In order to meet and assess progress towards global sustainable development goals (SDGs), an improved
understanding of geographic variation in population wellbeing indicators such as health status, wealth
and access to resources is crucial, as the equitable and efficient allocation of international aid relies on
knowing where funds are needed most. Unfortunately, in many low-income countries, detailed, reliable
and timely information on the spatial distribution and characteristics of intended aid recipients are rarely
available. Furthermore, lack of information on the past distribution of aid relative to need also hinders
assessments of the impacts of aid. High-resolution data on key social and health indicators, as well as
how aid distribution relates to these indicators are therefore fundamental for targeting limited resources
and building on past efforts.
In this study, we show how modern statistical approaches combined with a new geographic database of
aid distribution can be used to map the distribution of indicators with a level of detail that can support
geographically stratified decision-making. Based on geo-located survey data from Demographic and
Health Surveys (DHS) in Nigeria (2008 - 2013) and Nepal (2006 - 2011), Bayesian geostatistical models
and machine learning approaches were used in combination with a suite of spatial data layers to create
high-resolution predictive maps for (i) the rates of stunting in children under the age of five and (ii) the
household wealth index. An ensemble model was also exploited for aggregating different modelling
results to improve the modelling prediction capacity in Nigeria (for stunting 2008). By combining these
maps with information on the disbursement of aid for increasing food security and alleviating poverty
(AidData database - http://aiddata.org/), we quantified both the reported spatial distribution of aid relative
to stunting and poverty, as well as how changes in these indices overtime related to aid disbursement.
While many cases of aid disbursement lacked detailed spatial information, the results here demonstrate
the potential of this approach and highlight the value of spatially disaggregated data on the distribution
Funding: This research was supported by AidData at the College of William and Mary and the USAID
Global Development Lab through cooperative agreement AID-OAA-A-12-00096. The views expressed
here do not necessarily reflect the views of AidData, USAID, or the United States Government.
Note: This article contains high-resolution maps and images. The link above is for an optimized PDF (2 kb). For the full-resolution PDF (56 mb), please click here. The PDF may take a few moments to load.