it is worth asking whether constitutional design can indeed affect a nation's performance in international crisis bargaining and violent conflict. The question is surely relevant in an era of ubiquitous constitutional design. Somewhere between five and ten countries every year will be engaged in overhauling their constitutions, and might be interested to know whether anything they say might actually affect national security.3 And the question relates to international efforts to reduce the levels and costs of violent conflict. To understand why constitutional design might matter, we need to delve into the large literature in International Relations theory on the democratic peace. The central findings of this literature can be summarized briefly: (1) mature democracies generally do not go to war against each other; (2) democracies are, on the whole, no less warlike (or only slightly less warlike) than autocracies; and (3) democracies are likely to win the wars that they fight against autocracies.4 To illustrate, by one account, democracies win over three­quarters of their wars, and over 93 3See generally, Melkinsburg 2009. 4 The middle proposition of this trifecta is contested. For one view, see Reiter and Stam 2003 (autocracies initiate conflicts against democracies more frequently than democracies do against autocracies); and Quackenbush and Rudy 2006 (democracies initiate wars against autocracies more frequently than autocracies do against each other); see also Doyle 1983, 225 ("Liberal states are as aggressive and war prone as any other form of government or society in their relations with nonliberal states."); and Dixon 1994, 1 ("Kant was clearly wrong in his presumption that democracies are inherently peaceful."); but see Ray 2000, 300-302 (democracies may be more peaceful). 3
percent of wars they initiate.5 Putting propositions (1) and (2) together (and setting asides debate over proposition (2)), we can see that democracies are more likely to go to war against autocracies (because they are not fighting fellow democracies, while autocratic governments do fight autocracies). These statistical regularities, however, are not self-explanatory, and there is a large debate on the sources of the democratic peace.6 In thinking about the project of optimal constitutional design, we need to understand why it is that democracies may have certain advantages in war. In particular, it would be helpful to know whether constitutional choices may have anything to do with these outcomes, as the American framers imagined. A comparative inquiry into the effects of domestic constraints on war and coercive bargaining may have normative implications for internal separation of powers discourse in the United States as well. Our constitutional rules are argued by some to be unclear: the legislature declares war but the president serves as Commander in Chief and thus prosecutes war. There has been a major debate, revived since 9/11, over how much room to give the executive in warmaking.7 Only if we can determine that constitutional choices make a difference can we answer the contested normative question of optimal design. This paper examines the internal structure of constitutional war-making power as a possible source of democratic advantage in crisis bargaining and war. In particular, it asks whether legislative involvement in decision-making about war, or supervision of military 5 Reiter and Stam 2003, 29. For a different view on democracies and war, see Desch 2002. Even though Desch questions whether democracies enjoy any advantages in conflict, he does not support the contention that there is any disadvantage associated with democracy. See Reiter 2003; and Hemel 2007. 6 Indeed, there have been some recent suggestions that the democratic peace is illusory--that it is driven by trade patterns or liberal political economies rather than regime type per se. See, e.g., Gartzke 2007, 166-91; and McDonald 2009. Maoz and Russett argue that the democratic peace is driven by political stability. Maoz and Russett 1993. Others focus on alliance ties. See Ray 2000, 306. 7 Nzelibe and Yoo 2006; Posner and Vermuele 2011. 4
authorities, leads to better war selection and performance. In doing so, it draws on data from the Comparative Constitutions Project, which measures features of formal constitutions for independent nation-states since 1789, as well as data from the Militarized Interstate Disputes (MID) dataset of the Correlates of War Database, frequently used in the study of conflict.8 By examining internal features of state structure beyond regime type, the paper opens up a new line of inquiry into the domestic institutional sources of the democratic peace.9 Other lines of research in international law, such as scholarship on human rights, have gone much further to unpack the state and examine the domestic sources of international behavior. The argument here is that legislative involvement in decisions about war implicates a bargaining process between the executive and legislature, which can interact with the international level of state-to-state bargaining to affect outcomes of conflict. Legislatures can restrain overly ambitious executives, mobilize support, and affect the quality of signaling to other states. We focus especially on legislative powers to declare or approve war, or otherwise approve the actions of a commander in chief. These powers have implications for international interactions far short of full-fledged war, as crisis bargaining often occurs in the shadow of threats to go to war (see Figure 1). To be sure, constraints on formal decision-making about war are only a subset of potential tools that might be used in the bargaining process between executive and legislature. The budget process, for example, will determine who controls military capabilities, and a powerful legislature can do much with audit and oversight mechanisms as well. Regulations about the chain of command, including designation of a commander in chief, will play a role. 8 Militarized Interstates Dispute Data Set (v. 3.0) available at http://www.correlatesofwar.org/ (last checked April 4, 2014). 9 See also cites in Gibler 5
The relationship between civil and military authorities may also be specified in the constitution, for example through a requirement that the head of the military or minister of defense be a civilian. This can serve as a form of agency control to restrict aggrandizing militaries. Some constitutions also have restrictions on the armed forces, including limitations on whether officers can hold public office.10 Finally, a small number of constitutions outlaw anything but defensive war,11 or in very rare cases abolishing the armed forces entirely, as in post-war Japan's famous "peace constitution" and in Costa Rica.12 These constitutional constraints also make a difference in practice. In this paper, we focus on the decision to go to war, rather than these collateral institutions, which admittedly affect the decision-making environment for military action. An underlying assumption, which we will defend, is that constitutional assignment of powers is one factor that affects the interaction between legislature and executive. One need not assume that actual behavior perfectly conforms to the constitutional text to believe that the formal assignment of constitutional power can lead an actor to defend its prerogatives. The paper proceeds as follows. We first review, in Part I, the literature on domestic constraints on warmaking, and their role in war selection and success. We then examine specific constitutional institutions that are implicated in this argument, and describe the data from the Comparative Constitutions Project. Part III conducts an empirical analysis. We show that legislative involvement in war decision-making reduces the propensity of states to use force abroad. But we also find that, once war is triggered, legislatures are not associated with a greater 10 Some 11% of national constitutions have such a restriction. 11 See Kuwait Constitution [1962], art. 68; and Bahrain Constitution [1973], art.36. See also Dodging the Draft: Federal Constitutional Court Evades Review of Germany's Military Service Law, 3 German Law Journal (2002), available athttp://www.germanlawjournal.com/index.php?pageID=11&artID=151 12 Costa Rica Constitution [1949], art. 12; Japan Constitution [1947], art. 9. 6
probability of victory. These findings take us one step closer toward identifying the precise institutions which serve to constrain or empower democracies, and so sheds light not only on the democratic peace but also on optimal constitutional design. I. Why Do Democracies Succeed? The Virtues and Vices of Legislative Involvement in Warmaking A. Legislative power over warmaking Wars result from a complex set of calculations by domestic actors in (at least) two different states.13 Before going to war, states typically engage in what is known in the literature as crisis bargaining, a communicative process of signaling about capabilities and resolve.14 This bargaining process allows countries to, in many cases, anticipate the costs and benefits of a war and reach a negotiated agreement, without the costs that a war would impose. If this bargaining process worked perfectly, we would never observe actual war, which by definition involves the destruction of human and material resources. However, in a world of imperfect information about capabilities and intentions, bargaining problems, and agency problems in the decisionmaking process, armed conflict does occur. A key question for constitutional designers from the very beginning of the modern era has been how to "chain the dogs of war", as the American founders put it.15 Recognizing that war is the most serious decision that a leader can take, the American founders debated whether to allow 13 Fearon 1995; and Posner and Sykes 2013, 168-71 (summarizing argument about role of information on decisions to go to war). 14 Leng 2000. 15 Wormuth and Firmage 1989. 7
the executive to make the decision unconstrained, as might a monarch. Instead, the framers quite self-consciously located the power of declaring war in the U.S Congress.16 As Table 1 below shows, some 25% of all constitutions written thereafter have followed suit, vesting the war declaration function in one or two houses of the legislature.17 A further 38% of constitutions have some role for the legislature in approving declarations of war or deployment of troops. Altogether nearly half of all constitutions involve the legislature in the process of making decisions about war or the use of force. TABLE 1 HERE B. Why Legislative Involvement Might Matter How might legislative involvement change the propensity of states to become involved in war? Several lines of argument from the democratic peace literature are particularly relevant here. One set of arguments focuses on the institutional structures of democratic governance. Legislatures can, most obviously, constrain over-zealous executives who wish to go to war by raising the costs of doing so, requiring evidence to justify the decision, and limiting budgets.18 So, for example, writing on the US, William Howell and Jon Pevehouse argue that Congress has indeed constrained presidents in war policy, notwithstanding the dearth of formal declarations of war.19 They show that Presidents are systematically less likely to exercise military force when 16 U.S. Constitution, art. 1, sec. 8. 17 Data on file with author. 131 of these constitutions give it to the first house; 11 to the second house; 67 give it to both houses acting jointly. Some five cases (Nicaragua 1939 and 1974; El Salvador 1872; Morocco 1962 and Italy 1947 allow either of two houses to declare war.) 18 See Isaacharoff 2009. 19 Howell and Pevehouse 2008. But see Fisher et al 2008. 8
their partisan opponents retain control of Congress. 20 Since such partisan constraints are by definition more likely in democracies than in autocracies, we might observe that democrats engage in fewer foreign adventures. This is essentially a screening argument: the assumption is that the second look by the legislature will produce a better selection of wars actually prosecuted, including a higher proportion that are won. It may rely on the simple mechanism of a having a second decisionmaker, or it may have to do with particular institutional qualities of executive and legislature. Either way, the involvement of the legislature constrains, leading to better war selection. Legislative involvement has several other theoretical advantages for war policy beyond constraint. Legislatures can also mobilize support for popular wars, for example by demonstrating bipartisan support, or through elaborating on the rationale for going to war. This has been a theme of the literature on democracies and war since the time of Thucydides.21 Legislative deliberation and debate can be a crucial mechanism for transmitting information to the public, possibly generating support.22 This support in turn can enhance the quality of the signal given in crisis bargaining at the international level. If a decision to go to war involves both an executive and legislative branch, a counter-party will have an indication of the level of support for the war. It will assume that there is a good deal of domestic support, which in turn, may lead the party to believe the executive branch during negotiations. Legislative involvement thus provides a costly signal of seriousness and commitment. Not only will a country with 20 "The partisan composition of Congress, however, matters most for proposed deployments that are larger in size and directed at less strategically important locales. Moreover, congressional influence is often achieved not through bold legislative action but through public posturing--engaging the media, raising public concerns, and stirring domestic and international doubt about the United States' resolve to see a fight through to the end." But see Fisher and Hendrickson 2008. 21 Thucydides, The Pelopennesian War. 22 See generally Hess 1981. 9
legislative involvement avoid "bad" wars entered into by an overzealous executive, but it will have bargaining advantages vis-а-vis other states because executive representations of public support will be more credible. Signaling enhances probability of selecting "good" wars. Beyond enhancing the ability to send signals in international crisis bargaining, legislatures may contribute to the ability to receive signals. While the executive branch may have the exclusive ability to communicate with foreign powers on behalf of the state, it may have to provide evidence of these communications to legislators to gain support. Legislators can, through probing questions and their own stock of information, provide for alternative interpretations of evidence.23 We know that wars sometimes arise from misperception--the Six Day War between Israel and its Arab neighbors is a good example.24 Ceteris paribus, more institutions involve more independent evaluations of information, and hence may reduce the noise in received signals from abroad. Like the Condorcet jury theorem, more institutions assessing evidence independently might lead to higher quality aggregate decision-making. In short, legislatures may reduce the propensity to enter into war by screening international conflicts and restraining the executive before escalation; and increase the incentives to settle by sending and receiving more accurate signals. War is in the error term; if both parties have accurate information on capabilities and intentions they are likely to settle in the shadow of predicted outcomes.25 23 Scoville 2013. 24 Leng 2000, 251; and Oren 2003. 25 Gartzke 1999. 10
C. Why Legislative Involvement Might Hurt If these were the only considerations, one might expect that more constraints would always be helpful because they would raise the quality of signals sent and received, help generate more public support, and play a screening role to filter out wars without sufficient justification or popularity. The normative implication might be to add additional constraints on warmaking, such as supermajority approval requirements, judicial approval, or a public referendum before going to war.26 More decisionmakers would screen bad wars and generate more support for "good" wars, whatever that might mean in the local context. But there are, as always, countervailing considerations. Proponents of executive-centered decision argue that only the executive can act with the speed and decisiveness needed to be effective.27 Legislative involvement in war-making, in this view, can cause potentially harmful delays, leading to failure to launch "good" wars.28 Proponents of this view share certain assumptions with proponents of congressional ex ante approval, such as John Hart Ely, who argue that deliberation can also prevent or delay "bad" wars.29 But they see a greater risk from too little war than from too much war. Another argument sometimes advanced in favor of exclusive executive control is accountability. For example, in the United States, the president is the only actor elected with a 26 The Polish liberum veto amounted to a kind of unanimity rule, which ended up facilitating foreign interference with the Polish state. 27 Nzelibe and Yoo 2006. 28 Nzelibe and Yoo 2006. 29 Ely 1993. Recent events, however, suggest that Congress need not delay war if it chooses not to. For example, Congress passed the Authorization for the Use of Military Force within a week of the September 11 attacks, yet several weeks passed before large numbers of troops were actually deployed in Afghanistan. The delay in launching the 2003 invasion of Iraq was due not to any congressional opposition or deliberation, but rather resulted from the President's explicit choice to seek United Nations Security Council approval after obtaining congressional support. 11
national mandate. To be fair, some political science literature suggests that the House of Representatives is the governmental actor that is most responsive and accountable to public opinion.30 But proponents of executive control over warmaking argue that because the public attributes responsibility to the executive for foreign policy, the president should be given more power in this arena.31 Power, in other words, should follow accountability, and not the other way around. Finally, it is often argued that the legislative power of the purse provides enough of a mechanism for accountability. Should Congress, in the US context, decide that a war entered into by the executive is a poor choice, it can always cut off funds.32 In this sense, the executive can be held accountable to public opinion ex post control, in other words, can substitute for ex ante control, and so ensure that the executive is accountable for its actions. In short, speed and accountability are the primary virtues of an executive-centered war policy, and so suggest symmetric vices of legislative involvement. D. Discussion On close examination, many of the arguments for executive dominance are less powerful than they first appear. As Paul Diehl and I have argued, the argument that the executive will bear the blame for war policy, and hence should have the primary say, does not defeat the possibility that collectively the executive and legislature may be more accountable than either of them 30 Rudolph 2003 (presenting data and institutional reasons for the public's attribution of greater responsibility for the economy to Congress than to the president, among other actors); and Stimson, MacKuen, and Erikson 1995, 559-60 (indicating that regression estimates show the House to be more responsive than the president, who in turn is more responsive than the Senate). 31 Nzelibe and Yoo 2006, 2522. 32 Yoo 2006. 12
individually.33 The two branches of government represent different electoral constituencies, and so joint decision-making may be accountable to a broader national constituency and to many more diverse groups. To illustrate, suppose the executive is slightly more hawkish than the median voter and the legislature is much more dovish. In such a circumstance, allowing the executive to make decisions about war on its own will produce a more "accountable" policy than would assigning the power exclusively to the legislature, because the executive's position is closer to that of the median voter. Nevertheless, requiring the executive to bargain with the legislature before initiating hostilities has the potential to move the expected policy further in the direction of the median voter. Hence, certain negotiated outcomes will be closer to the position of the median voter than allowing either actor to act alone.34 Indeed, assuming that a sole executive is always superior to one constrained by a legislature is to assume that the executive is always closer to the median voter than is the legislature on war policy. There is no particular reason to believe that this assumption is valid. The assumption includes an additional requirement, not only that the president's position is always closer to that of the median voter in absolute terms, but also that legislature and the executive are always biased in the same direction. In other words, in the U.S. context, if the president is more hawkish than the median voter, Congress is even more hawkish; if the president is more dovish than the median voter, Congress is even more so. Only if this alignment were universally true would a system of sole presidential decision-making always be more representative than a system that requires bargaining before war. 33 Diehl and Ginsburg 2006. 34.In our example, it is also possible to produce bargain outcomes that are further from the position of the median voter than an executive­only policy. This would not be true were the legislative position closer to that of the median voter than the executives, in which case bargaining would always produce more accountable policy. 13
As mentioned above, even if the legislature is not involved in the initial decision to go to war, it retains an important voice in war policy because it can use the threat of cutting off funds ex post. However, this is unlikely to be a perfect substitute for ex ante involvement in decisionmaking, since decisions about war have a path-dependent quality. Once launched they are difficult to reverse, and ex post legislative involvement must confront problems of sunk costs and path dependency. For this reason, ex post constraint through funding has a very poor record in the United States; conversely, even some opponents of the initial decision to go to War in Iraq later supported the surge. President Barack Obama, who campaigned on ending the wars in Afghanistan and Iraq, found himself supporting a surge in Afghanistan. This suggests that ex post constraints are qualitatively different from ex ante constraints, implying that greater legislative involvement in the initial authorization stage might prevent the downstream choice of two bad alternatives, the worse of which is to terminate funding. For all these reasons, legislative involvement might be good for a nation's war-making policy, regardless of the nature of the counter-party. How might legislative involvement affect crisis bargaining? Might the need to obtain legislative support generate net benefits on the international plane? Some have argued that democratic states are at a disadvantage in conflicts with non­democratic opponents because democratic executives must consult with or secure approval from other domestic institutions. This contention is based on the alleged risk that potential opposition to or dissent from the executive's action will undermine the resolve, and thereby the bargaining position, of the country in international confrontations. However, the great weight of studies of the effectiveness of democracies in coercive bargaining and war suggest that better outcomes emerge from consultation and consensus.35 35 See Gaubatz 1999; Huth and Allee 2002; Bueno de Mesquita and Lalman 1992; Dixon 1994. 14
One reason consultation might be helpful is that it might make threats are more credible.36 Shultz contrasts two theories of the democratic advantage in war: an institutional constraint argument, that democratic leaders face higher political costs from waging war and hence are likely not to select losing battles; and an informational theory that suggests that the support of the public provides a better explanation for why war selection is more careful. As he puts it "The institutional constraints argument suggests that democratic leaders generally face higher political costs for waging war. As a result, when a state is challenged by a democracy, the target has reason to doubt that the challenge will actually be carried out. Targeted states should be more likely to resist when threatened by a democracy than when threatened by a state that is not similarly constrained. The informational perspective, on the other hand, suggests that democratic governments are better able to reveal their true preferences in a crisis. Relative to non-democracies, they are less likely to engage in bluffing behavior, meaning that the threats they do make are more likely to be genuine. As a result, the target of a threat made by a democracy should be less inclined to resist or further escalate the crisis." Shultz tests these competing hypotheses and finds that the likelihood of reciprocation is lower when the initiating state is a democracy than when it is not, a result that is consistent with the predictions of the informational perspective. When executive action is undertaken with broad domestic support, Schultz reports that democracies are highly effective, indeed more 36 Check this para. Schultz 1999. 15
successful than their non­democratic counterparts in similar circumstances.37 Even when democracies face dissent, their success rates are no worse than those of non­democracies.38 These findings suggest that, whether or not political opposition is mobilized, seeking support or approval for threats and action does not put democracies at a disadvantage. From a signaling standpoint, there appears to be no disadvantage--and indeed some potential advantages--to bargaining with the legislature. Relatedly, democratic controls can help democracies make commitments that endure over time, making them better "contracting partners" in international negotiations.39 There is, however, a potential dark side to more credibility of commitment. The international relations literature has identified the problem of "gambling for resurrection", in which an executive, having started a war that is not going well, "doubles down" on the effort to avoid being punished at the ballot box.40 While the focus has been on the executive, legislatures too might be subject to this pressure, though their discount rates may be different depending on the security of their seats. Gambling for resurrection might lead to a kind of democratic disadvantage once conflict has begun. As Alexis de Tocqueville said, "there are two things that will always be difficult for a democratic nation: to start a war and to end it."41 37 Schultz 2001, 174. 38 Schultz 2001, 174. 39 Lipson 2003. 40 Downs and Rocke, 1995: 56. 41 De Tocqueville 1850. A second "dark side" may occur when the legislature is more hawkish than the executive. Historical examples include the Senkaku islands dispute of the 1970s. Daniel Tretiak, The Sino-Japanese Treaty of 1978: The Senkaku Incident Prelude, Asian Survey, Vol. 18, No. 12 (Dec., 1978), pp. 1235-1249. The two countries had been negotiating a treaty of friendship for some time, but right-wing anti-treaty members of the Diet in Japan sought to derail the treaty by raising the longstanding 16
In sum, public debate and participation in decision-making may help to transmit clear and reliable information regarding the intentions of democracies to other states. In dictatorships, in contrast, information on intentions and capabilities may be more difficult to discern. This can impede bargained solutions and lead to more wars.42 Dictatorships may also lack means of ensuring credibility of promises they do make, both domestically and internationally. Democracies are better able to make commitments, in part because of domestic constitutional structures, which gives them an advantage in bargaining.43 This implies that dictatorships which share these constitutional features might be able to gain some similar advantages. II. Does Text Matter? Lessons from the US These theoretical advantages of legislative involvement have been debated over the course of American history as Congress and the President have sought to come to an accommodation over their respective roles in war time.44 Clearly the founding fathers intended a reduced role for the executive, relative to the British monarch. As Hamilton put it in Federalist 69, the Constitutional scheme would take the powers of declaring war, raising and regulating armies away from the executive, who would be restricted to an operational role.45 Over the course of American history, however, the formal language of the founders has given way to a more pragmatic accommodation. While the U.S. Constitution clearly assigns to Congress the power to declare war in Article I, the United States has fought only four declared wars in our issue of the islands. The Chinese government reacted by sending an armed flotilla of fishing boats to the island, though disclaiming responsibility. Eventually the situation simmered down. The Japanese government didn't react strongly because it was trying to keep the lid on various members of the Diet and the treaty was signed later that year. 42 Levy and Razin 2004. 43 Lipson 2003. 44 See Hallett 2012; Reveley 1981. 45 Federalist Papers 1789. 17
history: the War of 1812, the Spanish American War, and the two World Wars of the 20th Century. Since the Korean War, presidents have been routinely willing to conduct "limited" wars without Congressional authorization. This leads to the question of whether formal requirements matter at all. To be sure, the question is a bit tricky because constitutional language is obviously not dispositive on its own. Obama's decision referred to at the outset of this paper was surprising precisely because it seemed to be unnecessary according to constitutional convention, and he did not frame his request in terms of a declaration of war under Article I, Sec. 8. One might think, then, that text is superfluous. But textual assignments of powers have the possibility of affecting the bargaining among branches; indeed, the notion that institutions will seek to preserve their prerogatives is at the basis of Madisonian constitutional design. In the U.S., Congress has at times demanded executive recognition of its constitutional power in war policy, even if it has not insisted on formal declarations of war. The War Powers Act of 1973 (passed over President Nixon's veto in 1973) is one famous example. Even as both Republican and Democratic presidents have challenged the constitutionality of the War Powers Act, and asserted that they did not need congressional authorization, they have generally observed its terms. (To be sure they have on occasion bypassed even the moderate reporting requirements of the Act.46) The War Powers Act does shape executive decision-making. In 1993, when there was a concurrent resolution of Congress demanding withdrawal from Somalia, 46 Ronald Reagan's decision to invade Grenada was made without notification of Congress, though he later sought and received authorization to deploy Marines to Beirut. Bill Clinton initiated the bombing of Kosovo in 1999 without any authorization from Congress, and succeeded in defeating the Serbs less than two weeks before the 90-day period of the War Powers Act expired. In that case, Clinton inferred congressional "authorization" from an appropriations statute that neither mentioned the War Powers Resolution nor authorized the military action. See U.S. House 1999. 18
President Clinton complied. Presidents will also report the commitment of troops abroad, sometimes using the language "consistent with the war powers resolution".47 Congress relied heavily on the constitutional baseline in passing the War Powers Act. The War Powers Resolution begins with the statement that "[i]t is the purpose of this joint resolution to fulfill the intent of the framers of the Constitution of the United States and insure that the collective judgment of both the Congress and the President will apply to the introduction of United States Armed Forces into hostilities."48 The introduction goes on to recount the constitutionally assigned powers of the Congress and the President, and asserts that the President is only able to commit forces pursuant to a declaration of war or statutory authorization, or in an emergency.49 Discussions of the Constitution and checks and balances recur throughout the various committee reports that were generated as the legislation wound its way back and forth through the House and Senate. In nearly identical language the various House reports reproduce excerpts from the Constitution, and emphasize that the purpose of the legislation is the "reaffirm" Congressional powers granted by the Constitution.50 The Senate expressed a concern about Congress's powers with respect to "undeclared" wars and also extensively referenced the views of various Framers, Supreme Court Justices, and academics throughout history.51 This emphasis on "restoring" a constitutional balance of powers was present from the very earliest bill in the lineage of what became the War Powers Resolution, intoduced by 47 See e.g. White House, http://www.whitehouse.gov/the-press-office/2013/12/13/message-congressreport-consistent-war-powers-resolution 48 U.S. Congress 1973, sec. 2(a). 49 U.S. Congress 1973, sec. 2(b)-(c). 50 See, e.g., U.S. House 1973a; and U.S. House 1972. 51 See, e.g., U.S. Senate 1973; U.S. Senate 1972. 19
Representative Bertram Podell of New York on April 14, 1969. This bill began by noting that the Constitution grants the power to declare war to Congress and that Congress had been losing this power to the President.52 Constitutional considerations suffused the floor debates on the resolution as well. For example, in the debate about whether to override the presidential veto of the resolution, Representative Broomfield spoke in favor of an override saying: "This historic legislation will, if enacted into law, reestablish once and for all the traditional warmaking responsibilities which the framers of the Constitution assigned to Congress 200 years ago."53 Notwithstanding its invocation of the Constitution, Congress seems generally undermotivated to exercise its constitutional and legal prerogatives, possibly because of the accountability considerations mentioned above.54 If the war works out, the president will get much of the credit; if it fails, Congress does not want to share the blame and so avoids getting involved in the decision. But the baseline assignment of powers does give Congress a resource to be used in bargaining with the executive, should the political incentives be present. We need not assert that the specific constitutional rules are followed to accept the possibility that the assignment of formal powers does have implications for domestic bargaining among institutions. Regardless of whether the formal language of the constitution is followed, we believe that political institutions will tend to guard their formal assignment of decisionmaking rights over a particular category of policy. A formal provision may serve to empower the legislature in internal bargaining with the executive over war policy, though of course this will not be true for all countries at all times. (Consider the counterfactual of a constitution in 52 U.S. House 1972. 53 U.S. House 1973b. 54 See 20
which Art. 1, Sec. 8 did not provide Congress any war powers. The War Powers Act might not have been enacted at all.) Another possibility is that the presence of a constitutional provision asserting the power of war may serve as a proxy for other relevant legislative powers. So constitutions that assign war power to the legislature are systems in which the legislature is generally strong vis-а-vis the executive, and can rely on collateral powers to constrain it. Indeed, an assignment of legislative involvement in war power is correlated with a general index of de jure legislative power in constitutions at a level of .73.55 The long debate over legislative control of war suggests an empirical implication that has not been directly tested with fine-grained constitutional data. If proponents of legislative involvement are correct that legislatures can help with selection or screening, states that have more constraints on warmaking should do better at war whether they are democratic or autocratic. Ceteris paribus, we expect them to select fewer wars. But wars that they enter into should be won with a higher probability. III. Data Sou`rces and Methods A. Democracy & Constitutional Data 55 This index is developed in Elkins et al. 2009 and responds to Fish and Kroenig 2009. The correlation with the Fish and Kroenig index of de facto legislative power is .56. 21
How might we test these conjectures? The literature has tended to use democracy as a proxy for domestic constraints.56 Alternatively, it uses general indicators of political constraints such as POLCON, which is an indicator of the number of veto points.57 Executive constraints are defined generally according to the relative authority of a country's chief executive and the legislature, without regard to specific powers such as the ability to declare or fund a war. But as is well known, there is great internal variety within each of the categories of democracy and autocracy, including variation on internal constitutional arrangements.58 This variation can provide analytic leverage on the democratic peace debate, and the particular sources of democratic advantage. To take one example of the state of the literature, Schultz's paper relies on Polity III, which is a relatively crude measure of internal democratic constraints. So even if Schultz has shown that democratic institutions "inform," we don't really know which democratic institutions generate the bargaining advantage. Using new data from the Comparative Constitutions Project, which measures features of formal constitutions for independent nation-states since 1789, this paper begins to help identify the precise institutions which serve to constrain. In particular we focus on the extent to which internal involvement of legislatures in war-making have an effect on war selection and performance, as the informational theory would predict. To evaluate these conjectures, we use MID dataset v3.10, which includes all Militarized Interstate Disputes (MIDs) from 1816 to 2001. The dataset authors define MIDs as "cases in 56 See e.g. Layne 1994, 13 (structural argument "focuses on an independent variable--decisional constraints embedded in a state's domestic political structure--that is associated with, but not exclusive to, democracies.") 57 Henisz 2006. 58 See the studies in Ginsburg and Simpser 2013; and Versteeg and Galligan 2013. 22
which the threat, display, or use of military force short of war by one state is explicitly directed towards the government official representatives, official forces, property, or territory of another state."59 This excludes all conflicts between states and non-state actors, as well as disputes that do not include military elements. But it also includes conflict short of war, typically defined as a conflict involving more than 1000 battle deaths.60 We focus on conflict levels 5 and 6, the use of force and war respectively. To measure democracy, we have a number of potential indicators. In evaluating states as democracies or autocracies, many scholars rely on the Polity III dataset, which furnishes a combined index of a nation's regime type as well as specific indicators on "[the] competitiveness of political participation, [the] openness of executive recruitment, and [the] extent of constraints on the chief executive."61 POLITY has the advantage of longitudinal coverage back to the early 19th century, and we use it in our main regressions. Another standard measure with good temporal coverage is from Przeworski, Alvarez, Cheibub, and Limongi (PACL), who develop a binary indicator of countries as democratic or autocratic.62 For our primary independent variables of interest, we also draw on the Comparative Constitutions Project, a dataset coding formal provisions of national constitutions for all independent nation-states since 1789, a total of 893 unique constitutional systems.63 In 59 Jones, Bremer, and Singer 1996, 168; and Ghosn and Palmer 2003. 60 Small and Singer 1982: 205-06. 61 Mansfield and Snyder 2005; and Mansfield and Snyder 2002; and Narang and Nelson 2009. 62 Przeworski et al 2000; An alternative available only for post 1945 observations is the Unified Democracy Scores (UDS), a composite measure that is more reliable than its component metrics. Pernstein, Meserve, and Melton 2010. 63 See description at www.comparativeconstitutionsproject.org. 23
particular we examine the assignment of powers to initiate and approve war. We construct several variables of interest: WARLEG, which assesses whether either house of the legislature is involved in declaring war (roughly 24% of constitutions); WAREXEC, which asks whether the government or head of state declares war (50% of constitutions); and WARNONE, for constitutions which are silent on the war power (29% of constitutions).64 Legislatures can also be involved in approving declarations of war that are made by an executive, or decisions of the commander in chief. WARAPLEG captures whether the legislature is involved in approving war (some 38% of constitutions); and COMAPLEG, which asks whether the legislature must approve decisions of the commander in chief (some 12% of constitutions).65 In most of the regressions that follow, we use a compositve variable WARANYLEG which consolidates all the legislative variables into a single binary variable. A word is in order about the distinction between approving actions of the commander in chief and involvement in declaring war. In the era before the United Nations Charter, a formal declaration of war was an act of legal significance, indicating the view that the rules governing war were in place.66 The United Nations Charter, however, speaks of "armed conflict" and 64The remainder of constitutions assign the power to an "other" category or stipulate that it will be determined by law. Roughly 5% of constitutions seem to assign the power of declaring war to both the legislature and executive. An example is the Constitution of the Peoples Republic of China (1954), which gives the Chairman the power to "proclaim a state of war" (Art. 40) at the same time that it gives the People's Congress the Power to "decide on matters of war and peace." (Art. 27) 65 Examples include the Constitution of the Maldives, Art. 243(b) (2008) ("If the President, as Commander in Chief, authorizes or orders the employment of the Military Service in defence of the Republic or as part of an international undertaking, the President shall without delay submit the authorization to the People's Majlis. The People's Majlis may at any time approve the authorization, or revoke the authorization.") 66 The Hague Convetion (III) of 1907 explicitly required declarations of war. See Art. I ("The contracting Powers recognize that hostilities between themselves must not commence without previous and explicit 24
prohibits the "threat or use of force."67 This is a functional definition, indicating that the formal law of war is triggered not by a special state of affairs between the nations concerned, but by the presence of hostilities. Modern constitutions to some degree reflect this shift, by moving away from language about formal declarations of war, toward legislative approval of the deployment of troops or decisions of the commander in chief in general. From the perspective of crisis bargaining, this involves the legislature earlier in the temporal sequence, as it includes the mere deployment of troops and not only active hostilities (see row 3 in Figure 1).68 Some of these constitutional attributes are found in similar frequencies whether in democracy or autocracy, while other vary. Table 2 below shows the distribution of variables in constitutional systems by whether they are democratic. Democracies seem to have a greater propensity to involve the legislature in a supervisory role over decisions of the commander in chief. But legislative approval or declaration of war is equally likely in both regime types. Interestingly, democracies are more likely to be silent on war powers, something we will return to later. And, perhaps unsurprisingly, executives are likely to have the dominant role in autocracies. TABLE 2 HERE Within democracies, there are some differences by type of regime, as presidential, semipresidential or parliamentary. Table 2 shows that parliamentary democracies are more likely to be silent on the issue of war, and are also less likely to assign the power to the legislature. Semi- warning, in the form either of a declaration of war, giving reasons, or of an ultimatum with conditional declaration of war." 67 U.N. Charter, art. 2(4). 68 For a helpful analysis of the threat to use force in the context of U.S. law, see Waxman Forthcoming. 25
presidential systems are most likely to assign the power to the legislature and to stipulate legislative approval. But regime type is hardly a perfect predictor of the assignment of these constitutional powers. Using the Cheibub et al. coding of regime types, we see that the assignment of war powers in presidential democracies is fairly similar to that in civilian and military dictatorships.69 A more refined analysis of the specific constitutional attributes is required, rather than simply relying on the classic presidential-parliamentary distinction to understand the effect of institutions.70 B. Dyadic Analysis: Use of Force & War To determine whether these constitutional variables affect war, we estimate a series of logit models using cross-sectional time series data. We adopt many of the conventions of the democratic peace literature in conducting the analysis. First, we adopt a dyadic framework, in which the unit of analysis is the country-dyad-year. This is the standard approach, and makes sense given our emphasis on the communicative effects of legislative involvement. The MID dataset identifies, for any conflict, an initiator ("state 1") and a target state ("state 2"). Because many pairs of states in the international system will not face a realistic possibility of entering into a conflict with each other, we restrict the analysis to "politically relevant dyads."71 These are defined as dyads that are either contiguous or involve a great power, which might realistically encounter the state. This restricts the sample, but reasonably so. Including the Iceland- 69 Cheibub 2009. 70 Cheibub, Elkins and Ginsburg 2013. 71 Lemke and Reed 2001. 26
Zimbabwe dyad might introduce noise and reduce the accuracy of the estimates. A final sample restriction is that, because we are interested in the onset of wars, we do not examine dyad-years in which the conflict is ongoing. The dependent variable in this initial analysis is drawn from the levels of hostility indicator from the MID database.72 This indicator has six levels ranging from "No Hostility" to "War". We focus on the two highest levels of conflict, "War", and "Use of Force", which is a lower level of conflict that falls below the conventional number of fatalities defining war. A "War" or "Use of Force" for our purposes occurs when both states in the dyad have at least that level of hostility vis-а-vis each other. (All Wars count as Use of Force in this analysis.) The set of control variables potentially useful in a study such as this is large. Mansfield and Snyder, for example, include whether or not the country in question was a major power, whether the country in question was engaged in a civil war in each year studied, and the distribution of power between the country in question and its neighbors.73 Other common variables used in the literature are the number of allies possessed by a country in a given year, the number of years of peace, and measures of absolute power. In the regressions that follow, we use a standard set of variables that are arguably at the core of the democratic peace model.74 First we assess the relative power between the two states in the dyad. This measure draws on the Correlates of War measure of capabilities, and captures 72 Ghosn et al. 2004. We use the cwhost variable in the MID dataset. 73 Mansfield and Snyder 1996, 168. 74 We are grateful to Robert Schub for the do-files to allow us to run the model. 27
the share of total capability held by the first state in the dyad.75 We also ask whether the dyads are jointly democratic; the prediction of the democratic peace literature is that such dyads will not enter into conflict and so it is important to control for such dyads if we are to isolate the effect of constitutional variables.76 We also ask if the countries in the dyad are allies, with the obvious prediction that this will reduce the probability of conflict.77 Finally, we control for contiguity. States that are next to each other are more likely to enter into conflict, possibly because of the depth and frequency of their contacts, and also because territorial disputes are a major source of international conflict.78 Each regression takes the form of CONFLICTijt=0 + 1CONSTijt + 2RELCAPijt + 3JDEMijt+4ALLYijt+ 5CONTijt + e where CONFLICT captures whether country i and country j, in year t, experience a militarized interstate dispute [FORCE | WAR] between them; CONST represents an attribute of the formal constitution in force in country i and/or j; and RELCAP, JDEM, ALLY, and CONT represents the set of dyad-level control variables described above. In keeping with the literature, we cluster the standard errors at the level of the dyad. The constitutional variables tested are described in the previous section. We test sequentially whether they are found in state 1 and state 2, respectively, and denote this with the 75 Specifically cap_1/(cap_1+cap_2) 76 We rely on the POLITY II coding to define democracy as any state that scores at least a 5 in a particular year. 77 This is drawn from the Correlates of War Alliance variable in which alliance scores of less than 4 are counted. They include formal military alliances, as well as lower levels of affiliation. 78 Huth 2000. 28
suffix 1 or 2 at the end of the variables. For the regressions reported in Tables 3-6, we use WARANYLEG, which captures whether a legislature is involved in either declaring or approving war, or supervising the actions of the commander-in-chief. Tables 3 and 4 in the Appendix present the results. The basic results are consistent with the democratic peace literature. Relative capabilities matter, so that as the relative power of state 1 increases, so does the use of force or war. Democratic dyads are less likely to enter into conflict in every specification, as are allied dyads. Contiguous states are more likely to experience conflicts. Controlling for these standard factors, constitutional variables have some effect. Whether the dependent variable is the use of force or war, legislative approval of the actions of the commander in chief makes a state less likely to initiate conflict. For war, legislative declaration or approval also reduces the probability that a state will initiate conflict, and legislative involvement in the target state reduces the probability of escalation to either use of force or war. Finally, constitutional silence in either state also matters, reducing the probabilities of conflict. To investigate the effect of regime type, we take two different approaches. First, we conduct the regressions for subsamples depending on whether state 1 has a POLITY score of greater than 7. This allows us to see whether democracies or dictatorships are driving the overall effect.79 Results are in Table 5. This shows that the restraining effect of legislatures is not limited to democracies by any means; indeed the strongest effects are in dictatorships with regard to war. One way to interpret these results are that authoritarian legislatures do not prevent their leaders from entering into interstate conflicts, but that they do deter them from escalating to the 79 We also experimented with different cutoff points in the POLITY data. 29
point of war. The effects in democracies are a bit weaker, but seem to occur earlier in the sequence, limiting leaders from using force at all. Second, we examine the basic regression for use of force and war for dyadic subsets of democracies vs. dictatorships, dictatorships vs. democracies, and dictatorships vs. dictatorships. We leave out the fourth conceptual possibility, democracies vs. democracies, in light of the basic result of the democratic peace literature. Tables 6a and 6b examine the effect of legislative involvement, in state 1 and state 2 respectively, for each pair of regime types. The results are surely intriguing. Legislative war power in the initiator state matters only in dictatorships that are paired with democracies; legislative war power in the target state matters when the target is a democracy responding to a dictatorship, or a dictatorship responding to a democracy. In all cases, it reduces the probability of conflict. In dyads composed of two dictatorships, legislatures do not matter. We speculate that the signaling dynamics are different in a conflict limited to dictatorships. C. Effect on Outcomes To evaluate the effect of constitutional war powers on outcomes, we examine 3647 disputes for which an outcome is coded by the COW database. These outcomes range from a win by Side A to a win by side B, with intermediate outcomes indicating yielding by one side or the other, or a stalemate. (There are several other categories of outcome, with relatively small numbers of data points.) For purposes of this study, we aggregate the categories of outright win and yield by the counterparty to create categories Win_By_A and Win_By_B. Most disputes, however, do not have either outcome: over half of disputes end in stalemate. 30
For each of these outcome variables, we run a series of multinomial logit regressions with constitutional variables on the right hand side, controlling only for relative power. Many of the other variables used to predict dispute initiation are not likely to matter as much for outcomes, and so we use this simpler framework. For example, while contiguity predicts conflict, it will have no effect on victory, since one or the other state will lose. We examine sequentially the effect of legislative involvement in war in state A and state B on probability of "win by A"; then do the same for "Win by B". The results presented in Table 7 are, on first glance, in some tension with the result of the democratic peace literature that democracies tend to win those conflicts they enter into. Our results show that legislative involvement is associated with loss and stalemate, not victory, once a dispute actually escalates to open war. If the legislature is involved in state A, state A is less likely to win (Model 1). But state A is also less likely to win if the legislature is involved in the target state B! (Model 2). And state B is less likely to win if the legislature is involved in state A (Model 3). --Table 7 here-- To understand this result, recall again that most outcomes are not wins. Running the same regressions with stalemate as the dependent variable shows that legislative involvement on either side increases the likelihood of a draw (Models 5 & 6). This is consistent with the informational account of legislative involvement. If legislatures are screening wars, then legislative involvement in either side should increase the probability of settlement of all but the closest cases. Only when there is a genuine ambiguity over the likely outcome will conflict escalate. Like the selection of disputes for litigation, the selection of conflicts will tend to lead to only close cases being fought, particularly when there are better institutions for information 31
production.80 Legislative involvement, and particularly joint legislative involvement on both sides, will help force settlements and, perhaps perversely, increase the possibility of stalemate. But why would legislative involvement be positively correlated with loss? One possibility is that legislatures are equally subject to the "gambling for resurrection" dynamic that has been identified for democratically elected executives.81 That is, once a legislature does commit to supporting military adventures, it will have difficulty backing down when the conflict goes badly. To admit defeat is to risk electoral loss, for both the executive and legislature. While this concern might drive the legislature to screen bad conflicts in the first place, once a conflict achieves a certain level of seriousness, legislative involvement can actually prevent a helpful settlement from materializing. This is consistent not only with the idea that legislatures can produce information about the likely outcomes of conflict, but that they independently contribute to political resolve and commitment. For some disputes, this may lead the other side to capitulate well before escalation to war; but if escalation occurs, the commitment is difficult to undo. This is precisely the source of credibility in the pre-escalation bargaining process. The result helps us sort out the relative importance of screening and signaling. If legislatures only screened conflicts, we would expect fewer conflicts to escalate to war, but a higher probability of victory in those wars that do materialize. This is because the legislature would prevent the country from getting into losing battles. If the source of legislative advantage, on the other hand, was primarily signaling, we would expect that there would be fewer wars, but perhaps a lower probability of victory therein, since signaling resolve would prevent giving 80 Priest and Klein 1984. 81 Downs and Rocke 1995 32
ground down the road. While we cannot say that only signalling is at work, our results provide support for the idea that legislatures do more than screen. D. Endogeneity and Robustness [IN PROGRESS] These results do not deal with selection issues. We cannot discount the possibility that, for example, nations that are more inherently peaceful or more secure choose to involve the legislature in decision-making over war. Indeed, as a matter of optimal constitutional design, such nations ought to be less concerned with the potential costs associated with delay in crisis decision-making (since crisis is less likely), and so might choose to constrain the executive more thoroughly. In other words, the risk of executive aggrandizement outweighs the risk of delayed decision-making. Of course, one can construct the opposite account: nations which are more warlike might be more likely to want to restrain an executive. To test for these possibility, we ran a series of instrumental variable regressions with several plausible candidates. We generally observe that there is a good deal of serial continuity in constitutional design; countries tend to stick with institutions once adopted.82 We thus experimented with the using the values in the first constitution in a country's history. These would plausibly be associated with the formal provisions in a later constitution, but arguably not related to the propensity to enter into conflict many years after independence. However, in this particular case, there seems to be less serial continuity than generally found. More than two dozen nations have changed their constitutional rule on war over time; some of them have moved toward more legislative involvement while some have moved toward less. An initial scan does not present a pattern in which these switches correlate with conflict. For example, although 82 See Elkins et al 2009. 33
France moved toward more legislative involvement in the Fourth Republic right after World War II, implying a desire for more constraint; on the other hand Bolivia moved to less legislative involvement in 1826 right after its war for independence, switched back after a war ended in 1839, then switched three more times during the forty-year period between of peace the War of the Confederation and the War of the Pacific. These patterns suggest that reverse causation may not be a concern, but for a more rigorous test, the search for a good instrument is ongoing. We also ran the analysis on subsamples of different regime sub-types, using the classification from Cheibub et al. (2006) for the initiator state in the dyad. Coverage for the regime type variable is smaller than for the full dataset, as it is only available for 490 constitutional systems, significantly reducing the dyadic sample. The results for entry into conflict are strongest in semi-presidential democracies, and also significant in military dictatorships. They are just shy of statistical significance in parliamentary democracies (85% confidence level) with the predicted sign. They are not significant in authoritarian monarchies, presidential democracies or civilian dictatorships. The results for outcome show that legislatures lead to greater likelihood of loss, and are statistically significant in monarchies, military dictatorships, and parliamentary democracies. In terms of time periods, the effects on both entry into conflict and outcomes are strongest in the post-1945 period, and also significant in the period 1900-1945. There are, of course, more dyads as time goes on, a function of the expansion of the number of countries in the world. This may explain why the model does not hold in the earlier period, though further examination is needed. 34
In unreported robustness checks, we introduce controls for the number of years of peace between the countries in the dyad, along with squared and cubed terms. These are included to control for past conflicts among states, and are fairly standard in the literature on war. The peace-years polynomials did not affect the results and so we leave these out of the main reported specifications. E. Further Discussion: Authoritarian Legislatures and the Virtues of Silence One nice advantage of formal constitutional rules is that they are found in both democracies and autocracies. While the traditional view has been to characterize authoritarian constitutions as shams, a new line of research is beginning to examine how they function and may serve to solve internal problems of credible commitments, agency control and coordination within authoritarian regimes.83 Authoritarian legislatures function differently than do democratic ones; their primary function has been theorized to provide private information to the authoritarian center.84 They are unlikely, however, to provide much signaling value, since their public statements are likely to align with the leadership. Any effect on reducing the prevalence of war is more likely to be driven by screening, and perhaps the receiving of signals, than the sending of them. This paper has shown that the effects of legislative involvement in warmaking depend on the dyadic configuration, and seem to matter most when authoritarians confront democracies. Perhaps authoritarian legislatures are able to restrain the risk-taking leadership 83 See e.g. Gandhi 2008; Schedler 2013; Svolik 2013. 84 Maleski 2013. 35
behind closed doors, even as they engage in public bluster.85 Intuitively, we should expect authoritarian legislators to have an incentive to generate useful information; after all, if the regime falls, the legislators will as well. Constitutions that are totally silent on the war process are associated with less conflict. This is an interesting finding, and goes to a core question in constitutional debates: what is the virtue of trying to specify written constraints on political power? The founding fathers in the United States believed that words could constrain, and that explicit allocations of power would serve to restrain its exercise. This assumption rested largely on the Madisonian scheme of selfaggrandizing political branches that would serve to check each other. From this perspective, it is obvious that one needs to use words to "chain the dogs of war." Our finding on silence seems to be driven by the fact that parliamentary systems, in both dicatorship and democracy, are less likely to articulate rules about the use of force. The basic result for the effects of silence is strongest in both democratic parliamentary regimes and civilian dictatorships, which tend to be based in a single party. While there is no overall "parliamentary peace", silence seems to both frequent and constraining in parliamentary systems, hardly a surprise given the co-dependence of government and legislature. In an unreported analysis, we decompose legislative involvement into its subtypes. The results suggest that a requirement of legislative declaration of war seems to have little substantive effect. Legislative approval of declarations made by the executive, on the other hand, is associated with less conflict, as are requirements for legislative approval of actions of 85 A counter-example is found in the experience of the Kosovo war, when Milosevic feared a continuation of the war if the parliament rejected the peace plan. (Sell, Slobodan Milosevice and the Destruction of Yugoslavia, 310 (Duke 2002)). 36
the commander in chief. This finding might spark an interesting normative debate on optimal design--it may be that there is virtue in having the legislature as the second actor in the process rather than the first, though in practice the executive will always play a role of initiating. It may also be the case that legislatures in the position of declaring war are more likely to be found in systems in which they are the only actor involved. This would be consistent with the idea that legislatures affect war through adding information by screening, without adding to the clarity of signals. E. Conclusion The democratic peace literature has come a long way since Immanuel Kant. His initial musings on the unwillingness of popularly elected governments to commit their citizens to war have been refined through a generation of sophisticated theorizing and empirical analysis. Yet certain questions remain unanswered. We do not have a precise sense of which institutions in democracies tend to lead to the statistical regularities: what are the mechanisms of information transmission and bargaining advantage? Constitutional rules about warmaking may be part of the story. Our central finding is that a legislative role in decisions about war tends to lead to fewer conflicts. We speculate that this is because of a combination of screening and signaling. But surprisingly, once a conflict escalates to war, legislative involvement reduces the probability of winning, and is instead associated with stalemate and loss. Legislatures, it seems, can indeed chain the dog of war; but they seem to spur the dog on once loosed. 37
References Bueno de Mesquita, Bruce, and David Lalman. 1992. War and Reason: Domestic and International Imperatives. New Haven, Conn.: Yale University Press. Cheibub, Jose Antonio, Zachary Elkins and Tom Ginsburg. 2013. Beyond Presidentialism and Parliamentarism. British Journal of Political Science. Cheibub, Jose Antonio, Jennifer Gandhi and James Vreeland. 2009. Democracy and Dictatorship Revisited Dataset v 1.0. Desch, Michael. 2002. Democracy and Victory: Why Regime Type Hardly Matters. International Security 27 (2):5-47. Dixon, William J. 1994. Democracy and the Peaceful Settlement of International Conflict. American Political Science Review 88 (1):14-32. Downs, George W., and David M. Rocke. 1995. Optimal Imperfection: Domestic Uncertainty and Institutions in International Relations. Princeton: Princeton University Press. Doyle, Michael. 1983. Kant, Liberal Legacies, and foreign affairs. Philosophy and Public Affairs 12(3): 205-235. Elkins, Zachary, Tom Ginsburg, and James Melton. 2009. The Endurance of National Constitutions. New York: Cambridge University Press. Ely, John Hart. 1993. War and Responsibility: Constitutional Lessons of Vietnam and Its Aftermath. Princeton: Princeton University Press. Fearon, James D. 1995. Rationalist Explanations for War. 49 International Organization 49 (3):379-414. Fish, Steven and Mathew Kroeing. 2009. The Handbook of National Legislatures: A Global Survey. New York: Cambridge University Press. Fisher, Louis, Ryan Hendrickson, and Stephen R. Weissman. 2008. Letter to the Editor: Congress at War. Foreign Affairs 87 (3):167-9. Gandhi, Jennifer. 2008. Political Institutions Under Dictatorship. New York: Cambridge University Press. Gartzke. 1999. War is in the error term. International Organization 53 (summer): 567-587 Gartzke, Erik. 2007. The Capitalist Peace. The American Journal of Political Science 51 (1):166191. 38
Gaubatz, Kurt Taylor. 1999. Elections and War: The Electoral Incentive in the Democratic Politics of War and Peace. Stanford: Stanford University Press. Ghosn, Fatten, and Glenn Palmer. 2003. Codebook for the Militarized Interstate Dispute Data, Version 3.1. Correlates of War 2 Project. Available from . Last Modified 14 April 2003. Ginsburg, Tom, and Alberto Simpser, eds. 2013. Constitutions in Authoritarian Regimes. New York: Cambridge University Press. Ginsburg, Tom, and Paul Diehl. 2006. Irrational War and Constitutional Design: A Response to Professors Nzelibe and Yoo. Michigan Journal of International Law 27:1239-59. Hallett, Brien. 2012. Declaring War: Congress, the President and What the Constitution Does not Say. New York: Cambridge University Press. Henisz, Witold J. 2006. Political Constraint Index Dataset. Available at < http://wwwmanagement.wharton.upenn.edu/henisz/>. Accessed DAY MONTH YEAR. Hemel, Daniel. 2007. Picking up the Peaces: Democratic Norms, Structural Constraints, and International Security, B.A. Thesis, Harvard College (on file with author). Hess, Stephen. 1981. The Washington Reporters. Washington, D.C.: Brookings Institution Press. Howell, William G., and Jon C. Pevehouse. 2008. While Dangers Gather: Congressional Checks on Presidential War Powers. Princeton, N.J.: Princeton University Press. Huth, Paul K. 2000. Territory: Why Are Territorial Disputes between States a Central Cause of International Conflict. In What Do We Know About War?, edited by John A. Vasquez, 85110. Lanham, Md.: Rowman & Littlefield. Huth, Paul K., and Todd L. Allee. 2002. The Democratic Peace and Territorial Conflict in the Twentieth Century. Cambridge, UK: Cambridge University Press. Isaacharoff, Sam. 2009. Political Safeguards in Democracies at War. Oxford Journal of Legal Studies 29 (2):189-214. Jay, John. 1987a. No. 3: The Same Subject Continues: Concerning Dangers from Foreign Force and Influence for the Independent Journal. The Federalist Papers. ------. 1987b. No. 4: The Same Subject Continued: Concerning Dangers from Foreign Force and Influence. The Federalist Papers. 39
Jefferson, Thomas. 1789. Thomas Jefferson to James Madison, 6 September 1789. In vol. 15 of The Papers of Thomas Jefferson, edited by Julian P. Boyd. Princeton: Princeton University Press, 1958. Jones, Daniel M., Stuart A. Bremer, and J. David Singer. 1996. Militarized Interstate Disputes, 1816-1992: Rationale, Coding Rules, And Empirical Patterns. conflict management and Peace Science 15 (2):163-212. King, Gary and Langche Zeng. 2001. Logistic Regression for Rare Events Data. Political Analysis 9:137-63. Layne, Christopher. 1994. Kant or Cant: The Myth of the Democratic Peace. International Security 19 (2):5-49. Lemke, Douglas, and William Reed. 2001. The Relevance of Politically Relevant Dyads. Journal of Conflict Resolution 45 (1):126-44. Leng, Russell J. 2000. Escalation: Crisis Behavior and War. In What Do We Know About War?, edited by John A. Vasquez, 235-259. Lanham, Md.: Rowman & Littlefield. Levy, Gilat, and Ronny Razin. 2004. It Takes Two: An Explanation for the Democratic Peace. Journal of the European Economic Association 2 (1):1-29. Lindley, Dan and Ryan Schildkraut. ___. Is War Rational? The Extent of Miscalculation and Misperception as Causes of War. Manuscript, University of Notre Dame. Lipson, Charles. 2003. Reliable Partners: How Democracies Have Made a Separate Peace. Princeton, N.J.: Princeton University Press. Mansfield, Edward D., and Jack Snyder. 2002. Democratic Transitions, Institutional Strength, and War. International Organization 56 (2):297-337. ------. 2005. Electing to Fight: Why Emerging Democracies Go to War. Cambridge, Mass.: Belfer Center for Science and International Affairs. Maoz, Zeev, and Bruce Russett. 1993. Normative and Structural Causes of Democratic Peace, 1946-1986. The American Political Science Review 87 (3):624-38. McDonald, Patrick. 2009. The Invisible Hand of Peace: Capitalism, the War Machine, and International Relations Theory. New York: Cambridge University Press. Narang, Vipin, and Rebecca M. Nelson. 2009. Who Are These Belligerent Democratizers? Reassessing the Impact of Democratization on War. International Organization 63 (2):35779. 40
Nzelibe, Jide, and John Yoo. 2006. Rational War and Constitutional Design. The Yale Law Journal 115 (9):2512. Oren, Michael B. 2003. Six Days of War: June 1967 and the Making of the Modern Middle East. New York: Presidio Press. Pernstein, Daniel, Steven Meserve, and James Melton. 2010. Democratic Compromise: A Latent Variable Analysis of Ten Measures of Regime Type. Political Analysis 18 (4):426-449. Posner, Eric, and Adrian Vermuele. 2011. The Executive Unbound: After the Madisonian Republic. New York: Oxford University Press. Posner, Eric, and Alan Sykes. 2013. Economic Foundations of International Law. Cambridge, Mass.: Harvard University Press. Priest, George, and Benjamin Klein. 1984. The Selection of Disputes for Litigation. Journal of Legal Studies 13 (1):1-55. Przeworski, Adam et al. 2000. Democracy and Development: Political Institutions and WellBeing in the World, 1950­1990. Cambridge, UK: Cambridge University Press. Quackenbush, Stephen L., and Michael Rudy. 2006. Evaluating the Monadic Democratic Peace. (PDF). Paper presented at the Annual Meeting of the Midwest Political Science Association, April 20-23, Chicago, Ill. Ray, James Lee. 2000. On the Level(s), Does Democracy Correlate with Peace? In What Do We Know About War?, edited by John A. Vasquez, 299-316. Lanham, Md.: Rowman & Littlefield. Reiter, Dan, and Alan C. Stam. 2003. Democracies at War. Princeton, N.J.: Princeton University Press. Reiter, Dan. 2003. Exploring the Bargaining Model of War. Perspectives on Politics 1 (1):27-43. Reveley, Taylor W. 1981. War Powers of the President and Congress: Who Holds the Arrows and Olive Branch. Charlottesville: University Press of Virginia. Rudolph, Thomas J. 2003. Who's Responsible for the Economy? The Formation and Consequences of Responsibility Attributions. American Journal of Political Science 47 (4):698-713. Schedler, Andreas. 2013. The Politics of Uncertainty: Sustaining and Subverting Electoral Authoritarianism. Oxford, UK: Oxford University Press. 41
Schultz, Kenneth A. Do Democratic Institutions Constrain or Inform? Contrasting Two Institutional Perspectives on Democracy and War. International Organization 53 (2):233266. Schultz, Kenneth A. 2001. Democracy and Coercive Diplomacy. Cambridge, UK: Cambridge University Press. Scoville, Ryan. 2013. Legislative Diplomacy. Michigan Law Review 112:331-95. Small, Melvin and J. David Singer. 1982. Resort to Arms: International and Civil War,1816­ 1980. Beverly Hills, CA: Sage. Stimson, James A., Michael B. MacKuen, and Robert S. Erikson. 1995. Dynamic Representation. The American Political Science Review 89 (3):543-565. Svolik, Milan. 2013. The Politics of Authoritarian Rule. New York: Cambridge University Press. U.S. Congress. 1973. Joint Resolution Concerning the War Powers of Congress and the President: Pub. L. 93-148, H.J. Res. 542. 93d Cong., 1st sess., 7 November. U.S. House. 1969? H. Con. Res. 91-199. 91st Cong., 2d sess., Day Month. ------. 1972. Committee of Foreign Affairs. House Report No. 92-1302 Concerning the War Powers of Congress and the President. 92d Cong. 2d sess., 3 August. ------. 1973a. War Powers Resolution of 1973: Report No. 93-287 prepared for the House Committee on Foreign Affairs. 93d Cong., 1st sess., 15 June. ------. 1973b. H9642. 93d Cong. 1st sess., 7 November. ------. 1999. Emergency Supplemental Appropriations Act: House Resolution No. 1141, Pub. L. 106-31. 106th Cong., 1st sess., 21 May. U.S. Senate. 1972. Committee on Foreign Relations. War Powers: Report Together with Additional Views of Senator Fulbright and Individual Views of Senator John Sherman Cooper (to Accompany S. 2956): Report No. 92-606. 92d Cong., 2d sess., 9 February. ------. 1973. Committee on Foreign Relations. War Powers: Report to Accompany S. 440: Report No. 93-220. 93d Cong., 1st sess., 14 June. Versteeg, Mila, and Dennis Galligan, eds. 2013. Social and Political Foundations of Constitutions. New York: Cambridge University Press. Waxman, Matthew C. Forthcoming. The Power to Threaten Force. The Yale Law Journal. 42
Wormuth, Francis D., and Edwin B. Firmage. 1989. To Chain the Dog of War: The War Powers of Congress in History and Law. Champaign: University of Illinois Press. Yoo, John. 2006. The Powers of War and Peace: The Constitution and Foreign Affairs after 9/11. Chicago: University of Chicago Press. 43
TABLES & FIGURES Figure 1: Sequence of Play Classic War Model: Unitary States State A intitiatesState B responds[settlement|escalation to MID][settlement|war] "Chain the Dog of War" Model Legislature State A intitiatesState B responds[settlement|escalation to MID][settlement|war] Modern Constitutional Approach Legislature State A initiatesState B responds[settlement|escalation to MID][settlement|war] 44
Table 1: Distribution of Legislative Powers over War
N==745 national constitutions since 1789
Declare War
Approve War
Head of state/government
17 (2%)
3 (0%)
First or only legislative house 119 (15%)
143 (19%)
Upper house
11 (1%)
7 (1%)
Either house
4 (0.5%)
Both houses jointly
62 (8%)
133 (18%)
23 (3%)
[no approval needed]
Not specified
219 (29%)
387 (52%)
Table 2: Distribution of variables by regime type
Universe: 490 constitutional systems for which Cheibub et al (2006) variables available.
Variable (n)
Leg declares war
Leg approves war
Leg approves actions of commander in chief
Executive declares war
Const. silent on war
Democracy (56) .14
Democracy (26) .58
Democracy (35) .17
Dictatorship (187) .21
Dictatorship (158) .19
Monarchy (28)
Table 3: Constitutional Variables and Use of Force
(1) force
(2) force
(3) force
(4) force
Relative Capability of state1 Both democratic Allies Contiguous Leg involvement in war state 1 Leg involvement in war state 2 Exec declare war state 1 Exec declare war state 2 Silence on war state 1 Silence state 2
0.414** (0.162) -0.986*** (0.244) -1.108*** (0.148) 2.296*** (0.176) -0.235* (0.123)
0.403** (0.162) -0.963*** (0.249) -1.102*** (0.149) 2.293*** (0.176) -0.354** (0.140)
0.419** (0.167) -0.998*** (0.242) -1.097*** (0.155) 2.311*** (0.185) 0.146 (0.123)
0.356** (0.166) -1.018*** (0.242) -1.129*** (0.152) 2.283*** (0.176) 0.0609 (0.129)
-6.800*** -6.749*** -6.968*** -6.855***
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
(5) force 0.417** (0.168) -0.963*** (0.243) -1.030*** (0.155) 2.299*** (0.182) -0.562*** (0.178) -6.800*** (0.192) 197,048
(6) force 0.369** (0.169) -0.988*** (0.242) -1.058*** (0.154) 2.272*** (0.178) -0.625*** (0.219) -6.725*** (0.188) 197,059
Table 4: Constitution Variables and War
Relative capability state 1 Both democratic Allies Contiguous Leg involvement in war state 1 Leg involvement in war state 2 Exec declare war state 1 Exec declare war state 2 Silence on war state 1 Silence state 2
0.654** (0.279) -1.824*** (0.701) -1.328*** (0.289) 2.098*** (0.309) -0.578** (0.249)
0.629** (0.280) -1.766** (0.700) -1.319*** (0.291) 2.089*** (0.310) -0.877*** (0.279)
0.671** (0.283) -1.891*** (0.694) -1.308*** (0.295) 2.052*** (0.317) 0.0977 (0.229)
0.593** (0.295) -1.861*** (0.696) -1.423*** (0.305) 2.046*** (0.312) 0.0304 (0.258)
-8.492*** -8.392*** -8.708*** -8.651***
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
(5) war 0.666** (0.285) -1.847*** (0.699) -1.228*** (0.293) 2.032*** (0.311) -0.634** (0.275) -8.547*** (0.335) 197,048
(6) war 0.626** (0.302) -1.817*** (0.691) -1.311*** (0.305) 2.031*** (0.316) -1.198** (0.529) -8.485*** (0.345) 197,059
Table 5 Subsamples of democracies and dictatorships
Relative capability Allies Contiguous Legislative involvement state A Legislative involvement state B Joint democratic
0.522* (0.311) -1.721*** (0.358) 2.642*** (0.465) -0.636** (0.301) -0.696** (0.325)
-8.832*** (0.551)
0.309* (0.175) -1.241*** (0.165) 2.435*** (0.231) -0.212 (0.137) -0.204 (0.162) -6.850*** (0.253)
0.489 (0.415) -0.169 (0.525) 1.049** (0.473) -0.446 (0.406) -0.475 (0.414) -1.989*** (0.709) -7.937*** (0.453)
0.605** (0.304) -0.719*** (0.268) 2.054*** (0.253) -0.287 (0.247) -0.821*** (0.227) -1.016*** (0.261) -6.780*** (0.329)
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 6a: Effect of Legislative Involvement in State 1 By Dyad Type
VARIABLES Dyad type Level of Conflict
(1) Democracy v. Dictatorship force
(2) Dictatorship v. Democracy force
(3) Dictatorship v. Dictatorship force
(4) Democracy v. Dictatorship war
(5) Dictatorship v. Democracy war
(6) Dictatorship v. Dictatorship war
relative capability ally contiguous Leg involvement in war state 1
0.628* (0.363) -1.040*** (0.357) 1.953*** (0.294) -0.204 (0.263)
0.407 (0.387) -0.867* (0.479) 2.366*** (0.409) -0.416* (0.246)
0.422** (0.200) -1.162*** (0.155) 2.467*** (0.293) 0.0464 (0.152)
0.211 (0.414) -0.135 (0.623) 0.741 (0.508) -0.391 (0.424)
0.00907 (0.601) [omitted] 3.511*** (0.888) -1.168* (0.698)
0.591 (0.392) -1.478*** (0.351) 2.337*** (0.538) -0.392 (0.289)
-6.770*** -6.710*** -7.220*** -7.644*** -9.176*** -8.862***
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 6b: Effect of Legislative Involvement in State 2 By Dyad Type
Dyad type
(1) Democracy v. Dictatorship
(2) Dictatorship v. Democracy
(3) Dictatorship v. Dictatorship
(4) Democracy v. Dictatorship
(5) Dictatorship v. Democracy
(6) Dictatorship v. Dictatorship
Level of conflict force
relative capability ally contiguous Leg involvement in war state 2 Constant
0.628* (0.356) -1.010*** (0.356) 1.958*** (0.292) -0.578** (0.241) -6.679*** (0.331)
0.418 (0.397) -0.891* (0.493) 2.344*** (0.407) -0.754** (0.342) -6.522*** (0.316)
0.416** (0.197) -1.161*** (0.159) 2.469*** (0.294) 0.0349 (0.167) -7.215*** (0.347)
0.252 (0.405) -0.0882 (0.625) 0.767 (0.507) -1.226** (0.557) -7.534*** (0.398)
-0.0657 (0.628) 3.489*** (0.889) -2.871*** (1.098) -8.769*** (0.657)
0.632* (0.379) -1.485*** (0.352) 2.321*** (0.536) -0.214 (0.332) -8.916*** (0.704)
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 7: Effect on outcomes
win_by_A win_by_A win_by_B win_by_B
(5) stalemate
relative capability Leg involvement in war state 1 Leg involvement in war state 2
0.221 (0.152) -0.313*** (0.106)
0.0693 (0.153) -0.531*** (0.109)
0.0596 (0.198) -0.355** (0.140)
-0.0227 (0.199) -0.142 (0.137)
-0.0666 (0.115) 0.213*** (0.0788)
Constant Observations
-1.615*** (0.0969) 2,771
-1.457*** (0.104) 2,771
-2.222*** (0.124) 2,771
-2.265*** (0.136) 2,771
0.336*** (0.0729) 2,771
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
(6) stalemate 0.0164 (0.115) 0.248*** (0.0790) 0.279*** (0.0794) 2,771
Table 7: Effect of constitutional variables on outcomes, controlling for relative capability
Initiator win
Target win
Legislative involvement in declaration: initiator Legislative involvement in declaration: target Legislative approval of commander action: initiator Legislative approval of commander: target
Note: "win" includes yield by counterparty. *** p<0.01, ** p<0.05, * p<0.1

T Ginsburg, F de Vitoria

File: war-and-constitutional-design.pdf
Author: T Ginsburg, F de Vitoria
Published: Mon Apr 28 08:58:47 2014
Pages: 53
File size: 0.57 Mb

À propos de, 4 pages, 0.08 Mb
Copyright © 2018 doc.uments.com