Why are stunting rates of children in India (above 40%) higher than so many much poorer countries in sub-Saharan Africa? There are many theories out there: women’s status is lower in India, the international growth standards do not apply to India, and poor sanitation.
A new paper by Dean Spears at Princeton sheds more light on the sanitation theory. The paper uses three different empirical strategies to estimate the causal relationship between open defecation rates and child height. It concludes that open defecation is responsible for “much of the excess stunting in India”.
The 3 strategies? (1) use all DHS surveys ever conducted that have both stunting and sanitation data (about 140 country-years) and collapse them to the mean and then run regressions on child height and a range of controls country level controls, (2) use cross sections of NFHS surveys from India to construct district level panels (the same subdivisions of states, with mean values within each district) and run similar regressions to (1), and (3) use the full set of DHS data from Sub Saharan Africa and India, this time at the child level, running similar regressions to (1).
All three strategies show large and statistically significant relationships between open defecation and child height, controlling for a range of other variables.
The paper is an important contribution because (1) of the triangulation of results via multiple methods, carefully done, (2) the suggestion of why even high income groups in India have many more stunted children than similar high income groups from other countries (the effects of open defecation cannot be escaped even if households have improved sanitation within their residence) and (3) because it explores rural-urban differences (open defecation seems more important in urban areas).
But the paper does have some things that need attending to:
1. Not enough attention to women’s status—I would have liked to have seen more use of the women’s empowerment variables in DHS and NFHS data
2. Not enough attention to food consumption. This is harder, but there are some food consumption variables in some of the DHS data and it would have been good to see how the open defecation results hold up
3. A more general formulation of the previous two points: the regression specifications do not seem to be sufficiently driven by the nutrition conceptual model with food, health, care and watsan as underlying variables, often mixing in income and other variables with these underlying determinant variables.
4. The paper is difficult to follow and coy on the percent of the India-Sub Saharan Africa gap that can be reasonably attributed to open defecation
It is good to have more solid evidence on open defecation to back up some of the guesses in other papers (see here). And like all good papers, it makes the reader wonder why this analysis had not been done before.
The really big puzzle is why take up of improved sanitation is so slow in India (55% in 2006 and 64% in 1998: barely an improvement of one percentage point per year—at that rate it would take 50 years to get to 5% open defecation)?