As described in chapter 1, one of the motivations of this project is the effort to extend knowledge about the conditions under which great powers (and the developed world more generally) fight wars, so as to determine whether that knowledge can increase our understanding of the conditions under which developing states fight wars. I suspect few would object to such a motivation, but some might not agree with the specific body of existing knowledge I use as the basis for my extension. In this book I extend and adapt power transition theory, a theory developed to account for the incidence of wars fought for control of the international system among the very strongest of states. Some might question why I would focus on any structural theory of international behavior when strategic theories have gained such popularity and offered so many insights. Others might question why I would specifically select power transition theory as the best candidate among structural theories. In order to address such potential concerns, I describe power transition theory in some depth. Having summarized the theory, I suggest it is the best candidate structural theory because it has achieved so much empirical support and offered clues about a variety of international phenomena beyond war and peace. I then address concerns that strategic theories may offer a more promising avenue for extensions such as I attempt.

The need for a weaving together of the strands of this book seems greater than is the case with most volumes. The reason for this is that I address three interrelated tasks in this book. The first is an evaluation of the multiple hierarchy model extension of power transition theory to minor power regions, necessitating consideration of whether such an extension is justifiable. The second task is empirical identification of what a regional sub-system is. Unless regional sub-systems of the international system are objectively identified, extensions of great power theories to the minor power level cannot be empirically evaluated. The final task is an exploration of possible sources of persistent cross-regional variation in the relationships between power parity, status quo evaluations, and war onset. All three of these tasks have been accomplished, at least partially, in the preceding chapters. However, my book offers only a first perspective on each, and more can clearly be done to address all three. The third section below lays out steps likely to help any effort to move beyond my preliminary achievements.

Summary

Power transition theory was created to account for great power interactions. No clearer statement of this is needed than Organski and Kugler's own recognition that their work could:

Perhaps the most obvious criticism of the results reported in chapter 5 is they are based on analyses of a dataset which includes neither cases nor variables representing great power interference in minor power interactions. This may be especially egregious in some situations. For instance, it is hard to think of the outbreak of various Arab–Israeli wars without consideration of American and Soviet activity in providing weapons, intelligence, etc. At the same time, it is not hard to imagine that Israel and Egypt were at parity, one of them dissatisfied with the local status quo, and both itching to go to war with the other in, say, 1965. However, escalation in Vietnam might have drawn American strategic attention away from the Middle East, and thus the Americans might have pressured the Israelis into avoiding war at that point in time.

In these hypothetical cases the actions of external great powers affect whether minor powers wage war. In the first case superpower activity makes war more likely, while in the second it makes it less likely. In both cases the activity of one or both superpowers might have been an important causal factor in whether war occurred. In none of the analyses in chapter 5 is such activity represented in any way. Consequently, if external great power activity is systematically related to the occurrence of minor power war, the analyses in chapter 5 are potentially contaminated by omitted variable bias. Omitting variables does not mean I believe they have no effect.

In the analyses reported in the previous chapter, the Africa and South America dummy variables are always statistically significant. The statistical interpretation is Africa and South America are persistently different from the other regions in terms of war-fighting. I “know” these regions vary from other regions, but I have no idea why. I am specifically ignorant about the sources of cross-regional variation.

A consequence of estimating statistically significant region-specific coefficients is I can switch the regional dummies on and off and calculate how large an effect parity and dissatisfaction have on the probability of war within each specific region. When I do this I find a diminishing impact, substantively, of parity and dissatisfaction as I move from analysis of the Great Powers, through analysis of the Far East and Middle East, to analysis of South America and Africa. The diminishing impact and negative sign of the region-specific coefficients suggests an increasing pacificity. Regardless of whether parity and dissatisfaction are present, war is less likely in some regions than others. Since Africa's coefficient is the most negative, and since the corresponding substantive effect of parity and dissatisfaction on the probability of war is the smallest, we might conclude from this that Africa is the most peaceful region.

The notion of an African Peace is not intuitively appealing. News coverage of atrocities in Rwanda, of seemingly constant strife in the Horn of Africa, and within Congo/Zaire in the late 1990s, presents an image of a Hobbesian world, of the “war of all against all” (see Kaplan 1994, and Economist 2000a, 2000b).