: Trend Estimation of Child Nutrition Indicators at Micro-Level Administrative Units Using Night-Time Light Data

Bernard Baffour , Australian National University
Sumonkanti Das, Australian National University
Syed Abdul Basher, East West University
Salim Rashid, East West University
Alice Richardson, The Australian National University ACT 2601

Chronic undernutrition in children is an important population health problem with adverse outcomes in both the short and longer term - and is a specific SDG target. In order to provide an evidence base for the achievements of these goals, this study aims to estimate trends of chronic undernutrition (stunting) in under-five year old children in Bangladesh for micro-level administrative domains (64 districts and 544 sub-districts) over the period from 1997-2019 using satellite-based night-time light intensity data as a proxy measure of urbanization. Bayesian multilevel time-series models are developed using direct estimates of stunting and their standard errors for the target small domains, extracted from micro-data. These models borrow cross-sectional, temporal, and spatial strength in such a way that they can interpolate stunting levels in non-survey years for all small domains. The night-time light intensity makes a significant contribution to borrowing strength over space and time. Results show that the national-level trend is in steep decline over the period, from 60% to 30%. The trends at district level shows some districts with higher stunting levels over the last two decades have consistently higher vulnerability, while others vary more. At sub-district level, the direct estimates, which were too volatile - with up to 50% of domains missing - are considerably improved through using multilevel time series modelling. Findings of the study provide data-driven evidence for monitoring the progress in meeting nutrition goals at the detailed administrative domains in Bangladesh. Secondly, using accessible remote-sensed data like night lights intensity improves the precision.

See extended abstract

 Presented in Session 52. Modelling Subnational and Spatial Variation