The toilet gap: How much of differences across developing countries in child height can sanitation explain?


This page in:

Guest post by Dean Spears

A child’s height is one of the most important indicators of her well-being. Height reflects the accumulated total of early-life health, net nutrition, and disease. Because problems that prevent children from growing tall also prevent them from growing into healthy, productive, smart adults, height predicts adult economic outcomes and cognitive achievement.

Researchers studying height have long been puzzled by a paradox: Among developing countries, differences in average height are not very well explained by differences in income. In particular, children in India are shorter, on average, than children in Sub-Saharan Africa, even though Indians are richer on average.

What could explain this paradox? Because addressing widespread stunting is a health and economic policy priority, understanding determinants of children’s height is important. In a recently released Policy Research Working Paper (WSP 6351) I explore evidence for one possible explanation: open defecation. More than a billion people worldwide defecate openly without using a toilet or latrine. India, with some of the world’s worst stunting, also has one of the very highest rates of open defecation: more than half of the Indian population does not use any toilet or latrine.

Evidence in the medical and epidemiological literature has documented that germs in feces can stunt children’s growth. This is in part due to diarrhea, and in part due to enteropathy: chronic changes in the lining of the intestines that make it harder for the body to use nutrients. Well-identified econometric papers have also shown a causal link from sanitation to child height. For example, in a paper coauthored with Jeffrey Hammer about an experiment done in partnership with the World Bank Water and Sanitation Program and the government of Maharashtra, we find that a program that promoted rural sanitation also caused children to grow taller. Therefore, this new paper asks the quantitative, accounting question: how big is the effect of sanitation on child height? Big enough to account for important differences?

Open defecation around the world

Are the countries where many people defecate openly the same countries where the most children are stunted, and the average child is shortest? Yes, as demonstrated in the graph below. Cross-country differences in sanitation linearly explain 54 percent of the international variation in average child height. 

Each circle in this graph is a collapsed round of a Demographic and Health Survey, therefore, each represents one country in one year. The size of the circles is proportionate to the population of the country in that year. For example, the three largest circles at the bottom-right of the graph represent India in 1992, 1998, and 2005 – the three years when India had a DHS survey. One striking fact is that India’s circles fall on the trend line. Indian children are very short by international standards, but are exactly as short as widespread open defecation in India predicts.

Further analysis in the paper suggests that the association between child height and open defecation is not merely due to some other coincidental factor. It is not accounted for by GDP or differences in governance, female literacy, breastfeeding, or other forms of infrastructure such as availability of water or electrification. Because changes over time within countries have an effect on height similar to the effect of differences across countries, it is safe to conclude that the effect is not a coincidental reflection of fixed genetic differences.

Good toilets make good neighbors

If open defecation is indeed keeping children from growing to their genetic potentials – rather than merely being coincidentally correlated with height – we would expect open defecation to be more important for health outcomes where children are more likely to encounter whatever fecal germs are introduced into the environment. This means that population density should matter: living near neighbors who defecate openly is more threatening than living in the same country as people who openly defecate far away.

The graph below confirms that this is the case: child height is even more strongly associated with the average number of people per square kilometer in a country who practice open defecation. The density of open defecation per square kilometer, in this simple linear graph, can account for 64% of international variation in child height.

The graph plots the same 140 country years as before, and again the three large circles represent India. Once again, stunting among Indian children is no surprise: they face a double threat of widespread open defecation and high population density.

An “Asian enigma”?: Height differences between India and Africa

Households in India are less poor, on average, than households in Sub-Saharan Africa, but children there are shorter. Stunting is common even among relatively well-off families in India. However, widespread Indian stunting is not due to genetics: Indian babies who move to developed countries in early life grow much taller.

Because of the effect of open defecation on stunting, we can estimate how tall Indian children hypothetically would be if exposed to the better sanitation profile in Africa. Decomposition results in the paper – in the spirit of Blinder-Oaxaca – show that sanitation differences are able to completely explain this gap. Constructing a counterfactual sample of children’s heights in India’s most recent DHS weighted to match the exposure to open defecation in a set of pooled DHS surveys from Sub-Saharan Africa can eliminate the India-Africa gap.

This suggests that sanitation is very important, but it isn’t everything important. For example, in joint work with Diane Coffey and Reetika Khera, we show that children of women in India with lower social status grow less tall, identifying the consequences of intrahoushold rank by taking advantage of special properties of rural joint Indian families. Because these children are in the same household, the presumably are exposed to essentially the same sanitation environment. Further, because Indian households are richer, Indian children would be expected to be taller than African children, beyond merely eliminating the gap. Moreover, even matching African levels of open defecation, children in both regions would be much too short.

Stunting is often referred to as “malnutrition.” Sometimes in policy debates, this is taken to imply that the solution is to provide more food. But what these results suggest is that the disease environment is an important cause of “malnutrition,” too. If so, then far from merely a concern of infrastructure specialists, open defecation would a priority for health and nutrition policy – and for children’s well-being and the productivity of the next generation of workers.

Dean Spears is a PhD candidate in economics at Princeton University and a visiting researcher at the Delhi School of Economics. He is a co-founder of rice – online at – a research organization based in India and focused on the economics of early life health and human capital accumulation. He splits his time among Princeton, NJ; Delhi; and Sitapur, Uttar Pradesh.


David Cisewski
February 19, 2013

I think this information looks great and I agree there is a substantial problem concerning the sanitation and hygiene practices and their underlying effects on malnutrition (via enteropathy). It's great to see this issue brought into the public eye.

My only concern with the above graphs is that India represents a substantial portion of the data set with 3 prominent data points ('92,'98,'2005) and by far the largest population sizes. As a result, your regression analysis appears heavily skewed to reflect the findings of India. I am curious what these graphs would like look without India data. Just eyeballing the data, I am envisioning a pretty flat curve (limiting the overall findings). Would you be willing to supply regression excluding India?

David Cisewski.

Dean Spears
February 21, 2013

Hi, David,

Thanks much for your great question. I encourage anybody interested to take a look at the full paper, online at where these details and many others are covered.

What happens when you omit India? Table 2 omits each of 6 world regions in turn, including South Asia. The regression slope without South Asia is -0.988 height-for-age standard deviations linearly associated with moving from nobody openly defecating to everybody openly defecating (standard error 0.14), compared with -1.2 (se = 0.22) in the full sample. So there is still a downward slope, which is actually pretty similar. Another way of responding to your broader point would be to change whether or not observations are weighted by population size: is India treated as equally or more important than East Timor in fitting the line? The graphs like this in the paper plot these lines with and without weights.


David Cisewski
February 23, 2013

Thanks for the response, Dean. My mistake - I was under the impression the size of the circles indicated a weighted average. I look forward to reading the full paper.

K N Vajpai
March 15, 2013

I see Dean Spears article with all alien facts of comparing height of Indian children by just considering ‘malnutrition’ and ‘sanitation’ (toilet) as major reasons. In general there are certain scientific facts about the level of malnutrition among children in India, and therefore the results about overall growth pattern and the correlation . However, in my view knowing about a child by just measuring the height certainly seems a poorly researched correlation and poor scientific argument.

From this article it appears that because children in developed world take enough food, follow good infant feeding practices and have a toilet, so they are good in height. So, with this logic they fair in height with children of majority of African countries and may be Pathan and Gujjar children’s in India. With same logic what about the children in China, Thailand, Vietnam, Laos, Bhutan, Nepal and NE part of India, and for that matter in developed South Asian countries like Japan, Hong Kong and South Korea? Don’t we have to consider other major factors affecting the height of a child? In my view the researcher needs to think more scientifically and work further on all available facts before coming to any such conclusion.

Again, we certainly can’t learn a lot from measuring the height of a children scientifically, while, as a researchers we can draw ‘some’ inferences about a child’s lifestyle. Therefore, height can’t be treated as an major (only) indicator for absorbed nutrition, child’s early-life health and disease history. If this is the case, then what about genetic diversity within India and various environmental and social factors? Enough food can’t be treated as a major factor for better health, but, nutritive food should be. What about majority of Americans those are poor in taking nutritive food as well?

Also, there are confusing facts given in this research where in place of 74 % Household those didn’t had any toilet facility as per NFHS 2005-06 (Link:…) has been quoted 55%, similar is the case with Census 2011. Therefore, one should be cautious while referring this study in terms of quoted facts and figures as well.

The conclusions by quoting a few generic example of genetic history are among the poor facts in this research to suggest upon the children’s height pattern in India.

Taking stock from Dean’s research, I wonder, how come malnourished Indian children with no-latrines from its poverty ridden states like Bihar and UP succeeded in giving world class managers, engineers, doctors, scientists, politicians, diplomats and researchers?

Just a food for thought for other such researchers as well.

The truth
November 16, 2016

Its proven fact that when you live in a trash dump like India with 85% of the population urinating on the road, you will get shorter people, there is a reason why more people from India go to work in Africa instead of the majority of Africa wanting to move to India, they would rather go to Europe instead, the average person in Nigeria is 5 ft 4, majority of Africa is 5 ft 7, trust me, Africa is a lot cleaner than India. As for the Chinese its well documented that northern people in China average at 5 ft 11 while southern people of China average at 5 ft 6 simply because people in northern china at more wheat and less rice, its well know that most East Asian people have staple rice in their diet, this is also the case with people of southern India also, while wheat is mostly consumed in the north near Punjab, the average Jatt and Gujjar people are on average much taller and heavier than the other Indian people because they have been around wheat, dairy, and physical labor also which gives them their tall stature. People in southern China work on farms and they still don't grow simply because of their carbohydrate diet. I believe India's average can be 5 ft 11 if sanitation and wheat consumption is seen. Also the average Pashtun is 5 ft 6 around the cities in Afghanistan while the nomad mountain Pashtuns of Pakistan are 5 ft 7, the Jatts of Pakistan are usually from 5 ft 8 - 7 ft 5, considering most of Pakistan's pehlwan and stonelifters are Jatts and Gujjar people. Also in times of poverty, its the brain that is given the most nutrition which is why some biharis are smart because the few nutrients they get is given to the neurons and none for their bones.