(Page 2 of 30) - Balance of Power Dynamics and War Escalation authored by Bennett, Scott.

Introduction

International relations scholars have a problem. Despite decades of theorizing and

empirical examination, we still do not know much about the balance of power, and what we do
know about is actually suspect. In particular, although it will be surprising to many, we don’t
actually know what we think we know about how the balance of power affects the escalation of
disputes to war. What do we think we know? If one were to examine recent literature on
conflict onset, or to survey IR scholars, there would appear to be a developing consensus that
balance of power logic is somehow incorrect. Empirical research consistently finds dyadic
balances of power to be associated with more conflict rather than less, suggesting that the
traditional balance of power hypothesis that balance is conducive to peace if incorrect. In
additional, rational choice logic tells us that the observable balance of power should have no
effect on conflict propensities, because states can observe this balance, take it into account, and
adjust to it before we ever see it making a difference. Unfortunately, it is premature for us to
reach the conclusion that we know the effects of the balance of power on conflict. In fact,
existing evidence on the balance of power and conflict is limited and specific. The emerging
consensus that situations of relatively equal power are more conflict prone than situations of
imbalance has a solid backing only when applied to the onset of militarized disputes, and not to
their escalation. IR scholars have in fact not had a good way to study the relationship between
the balance of power and dispute escalation to war, and in particular we know very little about
the dynamics of the balance of power in disputes, and how the changing balance of power over
the course of a crisis affects escalation patterns. This is in large part because we focus our
attention on the militarized dispute or crisis as our fundamental unit of analysis, asking whether
the dispute as a whole escalates to war or not. When we analyze the dispute/crisis as a whole,
we are unable to study any patterns or events going on at any finer level, such as over time
within a dispute. This poses a fundamental challenge when we study the balance of power,
because one of the core conceptions of crises is that they are contests of brinksmanship, where
players take actions in response to one another intending to influence their behavior during the
course of a crisis or dispute. Such maneuvering – including the effects of the participation of
new allies joining the conflict – is necessarily missed by current studies.

The balance of power is exemplary of a class of problem that we cannot model

appropriately given current common research designs in quantitative international relations. Any
time we use a research design which imposes a temporal or spatial delimitation on cases through
a fixed unit of analysis, but where there is variation within that unit, we have a situation where
the imposed unit has limited our understanding. In particular, situations where we impose a
minimum temporal observation period on the unit of analysis (for instance, the year, or month, or
week) even though there are meaningful changes in variables over time within that unit pose an
important problem for researchers. I refer to variables that change over time, and where we care
about those changes, as “dynamic” variables. This is in opposition to static conditions that
remain fixed. When changes occur sub-unit, but our observations are only at the level of the
larger unit, we impose having one and only one observation of the variable in question.
Properly, this observation must be made at the beginning of the unit in question, because any
other measurement would include ex post information that should not be part of building the
case. But this poses an enormous problem for linking theory and analysis, because our theories
often tell us that the evolution of some variable (or situation) over time is critical to the outcome
of our cases.