The hypothesis of different marginal benefit paths across public service sectors
can be tested on municipalities or districts within a national territory, or within
administrative regions in a decentralised system. Econometric estimates from spatial
Tobit models (with spatially weighted variables accounting for diffusion effects)
applied to primary education and healthcare across districts in Niger, suggest more
articulated patterns than the above arguments ([27]): both sectors appear to benefit
from ‘autonomous’ gains for worse-off districts, but hardly any additional gain
is found to accrue to these districts in terms of average treatment effects, relative
to districts randomly selected after accounting for socio-demographic characteristics
(with no selectivity bias based on Tobit estimates). However, econometric results of
that previous study rely on two alternative values for (a) the eligibility thresholds
(chosen a priori based on initial-period average rates of public service coverage across
districts and individuals, without testing for looser or stricter selection criteria), and
(b) the diffusion cut-off distances, respectively.