Are New Democracies War-Prone?

Electing to Fight seeks to explain an alleged paradox: Democracies do not fight each other, but "the beginning stages of transitions to democracy often give rise to war rather than peace. Since the end of the Cold War, this causal connection between democratization and war has been especially striking" (p. 2). Why are democratizing states so belligerent? Mansfield and Snyder offer a parsimonious answer—unleashed nationalism: "In democratizing states, nationalism is an ideology with tremendous appeal for elites whose privileges are threatened. It can be used to convince newly empowered constituencies that the cleavage between the privileged and the masses is unimportant compared to the cleavages that divide nations, ethic groups, or races" (p. 2).

The intuition guiding this study—that shifts within states lead to conflicts between them—is powerful and familiar. Scholars of revolution have made a similar claim about how internal change within a state triggers aggressive external behavior; modernization theorists have long argued that states experiencing rapid economic growth often expand beyond their borders; and observers have noted that state collapse invites conflict, as neighbors prey upon a target of opportunity.

The twist in Electing to Fight is its argument that one particular kind of internal change—democratization—is especially likely to lead to interstate conflict. Through an impressive compilation of cross-national regressions and a handful of case studies, Mansfield and Snyder claim to show that democratization is the most dangerous form of regime change, and that countries in transition to democracy are much more dangerous than stable autocracies, stable democracies, or countries transitioning to autocracy. The policy implications of the analysis are clear: Promoting democracy can trigger war.

Before discussing the actual theory presented in the book, it is important to clarify what it does not claim. First, it concedes that countries making successful transitions to democracy rarely go to war: "[T]here is only scattered evidence that transitions culminating in a coherent democracy influence war" (p. 101). Second, it acknowledges that democratizing states rarely initiate war; the authors report that only "ten case studies . . . comprise all of the democratizing states in our data set that initiated interstate wars." Yet Table 4.1 (p. 81) reports that the book covers a time period (1816 to 1992) in which there were 90 incomplete democratic transitions in 64 countries and 50 complete transitions in 35 countries. Given this observation, one must wonder about the book's subtitle. Electing to Fight is not about "why emerging democracies go to war," but rather about why failed democratic transitions under very special circumstances lead to war some of the time.

The results would be even more meager if the analysis focused on the latest wave of democratization. In the "third wave" of democratization, from 1973 to 2004, there were 179 instances of democratization, defined as countries moving from either Not Free to Free (25 cases) or Not Free to Partly Free (154 cases), as determined by Freedom House rankings. (This set of cases includes countries that experienced democratization more than once, after regressing from democracy earlier during this period.) Between 1973 and 2004, there were only 30 instances of a democratizing state experiencing political violence—civil war or interstate war—within five years after democratization. (The data set for political violence I use here comes from the Center for Systemic Peace, http://members.aol.com/cspmgm/warlist.htm.) In other words, during the third wave conflict occurred after democratization only 16 percent of the time, and most of these conflicts were internal rather than interstate wars. During the third wave, emerging democracies have rarely gone to war.

But then Mansfield and Snyder never claim the contrary. Instead, their argument is that democratizing states are more likely to go to war than either stable democratic or authoritarian regimes (p. 29). Even this narrower claim seems counterintuitive to the casual student of twentieth-century history. In recalling the millions slaughtered in the twentieth century, we think of ruthless dictators such as Stalin, Mao, Hitler, Pol Pot, Idi Amin, and Saddam Hussein, not leaders of democratizing states such as Adolfo Suarez in Spain, Lech WaŐsa in Poland, or Nelson Mandela in South Africa. But the reader of Electing to Fight has to read more carefully to grasp that the universe of belligerent democratizing states in question in this book is much smaller than all cases of democratization.

The authors add two crucial qualifiers. First, Mansfield and Snyder are concerned not with democratizing countries in general, but rather with a subset that they call "stalled transitions" or cases of "incomplete democratization." In their view, these are cases that move away from autocracy but fail to consolidate democracy. In the numerical language of Polity III, which is the coding system used in this book, these are countries stuck between a score of 6 and -6, while regimes with scores greater than 6 are coded as coherent democracies and regimes with scores less than -6 are coded as coherent autocracies. Remember, successful cases of democratization are not war-prone.

Yet the universe of belligerent democratizers is actually smaller still, since Mansfield and Snyder tag on a second important qualifier to every statement about the relationship between democratization and war. Stalled transitions lead to war only in countries that have weak political institutions. Without the presence of this variable, there is no bivariate relationship between democratization and war.

What are weak institutions? Mansfield and Snyder give a conventional description, writing that "the chance of war arises mainly in those transitional states that lack the strong political institutions needed to make democracy work (such as an effective state, the rule of law, organized political parties that compete in fair elections, and professional news media)" (p. 2). Note that the first two institutions in this list (an effective state, the rule of law) can be present in both autocracies and democracies, while the remaining two are specifically associated with liberal democracies.

Later in the book, Mansfield and Snyder admit that measures for such institutions "are scarce" (p. 87), compelling them to use a proxy from the Polity III data set called "domestic concentration"—that is, "the degree to which domestic authority is concentrated in a state i's central government in year t-1." For students of democratization, this seems an odd way to measure the strength of political institutions, since most analysts tend to equate concentration of power with democratic erosion and dispersion of power with democratic consolidation. Checks and balances or veto players are often good for effective government. Moreover, there are plenty of regimes today with highly concentrated levels of decision making but very ineffective states, such as Angola. Given how much of the authors' argument rests on a measure of institutional strength, the conflation of state-power measures with regime attributes is confusing.

Indeed, Mansfield and Snyder at times use the terms "state" and "regime" interchangeably. It would have been easier to follow their argument if they had adopted the more standard practice (for students of democratization) of treating regime type and "stateness" as two separate categories, since some autocratic regimes can have little state capacity (think of Angola) while others have strong states (think of China), and some democratic regimes have weak states (think of the Philippines) while others have strong state capacity (think of the United States). Given how much of Mansfield and Snyder's argument rests on assessments of weak or strong institutions, it would have been useful to help the reader understand when they are talking about the type of institutions (democratic or autocratic) as opposed to the capacity of those institutions (strong or weak).

Moreover, the qualitative discussion of their case studies would have been a good place to describe a correlation between the Polity variable—domestic concentration—and the more recognizable features of weak or strong institutions. But in their case studies, qualitative descriptions of the level of "rule of law," "professional news media," or "organized political parties" are disappointingly thin.

Once the importance of "stalled transitions" and "weak institutions" to the authors' causal argument becomes clear, the reader begins to question why Mansfield and Snyder continue to label the independent variable "democratization." The book really describes why failed state-building and failed democratization lead to war. As Thomas Carothers has forcefully explained in the pages of this journal ("The End of the Transition Paradigm," January 2002), a stalled transition is not democratization—the latter connotes motion, the former the absence of motion. Moreover, in Mansfield and Snyder's discussion of their case studies, the weakening of states—not regime type—does most of the explanatory work. With reference to their entire data set, the authors report that "states with weak institutions are not especially war-prone unless they are at the same time undergoing incomplete democratic transitions" (p. 33). This claim would have been strengthened if the alternative hypothesis—that weakening institutions increase the probability of war—had been tested independently and the results reported.

The authors' restricted set of cases of democratization is also a small fraction of the total set of cases of war. Between 1816 and 1992, there were 79 interstate wars and 108 "extra-systemic" wars (that is, colonial conflicts against nonstate actors). In these conflicts, 47 allegedly democratizing states were involved in war. But as Mansfield and Snyder state explicitly, "we survey ten case studies that comprise all of the democratizing states in our data set that initiated interstate wars" (p. 169, emphasis added). Even if democratizing regimes are war-prone, they did not cause the vast majority of wars over this 178-year period. Moreover, when the scale of various wars is taken into account, the asymmetries between wars started by autocrats and wars started by democrats are even greater. The Second World War and the 1879 War of the Pacific (a brief conflict involving Chile, Bolivia, and Peru) should never be treated in any data survey simply as "two observations."

With regard to their case studies, the authors report that "of the ten cases reviewed in this chapter, seven—France [1849–70], Prussia/Germany [1848–70 and after], Chile [1879], Serbia [1877, 1885, 1912, 1913, and 1914], Iraq in 1948, Argentina [1982], and Turkey [1974]—provide clear support for our theory," but "two cases (the United States in the Mexican War and Thailand in the Franco-Thai War) neither clearly support nor refute our theory," and "four wars with ostensibly democratizing initiators are false positives: Guatemala's two wars, the Franco-Spanish War of 1823, and the Iran-Iraq War" (p. 225). This result is hardly robust.

The results are even less compelling when one looks more closely at the cases allegedly confirming the theory. Most of them are not cases of "democratizing" states, but instead cases of regime collapse or a return to autocracy. Long ago, theorists of democratization and revolution counseled against conflating the analysis of regime failure with what comes later. Greater clarity in classifying regimes is essential if the kind of regime is to be assigned so much causal power in the theory.

Space does not permit a full reassessment of the coding of all ten cases. For several, though, one can use the language provided by Mansfield and Snyder themselves to ask why democratization is getting blamed for war. For instance, in the discussion of belligerent France in the nineteenth century, Mansfield and Snyder report that Polity III codes the Orléans constitutional monarchy (1830–48) as a democratizing regime which "completed its democratic transition in 1848" (p. 187). Yet their own narrative undermines this claim of completeness, since they report that Emperor Napoleon III seized power in a coup d'état in 1851. Comparativists usually code coups as a return to autocracy. Therefore when France goes to war with Russia in 1854, it is odd to call Napoleon III's Second Empire a "democratizing participant" (p. 187).

The same "concept-stretching" must be deployed to label Prussia and then Germany as democratizing states in the various wars fought between 1864 and 1945. Mansfield and Snyder report that Otto von Bismarck became minister-president of "autocratic" Prussia in 1862 (p. 194); two years later, however, Prussia gets recoded as a democratizing state. Then, according to the authors, "Prussia/Germany shows up as a war initiator undergoing incomplete democratization only for the wars of 1864 and 1866, but the causal processes set in motion by Germany's pattern of incomplete democratization left their mark on all of the wars between 1848 and 1945" (p. 196). Again, leaving aside the coding of Bismarck's Prussia as a democratizing state, it is hard to defend the claim that Germany's "tortured path toward democracy created the impetus toward its five aggressive wars between 1864 and 1939" (p. 6, emphasis added)—in other words, Hitler was the ruler of an "emerging democracy" (remember the subtitle of the book). Most scholars (see, for instance, Nancy Bermeo's Ordinary People in Extraordinary Times, 2003, p. 23) mark January 1933 as the initiation of autocracy in Germany, so surely it was a dictatorship that started the Second World War, not an emerging democracy. It was democratization in West Germany after 1945—when political institutions were weak, not strong as during the Nazi period—that made this country a more peaceful neighbor. (Of course, U.S. military power also helped to keep the peace, but occupation, as we know from Iraq today, is not a sufficient condition for peace.)

The same concept-stretching is on display when the authors categorize as democratizing regimes the Ottoman Empire in 1911, 1912, and 1913 (the Young Turks became increasingly authoritarian after 1908, the parliament was shut down in 1912, and a coup occurred in 1913), or the military junta in power in Argentina during the Falklands War, or the Iraqi monarchy in 1948. If one does a library search for books on democratization with respect to these cases, nothing comes up; specialists on these countries and historical periods simply do not use the label "emerging democracy" to describe the kind of political change that was taking place.

In Mansfield and Snyder's data set of democratizing countries, there are also some very strange "false positives" (cases where they acknowledge that their theory's causal mechanisms do not apply)—the Soviet Union in 1956, Cambodia in 1975, and Iran in 1980 (p. 285)—but these thankfully get left out of the case-study discussion. Their appearance in the data set, however, should have given the authors pause concerning both their data and their model.

The objection to calling regimes led by juntas, emperors, and fascist dictators "emerging democracies" is not just a matter of proper labels or normative claims. Rather, it is crucial to understanding the central causal claim of the book about the pernicious effects of elected leaders using nationalist rhetoric to mobilize voters. After a coup or a suspension of elections, why should the voters be blamed for what belligerent leaders do? More generally, for a book called Electing to Fight, there is very little discussion about voter preferences for war. In all the case-study discussions, elites are seen capturing institutions and manipulating coalitions to go to war; voters are not "electing to fight."

The same problem of miscoding taints the ad hoc discussion of democratizing states initiating war in the 1990s, the final empirical chapter of the book. Again, due to space limitations, I can consider only a few examples. Before discussing them, however, let us again be clear: Despite Mansfield and Snyder's assertion that the causal connection between democratization and war is especially striking since the end of the Cold War, most countries that democratized in the 1990s did not go to war. Even amid the tumultuous transitions in the communist world after the collapse of the Soviet empire, the overwhelming majority of new states in the region avoided interstate war.

Nagorno-Karabakh is the first war discussed in this chapter, but its timing undermines the book's argument. This conflict began in February 1988, two years before the first multiparty elections in either Armenia or Azerbaijan, and well before either republic had become an independent state. The obvious cause of this conflict was Soviet imperial breakdown. The kind of regime in place in either country (or republic, before independence) is inconsequential for explaining the conflict.

Another case, the Russian war against Chechnya in the years 1994–96, is intriguing, but also revealing for what the authors leave out. Incomplete democratization contributed to the beginning of this war, although again, the real driver was state collapse, not regime type. Yet what the authors do not discuss is that democratic institutions also helped to end this war. Independent media coverage, most notably by NTV television, undermined popular support for the war in Russia. Yeltsin and his campaign team then responded to these electoral preferences by ending the war in April 1996, three months before the presidential election. The faction that had originally pressed for intervention, the so-called "party of war," was dismissed from government in 1996. It is the absence of these democratic institutions in autocratic Russia today that has prolonged the second Chechen war, now in its eighth year. Independent opinion polls show that strong majorities in Russia no longer support the war, but Putin's regime does not provide a mechanism for these preferences to shape policy. More democratization in Russia would lead to more peace, the opposite of the claim made in Electing to Fight.

The Kosovo War demonstrates the same pattern. By the time that Slobodan Milo√eviæ launched his project of ethnic cleansing in Kosovo in 1999, few analysts considered the Serbian regime "democratic." Milo√eviæ was a dictator, albeit a weak one, who was then overthrown in the first "color" revolution in October 2000. This democratic breakthrough made the regime more democratic, albeit still not a consolidated democracy. Nonetheless, since October 2000 Serbia has not initiated any wars. Democratization has made Serbia more peaceful—again, the exact opposite of what Electing to Fight claims or predicts.

Finally, the policy suggestions following from the analysis in Electing to Fight—which, it is only fair to note, are often made by others invoking the book and not by the authors themselves—are misleading and impractical. Even without invoking the violence, repression, and disorder involved in producing strong state institutions in Europe in the eighteenth and nineteenth centuries or South Africa in the twentieth century, there is simply no just or practical way to ask citizens to accept disenfranchisement until their elites build strong institutions, especially when today's autocracies have such a poor record of actually building them. Moreover, as confirmed by Michael Bratton and Eric Chang's recent article, "State Building and Democratization in Sub-Saharan Africa: Forwards, Backwards, or Together?" (Comparative Political Studies, November 2006), successful democratic transitions help to strengthen state institutions (think of Poland in 1989, South Africa in 1994, or Ukraine in 2004). And while gradualism can sometimes be useful for producing democracy and strong states, it is not a necessary condition for producing either (think of the Czech Republic, Romania, or Mongolia).

The historical evidence provided in this book fails to support the notion that voters in new democracies are the cause of war. Therefore, to reduce our support for democracy out of fear that electorates will be war-prone is not only questionable from a normative perspective; it is also unjustified empirically.