Following Kant, many scholars have argued that rulers often benefit mo
re from war than do their subjects, and thus that rulers with more aut
onomy from subjects will initiate more wars. They usually test this ar
gument by focusing on whether democratic states are less prone to init
iate wars than autocracies, and generally find little or no relationsh
ip, These are not adequate tests of the general argument, since they t
urn both ruler autonomy and the interests of actors into rough dichoto
mies (democracy vs. autocracy, rulers' interests vs, interests of all
subjects), and they ignore opportunity costs. This article uses a mode
l of state policy formation based on agency theory to provide a better
measure of ruler autonomy by differentiating between institutional au
tonomy and resource autonomy, We also use a more nuanced specification
of the interests of different groups of subjects, taking their opport
unity costs into account. This model allows us to derive more precise
propositions about the relationship between ruler autonomy and war ini
tiation. An analysis of war in four Western European states (England,
France, Sweden, and Spain) between 1400 and 1700, using legit regressi
on and qualitative comparative analysis, provides some support for the
central propositions of the theory.