Abstract. In many developing countries, public service provision continues to fall short of
demand. In the presence of severe infrastructure backlogs and different returns on public investment
expenditure, marginal benefit incidence theory envisages that measures aimed at maximizing
average access rates have contradictory impacts in the medium term. While relatively uniform
expansion of access coverage across target areas can be achieved in some sectors, geographical
disparities may persist or worsen in others. This study revises and extends a previous modeling
approach by testing for endogenous eligibility, geographically-varying functional relationships, and
number of uncompensated losers (numbers effect) as an additional social welfare objective. Relative
to medium-term changes in access rates in primary schools and healthcare, spatial and geographically
weighted regression models are applied to districts in Niger. Results point to an eligibility
threshold which exceeds the average coverage rate for primary education, some evidence of numbers
effect as a target for healthcare, and substantial spatial heterogeneity particularly for primary
schools.
1