19
preferences over autonomy and differences in political institutions can create significant barriers to
the ability of each side to bargain and reach an agreement.
Given this model of self-determination politics, I am able to make a number of predictions
about when autonomy agreements will be more or less likely. The following hypotheses are derived
from the theory presented above. I present the hypotheses in two sets: predictions about the effects
of state characteristics and predictions about the effects of group characteristics.
Internal Characteristics of States
Internal divisions make it more difficult for states to bargain over autonomy and reach an
autonomy agreement. Because any disadvantaged factions must be compensated prior to a
successful agreement, the internal bargain between state factions becomes more difficult to resolve
when there are greater numbers of factions. However, the existence of multiple veto factions in the
state can also enhance the ability of the state to commit credibly to a new autonomy agreement. The
relationship between veto factions and agreements over autonomy will therefore be non-linear, with
agreements being made in the middle cases where there is more than one, but not many veto
factions.
28
Hypothesis 1: Agreement is most likely when there are a moderate number of veto factions.
Additionally, two other factors affect the ability of state factions to be adequately
compensated for potential loses due to a new autonomy agreement, and therefore the likelihood of
agreement. These factors are the size of potential losses and sufficiency of the state’s resources.
This leads to hypotheses two and three.
Hypothesis 2: The higher the rents from the contested territory, the less likely the state/group dyad is to reach
an agreement.
Hypothesis 3: The lower the amount of fungible resources the state has, the less likely the state/group dyad is
to reach an agreement.
28
Empirically, a moderate number of veto players is 2 – 4 inclusive. This is value that the middle 50% of cases takes on.