Way back in 2019, we blogged about the “meaning of stunting”, reflecting on some thought-provoking papers recently out that re-examined the meaning of stunting and sparked discussion for those of us who study human development.
Here we turn our attention to a new NBER working paper by Weaver, Sukhtankar, Niehaus, and Muralidharan (2024) which examines income (specifically cash transfers), nutrition, and child development in the Indian state of Jharkhand. The starting point for this paper is the rate of stunting Indian children under 5 (36%)—stubbornly high, and higher than many poorer countries. This paper hits on a number of interesting (and interrelated) points. But first, a short summary of this study.
Weaver and co-authors study a cash transfer program targeting very poor mothers with very young children (under 2 years) in Jharkhand. The program operated through existing integrated children development centers (“anganwadi centers”, or AWCs), replacing a status-quo approach of offering take-home rations. The program randomly selected new mothers to receive monthly transfers to her bank account. Mothers without an existing bank account were given assistance to open one. The intervention can be characterized as a “framed” cash transfer: beneficiaries also received messages encouraging them to spend the transfers on nutritious food for the mother and child through monthly recorded automated calls (IVR) calls (calls that also informed them that a transfer had just been sent to their accounts). Both treatment and control mothers were given messaging on flyers and verbally by AWC workers, so the messaging intervention was light-touch (moreover only about half received messages, on average, per month due to poor connectivity). Overall, this study is ambitious – focusing on a government intervention at scale (and the implementation challenges that go with that), as well as in-depth and multiple rounds of data (including detailed consumption tracking with measures of weight and volume of ingredients in past 24 hours).
They show that the program increased both the quantity of calories and the quality of diets of mothers and children. Still, there were no significant improvements in children’s anthropometrics, on average. This is consistent with systematic reviews which find no (or very modest, if any) impacts of cash transfers on anthropometrics (Manley et al 2020, for example find effects in 27 out of 46 studies, with a 2.1% average decline in the rate of stunting). Though their point estimates generally lie within range of the significant effects reported in studies of cash plus behavior change interventions in Bangladesh, Myanmar, and Nigeria). However, in areas with low rates of open defecation, anthropometrics improved by 0.12 to 0.16 SDs for height and weight for age z-scores, respectively.
Meanwhile, there were positive impacts on child development outcomes as measured by Ages and Stages (which captures both cognitive and motor skills), of similar magnitude (0.12 SD improvement at age 3). Lastly, the transfers crowded-in early childhood services, including increases in year 1 of vaccination rates, uptake of iron/calcium tablets, receipt of nutrition knowledge by mothers, and uptake of deworming medication, which, by themselves, might have a direct effect on the health and nutrition of the target population.
Here we highlight three points – among many others this thoughtful paper brings to mind.
On the complementarities between cash and sanitation
Cash is not sufficient to deliver improvements in stunting in this setting where children are exposed to infections (in this case, in the form of open defecation). This is an intuitive finding which likely contributes to the meta finding of little to no stunting effects found.
Protection from infection is, of course, broader than reducing open defecation. Different environments require different WASH (water, sanitation and hygiene) interventions. The nature of the disease environment is context specific, and stand-alone interventions that are not universal might have limited impact. Current large trials of WASH interventions (Wash benefits/SHINE, where they tested the role of nutrition, water, sanitation provided to households or small compounds in a factorial design) failed to yield a significant impact on anthropometric outcomes. Rather, the conclusion reached in this consensus statement was the need for greater (universal) coverage/population level interventions to break the re-transmission of pathogens. Can we test this? Appendix B in Weaver et al lays out power calculations to show that an ideal test in their setting of the sanitation and income interaction would not be feasible. Luckily, non-experimental methods of population-level interventions (see Duflo et al 2015) or structural estimates (Abramovsky et al 2019) are alternative ways to uncover these interactions.
On the complementarities between cash and messaging for behavioral change
Lots of literature points to the importance of behavioral change for improved child nutrition, suggesting that cash is not sufficient to deliver improvements. Here they do not find evidence that the light-touch additional messaging yielded better results. The less resource-intensive approach tested in Jharkhand was something feasible at scale in a setting with weak state capacity and overworked community health workers. While not impactful, it should not distract from the importance of messaging and counseling for behavioral change (as noted in the findings from Bangladesh, Myanmar, and Nigeria noted above). We still need to find impactful and feasible approaches. Improved knowledge (‘CARE’ in the WHO concept of nurturing care framework) about proper feeding and hygiene practices, the importance of breastfeeding, having a diversified and nutrient dense diet – is needed. For example, in India, the fraction of children 6-23 who have a minimum acceptable diet (defined as consuming 5 out of 8 food groups) is a dismal 11% (SOWC 2023), and almost one in five children 6-23 months are “zero food” children – children who did not receive any animal milk, semi-solid or solid food in the last 24 hours (Karlsson, Kim, and Subramanian 2024).)
The rationale of bundling cash with information and behavioral change is compelling and intuitive: cash transfers help poor households to act on knowledge, in line with other studies in the literature. It does seem we have turned a corner 10 years after Alderman 2014 noted that the “potential of transfer programs to be nutrition sensitive remains largely untapped.” But the ‘how’ to do it at scale still remains a critical challenge.
On the wide expanse of child development and timing of investment during the early ages
In our previous blog we asked whether we are over-stating the extent to which stunting is a good stand-in measure for human capital, and, if so, do we then bias policy direction away from potentially effective interventions. This study reminds us of the importance of viewing anthropometrics as one of several markers to capture some domains of child development, as noted by the authors. Beyond resources, inputs such as proper nutrients, protection from infections, and positive and sensitive caregiving are needed to reach the children full potentials, with different timing of critical and sensitive windows.
Another challenge of interventions is the different forms of malnutrition and their sensitive age patterns. Wasting, for example, is a very dynamic process, with an incidence that peaks between birth and three months of age [Mertens et al 2023]. And children do not start on a level playing field at birth. Children who are premature or are small for their gestational age (a notable phenomenon in India, in stark contrast with many Sub-Saharan countries) start with a significant disadvantage, with higher mortality risk, and the children who survive are more vulnerable to infections and are at a higher risk of stunting. In short, we need to start early to yield the potential protective effects of cash and health/nutrition interventions on both wasting and stunting (if any). This entails starting before birth -- including focusing on women’s health and nutrition status, with the objective of preventing wasting or faltering, early.
The program studied here rightly recruits pregnant women in the AWC catchment area, but starts the cash transfers shortly after birth, missing a key window of vulnerability of time in utero, especially in the context of the heightened vulnerability of the nutritional profile in India. Future work should ideally look at the integration and sequencing of cash and direct health and nutrition interventions for the entire window of the first 1,000 days (e.g. looking at the combined impact of the other Indian programs PMMVY and JSY with the current intervention). And ideally work to find new innovative designs that have strong complementarities with WASH and nutrition interventions at scale.
Join the Conversation