What Explains the Reductions in Infant and Child Mortality in Matlab, Bangladesh?

Lauren Hale, Stony Brook University, State University of New York (SUNY)
Julie DaVanzo, RAND
Mizanur Rahman, Pathfinder International
Abdur Razzaque, ICDDR,B: Centre for Health and Population Research

Infant and child mortality rates have decreased substantially in Matlab, Bangladesh over the last two decades. We use data from the Matlab Demographic Surveillance System on nearly 94,000 singleton live births that occurred between 1987 and 2002 to investigate the extent to which the change in mortality over time can be explained by changes in social characteristics and reproductive behaviors. We estimate Cox Proportional Hazards models for four periods of infancy and childhood. Preliminary results show that changes over time in reproductive patterns and in other social characteristics explain up to 35% of the improvement, but a large percentage in the improvement can’t be explained by observable variables over time. In the full paper, we will do a decomposition to assess which specific variables’ changes played the largest role in the drop in mortality. We will also assess the extent to which the effects of explanatory variables changes over time.