Lake, Hierarchy in International Relations, APSA 2004
29
The key problem in any NEGD is the non-equivalence of the groups. We cannot
assume that members are randomly assigned. Especially in this case, we know that it is
likely that states intentionally select themselves into one or the other category. To
properly identify the effect of hierarchy on defense effort, therefore, we need to examine
the possible covariates associated with this selection decision. This paper does not
present a fully developed theory of the formation of hierarchy (see Lake 1999). Prior
theory, and an eclectic set of auxiliary arguments, suggest the following covariates. Data
and sources are, again, discussed in the appendix.
• Countries that are more threatened by their neighbors are more likely to
subordinate themselves to the will of a dominant state in exchange for the
protection offered by that power and, thus, are more likely to enter hierarchical
relationships. The countries that are most threatened, in other words, are most
likely to be willing to pay the high price that subordination entails. In the absence
of a hierarchical relationship, we would expect states in the experimental group to
carry a higher than normal defense burden. To identify the effect of hierarchy,
therefore, we must control for foreign threats. These threats are measured in two
ways: a) the highest defense effort of any contiguous state and b) whether the
country was involved in a militarized interstate dispute (MID), as defined by the
Correlates of War MID dataset (MID involvement). For most countries most of
the time, their most pressing threats come from their immediate neighbors.
Involvement in a MID indicates the presence of issues with a strong potential to
escalate to violence and threaten the state. To offset problems of simultaneity, and
the possibility that defense spending in some countries might be responding to
Convention