Without exaggeration, Jessica Weeks’s Dictators at War and Peace will shape fundamentally the study of International Relations for years. It makes a major contribution to the study of non-democratic conflict which, given contemporary international politics, appears at least as relevant as the much larger body of work on democracies and war. Furthermore, Weeks tackles a set of regimes inherently more challenging to study—for both theoretical and empirical reasons—than democracies. This is an important book.

In arguing that two factors, militarism and personalism, underpin autocratic foreign policy, Dictators swings for the fences in covering a diverse range of regimes parsimoniously. Its elegance is enhanced by its use of Dan Slater’s intuitive, jargon-free ideal types: “Machine” (non-personalist, non-militarist), “Junta” (non-personalist, militarist), “Boss” (personalist, non-militarist), and “Strongman” (personalist, militarist). ^{Footnote 1} The range of behaviors Weeks seeks to explain with this two-variable theory is no less ambitious: the initiation of conflict, the likelihood of defeat, and the probability of being punished after a military loss.

Weeks’s argument is straightforward. As long as the leader of a state can be punished for bad performance by a relatively independent set of elites, many of the foreign policy virtues associated with democratic exceptionalism should ensue. On the other hand, personalist dictators need not fear this source of punishment, and tend to be violent, risk-seeking individuals to boot. Personalist autocrats will therefore initiate more and riskier interstate conflicts. Weeks goes on to claim that military associations for either the personalist leader or dominant elites increase the state’s predisposition for war due to leaders’ belief in war’s virtue and effectiveness as well as an aversion to diplomacy.

Weeks tests each of her dependent variables quantitatively. The empirical evidence gets thinner, largely for data availability reasons, the further one gets from the decision to initiate a conflict. The chapter on conflict initiation tests a host of multivariate models, while punishment for lost wars is described with a table of frequencies. In subsequent chapters, Weeks examines several regimes qualitatively to explore their decision-making processes.

Because of its agenda-setting nature, the book deserves serious analysis as the field decides how to build on it. The theory’s elegance clearly comes at a cost in terms of explanatory power; many regimes fall outside the model. Even excluding, as Weeks does, the 1,452 “new or unstable” country-years in her dataset, Dictators’ “other” category represents more than twice the number of country-years as any other regime type.

Nonetheless, within this set of cases, the book presents a powerful finding. Other work before Dictators has argued that the size of the selectorate shapes foreign policy. ^{Footnote 2} However, Dictators finds that a little bit of accountability goes a remarkably long way. Empirically, there is little distinguishing democracies from Machines in terms of conflict behavior. However, Weeks uncovers massive statistical differences between personalist (Strongmen and Bosses) and non-personalist (Machines and Juntas) regimes. Weeks’s qualitative exploration of the reigns of Stalin and Saddam Hussein further illustrate the ambitions, violence, and lack of restraint inherent to these two autocrats.

On the other hand, it is not clear that adding militarism, in its current form, delivers much additional explanatory power. Indeed, by my reckoning there is only one instance where the military version of an autocracy differs statistically from its civilian counterpart. Relative to Machines, Juntas initiate more disputes, but do not significantly differ in terms of war outcome or punishment. Among personalist regimes, there is no significant difference for any dependent variable between Bosses and Strongmen. It is also noteworthy that there are no cases involving a Strongman in the book’s qualitative section.

According to Weeks, militarism influences the dependent variables through two different mechanisms based on the level of personalism: “narrow parochial interests” and the “military mindset” (p. 6). While Juntas’ corporate incentives (as well as their mindset) differ from Machines’, axiomatically a personalist dictator is not responsible to other elites. He has the narrowest of parochial interests: himself. A Strongman must therefore differ from a Boss largely via the psychological, “military mindset” explanation.

Dictators’ statistical analysis of militarism therefore tests two separate mechanisms. Of the two only the parochial interest receives support. Militarism adds modest additional explanatory value, and only through the corporate interests rather than the psychological mechanism.

Given the empirical pattern, Weeks’s argument can be reframed as a theory resting on the political economic interests of the dominant actor(s). Since the personalist leader enjoys exclusively the entire machinery of the state as well as the rent it produces, personal background probably matters little. L’etat c’est lui. On the other hand if a secure elite dominates foreign policy, then the elites’ parochial interests matter greatly. Potential interests need not be limited to military. Given that juntas are by far the smallest category in the Dictators data set, future research should go further afield. Whether the dominant elite group profits from international trade is one example of a relevant “parochial interest.” ^{Footnote 3} The political economic incentives for international conflict likely differ for communist and religious elites as well.

If one must single out one category of elites when discussing autocratic conflict behavior, the military would seem to be the logical choice. But the assumption that militaries are more militaristic is actually a problematic one. According to Weeks, in juntas the military “audience can benefit from arms buildups and war.” The conflation of these two activities is an unusual move. The Tsarist grand duke that famously hated war “because it spoils the armies” epitomizes the balance of civil-military relations work suggesting that military regimes are often reluctant to use military force abroad, especially for missions not essential to the state’s survival. ^{Footnote 4}

Weeks acknowledges this disagreement, but then supports sticking with this assumption using evidence of a civil-military divide from a 1998–1999 survey of American elites. ^{Footnote 5} But the United States is unlikely to be representative of democratic militaries, much less non-democratic ones. For example, today’s Israeli Defense Forces as a corporate body appears more dovish on military conflict compared to either that country’s current civilian leadership or its public opinion. While Weeks shows the military driving Japanese aggression up to World War II, the Wehrmacht was generally more cautious than Hitler. A study of Russia cited by Weeks also shows this complexity: military elites are both “more hesitant than their political counterparts to embrace an interventionist foreign policy agenda” but more “activist in considering the use of force as an instrument of foreign policy.” ^{Footnote 6}

That said, most recent statistical analysis supports a link between military regimes and conflict propensity. ^{Footnote 7} Indeed, I have never seen a more stark division between the conclusions of theoretical/qualitative and quantitative approaches to a research topic. But even this statistical link often comes with caveats. Horowitz and Stam find that only military leaders without experience in war are more belligerent. ^{Footnote 8} Carter finds that military regimes are more likely to pick fights against stronger opponents. ^{Footnote 9} Clearly, we are just beginning to understand the role of militaries in non-democracies (most qualitative work having been done on Western militaries).

Militaries vary in both their interests and their intended function. In many countries the army is a tool of development. Other militaries, such as Egypt’s, are funded through their participating in the domestic economy. In still other countries, military service is a means of maintaining a pre-existing civilian elite through patronage or social prestige, and thus reflects an antecedent civilian group’s parochial interests. This is likely to change the military’s international incentives and thus their propensity for conflict.

Finally, the military’s preferences evolve over time through such forces as the balance of power, the economy, or dominant military technology. Regarding World War I, T.E. Lawrence bitterly observed “Napoleon had said it was rare to find generals willing to fight battles. The curse of this war was that so few could do anything else.” ^{Footnote 10} Weeks’ concept of “military militarism” may be largely passé. In looking at the ideal illustration of Japan, Weeks chooses the last best case for her theory, what Vagts called “the one remaining stronghold of old-style great power militarism.” ^{Footnote 11}

In short, in addition to its own significant contributions Dictators points us in exciting directions for future research on autocracies and conflict. Personalist leaders and military juntas are just two corporate interests that, un-checked, can shape conflict propensity. Military regimes’ relationships to war defy simple assumptions. While Weeks has not given us a definitive answer about military regimes, nobody else has either.

I am honored to engage with Weeks in this critical dialogue, and hope this will not be the last time my own work will be placed alongside Dictators. I will assign this book at every opportunity and reckon with the argument for the rest of my career. I hope others do likewise.