He Who Lets Slip the Dogs of War

July 1, 2016

Michael C. Horowitz, Allan C. Stam and Cali M. Ellis, Why Leaders Fight (Cambridge University Press, 2015).

This is a book that should have been unnecessary. Its main argument is that leaders, particularly chiefs of state and government, are important to the decision to go to war. As the authors argue repeatedly, we all know this. Hitler made a difference to history that some other leader of Germany might not have. People matter. Human agency should not be ignored. Why do we have to prove it?

The problem is an academic one. American political science, in particular its international relations theory, pays little attention to the differences leaders make. Academia attributes war to institutional, structural, and internal factors, leaving little intellectual headroom for the voluminous biographies that sell so well in the mass market and fascinate war and peace practitioners like me.

The solution the authors propose to this disconnect is to apply the quantitative techniques of modern political science to code “leader attributes” that make people more or less prone to take on the risks that the decision to go to war entails. Their claim is that leaders matter “in systematic and predictable ways,” though the “predictions” are like the weather, subject to considerable uncertainty.

They approach this quantitative challenge by assembling their own comprehensive data base of “Leader Experience and Attribute Descriptions” (LEAD) covering more than 2,400 leaders in power worldwide between 1875 and 2004. The attributes that contribute to the risk of leaders initiating interstate wars include coming to power in a coup or after fighting in a rebellion, military service without combat experience, and age (older leaders are willing to take greater risks, especially in democracies). While there is some unexplained inconsistency with their graphs, the authors find other attributes have less effect, such as the level of education, parental stability (except perhaps illegitimacy), poverty, and birth order. There is not enough data on women to determine the impact of gender, but those few who have made it to power in recent decades seem no more or less prone to go to war than men.

The attributes that make a difference are often more important in autocracies than democracies, which is consistent with democratic peace theory. Personal attributes, in other words, interact with the institutional environment as well as with particular circumstances. Risk-prone leaders do not always find occasion to make war. Less risk-inclined leaders like Woodrow Wilson may find themselves in situations that push them towards war. The authors also argue that sometimes a leader with risky attributes like Costa Rican President Figueres Ferrer decides to take a different risk: He eliminated his army in order to maintain peace.

The point is not that personality explains all, but rather that “the systematic inclusion of individual leader-level variables tells us an important story that traditional international relations theory overlooks.” I find the evidence persuasive and suggestive, though I confess that I have not delved into what must be a voluminous technical annex online.

There are, however, some problems with the book. The first is one that plagues many studies that draw on large datasets: the accuracy of the depictions of certain cases. For example, the authors use a database of interstate wars that codes Franjo Tudjman, the bête noire of this book, as responsible for the initiation of the Bosnian war. Slobodan Milosevic deserves a good deal of blame not only for that war but also for those Serbia fought with several other now independent countries once part of former Yugoslavia. He escapes without much blame.

Second is an issue the authors themselves point to in their conclusions: Interstate war has not in recent decades been common. Civil wars, often with extensive international involvement, are far more common. The approach used in this study needs to be applied to them as well. My guess is that will be methodologically difficult. As one example, the war for Croatia’s independence that Tudjman is said to have initiated looked from Belgrade like a civil war, one in which Milosevic claimed to be repressing a rebellion.

Third, and most profound, is that the authors fail to take into account the work of Daniel Kahnemann’s Thinking Fast and Slow and other psychologists on the perception of risk. That work has made an enormous difference in economics, where it has spawned new insights by undermining the notion that people are rational actors. I can only wonder how taking Kahnemann’s work into account might affect the arguments in Why Leaders Fight.

David C. Gompert, Hans Binnnendijk, and Bonny Lin have dissected the many mistakes leaders make in deciding to go to war in their Blinders, Blunders, and Wars. Why Leaders Fight takes a systematic approach to demonstrating that personal attributes are part of the cause for error. But I’d still like to see someone put these puzzle pieces together with Kahnemann’s work on how people perceive risk and react to it. That would be a major contribution.


Daniel Serwer is professor and director of conflict management at the Johns Hopkins School of Advanced International Studies.