Global Power Transitions and Regional Interests

Abstract

A number of studies have examined and largely validated power transitions as necessary conditions for war, yet the second critical element of power transition (PT) theory—a challenger's dissatisfaction with the status quo—has been analyzed to a much lesser degree. This paper is intended to address this research gap. Global power transitions between major powers can indeed be destabilizing, potentially triggering major conflicts, but we also argue that violent conflict is unlikely unless there is a clash of interests between the global contenders in the critical regions of their vital interests. Our study thus provides a modification of the original PT theory by identifying its two essential elements—(1) relative power and (2) attitude toward the status quo—at two different structural levels. Power transitions between major powers are still seen as critical at the global level, as originally stated by Organski and Kugler, whereas dissatisfaction with the status quo concerns the regional context of their interests. The paper provides a theoretical link between these two levels and quantitatively tests the argument, with suggested implications for further refinements of PT theory.

KEYWORDS:

• power transition theory
• regional interests
• major powers
• deterrence
• status quo

We would like to thank Christopher K. Butler, Chuck Finocciaro, Gregg Johnson, and Jason Sorens for their comments on an earlier version of this paper.

Notes

1. Zagare et al.'s models explore the impact of capability, credibility, and status quo dissatisfaction on conflict probability.

2. In this sense, for example, Japan's decision to attack the United States in 1941 should not be seen as the result of their dissatisfaction with the global status quo. Rather, it resulted from their own rising interests in Asia and the clash of these interests with the United States.

3. As Danilovic (2002) reports, the Cold War containment policy as an “indiscriminate” global policy was rather a lone exception for the United States as a major power. Even in the case of the U.S., she reports different behavior across regions.

4. As these challenges and demands can be manifested both verbally and behaviorally, Danilovic (2002) bases the typology of such moves on the MID coding rules as follows: (1) the threat of force; (2) the display of force; or (3) the use of force short of war.

5. Note that only the latter two categories of this dependent variable apply to situations where another major power stepped in to resist the challenger's attempt to change the status quo.

6. It is reasonable to object to this two-stage estimation for not accounting for possible selection effects. Indeed we also use Heckman's censored probit to estimate whether the status quo challenge is likely to occur (selection equation) and if it does, whether it is likely to escalate to major power crisis and/or major war (outcome equation). Note that Heckman's probit is disadvantaged by requiring binary dependent variables in both equations. If the correlation between the error terms is not statistically significant, then we can proceed with the standard approach of assessing both stages as statistically independent events as described above since the coefficients should not be suspected of statistical selection bias.

7. In her dataset, Danilovic (2002) classifies geopolitical regions as follows: West Africa, Central Africa, East Africa, Southern Africa, North Asia, South Asia, Southeast Asia, East Asia, North America, South America, Central America, and the Caribbean, Western Europe, Eastern Europe, Middle East, and Oceania.

8. To address the concerns regarding potential selection bias, the onset and outcomes of crises were estimated with Heckman's censored probit and the results showed a statistically insignificant correlation in the error terms (ρ) . Since there is no evidence of statistical selection effects between the selection and outcome models, the standard probit or logit estimates would not be affected by selection bias. Since Heckman's probit is disadvantaged by requiring binary choices for both the selection and outcome equations, we opt to estimate them with separate models due to the polychotomous nature of our crisis outcome variable.