Child #115181 in the Demographic and Health Survey we’re looking at is 38 months old. Let’s call her María. Her older brother, child #115201, is 51 months old. Let’s call him Alejandro. Despite their 13-month age difference, María and Alejandro are both 92cm tall. María is rather short for her age – she’s at the 18th percentile of the reference population of well-nourished children. She’d be 96 cm if she were average. Alejandro is extremely short – he’d be over 10cm taller if he were average height for his age. He’s so short for his age, in fact, that he’s not even at the 1st percentile. Technically Alejandro is ‘stunted’ – at ‑2.64 his ‘height-for-age z-score’ is less than 2, meaning his height-for-age is less than 2 standard deviations below the median of the reference population.
Since they’re both past their second birthday, the chances are that María has escaped stunting but Alejandro will stay stunted. His life chances are much worse than his younger sister’s. Alejandro’s extreme shortness was caused by a mix of cumulative exposure to infections and severe undernourishment in utero and infancy. These together will have affected the development not just of his body but also of his brain. María will overtake him cognitive and socioemotional skills and stay ahead. Alejandro will do worse at school and leave school earlier. He will earn less in adulthood – in part because of his less developed cognitive skills but also because of his shorter stature. He is also at greater risk of non-communicable diseases in later life.
The aggregate stunting income penalty
Moving beyond María and Alejandro to other stunted and non-stunted children in their country, we can see how entire countries pay a penalty – in terms of a lower per capita lower income – for not addressing stunting in early childhood. Several studies have tried to quantify this aggregate penalty by looking directly at the association between stunting and adult height, and height on earnings, or by using data linking stunting in early childhood to earnings in adulthood from a seminal small-scale efficacy study conducted in 4 impoverished villages in rural Guatemala.
In a policy research note that came out today, we and our co-authors assembled a broader evidence base to re-estimate the aggregate income penalty that countries incur from childhood stunting. We calculate the effects of childhood stunting on lifetime income operating through all potential pathways: reduced schooling (‑1.6 years on average), reduced height in adulthood (‑6cm on average), and lower cognitive skills (‑0.6 standard deviations on a typical test). For each of these pathways, we translate the deficit into the lower labor market returns – and hence lost adult income – that a stunted child incurs when she join the labor force.
These numbers allow us to put a number on the following thought experiment: how much higher would a country’s GDP per capita be today if none of its current workers had been stunted in childhood? The answer obviously depends in part on the prevalence of stunting when today’s workers were children, which depends in part on the median age of today’s workers. In the case of Chile, the rate of stunting when today’s median-age worker was a child was just 8%. In Bangladesh, it was 70%.
We adopt a method used in the growth literature called ‘development accounting’ to combine these stunting rates with the income penalties in terms of schooling, cognition and height incurred by a worker who was stunted in childhood. We find that the average country’s GDP per capita is 7% lower than it would have been if none of its current workers had been stunted in childhood. In Africa and S Asia, the average is even higher (9-10%).
But what are the returns to interventions that reduce stunting?
The potential economic benefits then of interventions to reduce stunting are sizable, even if they occur many years in the future. Realizing these benefits requires interventions that reduce stunting. The more effective they are in reducing stunting and the cheaper they are, the larger will be their economic returns.
There is one set of interventions for which researchers have estimated both the per capita costs and the overall impacts on stunting prevalence. It is a package of 10 key ‘nutrition-specific’ interventions reaching pregnant women and children in the first 1,000 days, starting at conception, including things like micronutrient supplementation, energy protein and calcium supplementation in pregnancy, nutrition education, complementary feeding, and breastfeeding promotion. Researchers estimate that if these interventions were scaled up across 34 high-stunting countries from current levels to 90%, stunting rates would fall by 20%.
This isn’t an especially large impact, but as we have seen above, even a small change in stunting is associated with large changes in income. Moreover, these interventions might have impacts on other key child outcomes that are also part of the Sustainable Development Goals (SDGs), such as child survival, and their combined cost is not that large (less than $5 per capita).
Putting these relatively low program cost figures together with our earlier estimates of the long-term income gains from reducing stunting, we come up with a benefit-cost ratio and a rate-of-return for this nutrition package. We factor in that the costs are incurred today while the benefits – in terms of higher incomes – don’t kick in until the child joins the labor force in 15 or so years’ time. We also account for the fact that countries start from different coverage rates of the 10 interventions, which need to scaled up to 90% coverage. We estimate that for every dollar spent, the program achieves $15 in benefits, assuming a discount rate of 5%, and its rate-of-return is 17%.
These figures likely underestimate the social returns, as they ignore the possible benefits to society as a whole from having more educated workers and mothers. Our estimate is still reasonably high even if we make more conservative assumptions. In the ultra-pessimistic case, where we double the program cost to almost $10 per capita, halve the program’s effect on stunting to 10%, and halve the assumed impact of stunting on education, height and cognition, we come up with a rate-of-return estimate of around 10%.
What else beyond direct nutrition-specific interventions?
Scaling the nutrition package up to 90% coverage – no mean feat – reduces stunting by only 20%, well short of the 40% reduction called for in the SDGs. Even if we assume on top of this the recent downward trend in stunting continues, we still fall short.
Fortunately, there are other weapons available. Recently there have been attempts to synthesize the evidence of the effects of these so-called ‘nutrition-sensitive’ interventions that address the underlying determinants of nutrition. They include conditional cash transfer (CCT) schemes that improve child nutrition by providing households with additional resources to purchase more and more nutritious food, and by encouraging growth monitoring and pre- and post-natal consultations. They include agricultural and livelihood interventions that enhance the intake of micronutrient-rich and animal-source food. Water and sanitation-related interventions, including infrastructure investments as well as interventions that promote behavior change like handwashing, can also help by reducing the disease environment to which children are exposed. Programs that promote better parenting, and increase early stimulation and learning, not only lower the long-term cost of stunting but also increase cognitive and socioemotional development even beyond the first two years of life.
Known knowns, and known unknowns
We know children like Alejandro who start life stunted will suffer an income penalty over their lives, and therefore a country’s GDP today is smaller today to the extent it failed to eliminate childhood stunting when its current workers were children. We know that even though scaling up direct nutrition interventions to 90% coverage would reduce stunting by only 20% or so, the costs are sufficiently small, and the effects on income sufficiently large, for this investment to have a healthy rate-of-return of the order of 17%. And we know that several complementary nutrition-sensitive interventions are available to help us get to the 40% stunting reduction called for by the SDGs.
What we don’t yet know is how much stunting would be reduced by if each were scaled up, and how much this scaling-up would cost. A rate-of-return analysis of each of these interventions would have to factor in that each has benefits to society that go well beyond their nutrition impacts: getting children into school and keeping them there; improving adult nutrition; reducing the cost of getting drinking water; and increased cognitive and socioemotional development. Moreover, the returns to each intervention is likely to depend on the scale of the others. Not a straightforward exercise, then, but one that would help guide policymakers ensure that more children start life like María and fewer like Alejandro, and avoid the large per capita income losses associated with childhood stunting.
Feature Story: http://www.worldbank.org/en/news/feature/2017/03/30/well-designed-early-childhood-development-programs-can-pay-big-dividends.
Video: http://www.worldbank.org/en/news/video/2016/10/25/policy-research-talk-effective-investments-in-the-early-years
Since they’re both past their second birthday, the chances are that María has escaped stunting but Alejandro will stay stunted. His life chances are much worse than his younger sister’s. Alejandro’s extreme shortness was caused by a mix of cumulative exposure to infections and severe undernourishment in utero and infancy. These together will have affected the development not just of his body but also of his brain. María will overtake him cognitive and socioemotional skills and stay ahead. Alejandro will do worse at school and leave school earlier. He will earn less in adulthood – in part because of his less developed cognitive skills but also because of his shorter stature. He is also at greater risk of non-communicable diseases in later life.
The aggregate stunting income penalty
Moving beyond María and Alejandro to other stunted and non-stunted children in their country, we can see how entire countries pay a penalty – in terms of a lower per capita lower income – for not addressing stunting in early childhood. Several studies have tried to quantify this aggregate penalty by looking directly at the association between stunting and adult height, and height on earnings, or by using data linking stunting in early childhood to earnings in adulthood from a seminal small-scale efficacy study conducted in 4 impoverished villages in rural Guatemala.
In a policy research note that came out today, we and our co-authors assembled a broader evidence base to re-estimate the aggregate income penalty that countries incur from childhood stunting. We calculate the effects of childhood stunting on lifetime income operating through all potential pathways: reduced schooling (‑1.6 years on average), reduced height in adulthood (‑6cm on average), and lower cognitive skills (‑0.6 standard deviations on a typical test). For each of these pathways, we translate the deficit into the lower labor market returns – and hence lost adult income – that a stunted child incurs when she join the labor force.
These numbers allow us to put a number on the following thought experiment: how much higher would a country’s GDP per capita be today if none of its current workers had been stunted in childhood? The answer obviously depends in part on the prevalence of stunting when today’s workers were children, which depends in part on the median age of today’s workers. In the case of Chile, the rate of stunting when today’s median-age worker was a child was just 8%. In Bangladesh, it was 70%.
We adopt a method used in the growth literature called ‘development accounting’ to combine these stunting rates with the income penalties in terms of schooling, cognition and height incurred by a worker who was stunted in childhood. We find that the average country’s GDP per capita is 7% lower than it would have been if none of its current workers had been stunted in childhood. In Africa and S Asia, the average is even higher (9-10%).
But what are the returns to interventions that reduce stunting?
The potential economic benefits then of interventions to reduce stunting are sizable, even if they occur many years in the future. Realizing these benefits requires interventions that reduce stunting. The more effective they are in reducing stunting and the cheaper they are, the larger will be their economic returns.
There is one set of interventions for which researchers have estimated both the per capita costs and the overall impacts on stunting prevalence. It is a package of 10 key ‘nutrition-specific’ interventions reaching pregnant women and children in the first 1,000 days, starting at conception, including things like micronutrient supplementation, energy protein and calcium supplementation in pregnancy, nutrition education, complementary feeding, and breastfeeding promotion. Researchers estimate that if these interventions were scaled up across 34 high-stunting countries from current levels to 90%, stunting rates would fall by 20%.
This isn’t an especially large impact, but as we have seen above, even a small change in stunting is associated with large changes in income. Moreover, these interventions might have impacts on other key child outcomes that are also part of the Sustainable Development Goals (SDGs), such as child survival, and their combined cost is not that large (less than $5 per capita).
Putting these relatively low program cost figures together with our earlier estimates of the long-term income gains from reducing stunting, we come up with a benefit-cost ratio and a rate-of-return for this nutrition package. We factor in that the costs are incurred today while the benefits – in terms of higher incomes – don’t kick in until the child joins the labor force in 15 or so years’ time. We also account for the fact that countries start from different coverage rates of the 10 interventions, which need to scaled up to 90% coverage. We estimate that for every dollar spent, the program achieves $15 in benefits, assuming a discount rate of 5%, and its rate-of-return is 17%.
These figures likely underestimate the social returns, as they ignore the possible benefits to society as a whole from having more educated workers and mothers. Our estimate is still reasonably high even if we make more conservative assumptions. In the ultra-pessimistic case, where we double the program cost to almost $10 per capita, halve the program’s effect on stunting to 10%, and halve the assumed impact of stunting on education, height and cognition, we come up with a rate-of-return estimate of around 10%.
What else beyond direct nutrition-specific interventions?
Scaling the nutrition package up to 90% coverage – no mean feat – reduces stunting by only 20%, well short of the 40% reduction called for in the SDGs. Even if we assume on top of this the recent downward trend in stunting continues, we still fall short.
Fortunately, there are other weapons available. Recently there have been attempts to synthesize the evidence of the effects of these so-called ‘nutrition-sensitive’ interventions that address the underlying determinants of nutrition. They include conditional cash transfer (CCT) schemes that improve child nutrition by providing households with additional resources to purchase more and more nutritious food, and by encouraging growth monitoring and pre- and post-natal consultations. They include agricultural and livelihood interventions that enhance the intake of micronutrient-rich and animal-source food. Water and sanitation-related interventions, including infrastructure investments as well as interventions that promote behavior change like handwashing, can also help by reducing the disease environment to which children are exposed. Programs that promote better parenting, and increase early stimulation and learning, not only lower the long-term cost of stunting but also increase cognitive and socioemotional development even beyond the first two years of life.
Known knowns, and known unknowns
We know children like Alejandro who start life stunted will suffer an income penalty over their lives, and therefore a country’s GDP today is smaller today to the extent it failed to eliminate childhood stunting when its current workers were children. We know that even though scaling up direct nutrition interventions to 90% coverage would reduce stunting by only 20% or so, the costs are sufficiently small, and the effects on income sufficiently large, for this investment to have a healthy rate-of-return of the order of 17%. And we know that several complementary nutrition-sensitive interventions are available to help us get to the 40% stunting reduction called for by the SDGs.
What we don’t yet know is how much stunting would be reduced by if each were scaled up, and how much this scaling-up would cost. A rate-of-return analysis of each of these interventions would have to factor in that each has benefits to society that go well beyond their nutrition impacts: getting children into school and keeping them there; improving adult nutrition; reducing the cost of getting drinking water; and increased cognitive and socioemotional development. Moreover, the returns to each intervention is likely to depend on the scale of the others. Not a straightforward exercise, then, but one that would help guide policymakers ensure that more children start life like María and fewer like Alejandro, and avoid the large per capita income losses associated with childhood stunting.
Feature Story: http://www.worldbank.org/en/news/feature/2017/03/30/well-designed-early-childhood-development-programs-can-pay-big-dividends.
Video: http://www.worldbank.org/en/news/video/2016/10/25/policy-research-talk-effective-investments-in-the-early-years
Join the Conversation