Human Development

Sustained health benefits of ICDS: Later life evidence

  • Blog Post Date 26 July, 2019
  • Print Page
Author Image

Gaurav Dhamija

Indian Statistical Insitute, Delhi Centre; Shiv Nadar University

Child malnutrition is a serious concern for India where more than half of the children under age five are moderately or severely malnourished. The Integrated Child Development Services (ICDS) launched in 1975, target long-term nutrition and holistic development of children. Analysing data from the India Human Development Survey-II, this article evaluates the causal impact of ICDS exposure in rural areas on children’s health outcomes in later years of their lives.

Malnutrition or poor health in the initial few years of a child’s life has a strong association with impaired growth through poor lifetime health (Dhamija and Sen 2018, Victora et al. 2008), lower ability to learn (Morgane et al. 1993), lower schooling attainment (Alderman et al. 2006), poor employment status (Jurges 2013), and lesser earnings possibilities (Bleakley 2010) in the long run. The strong associations between early life health, nutrition status, and later life outcomes raise serious concerns for a developing country like India where more than 110 million children are under the age of five years, and 38% and 36% of them are reported to be stunted and underweight respectively (International Institute for Population Sciences (IIPS) and Inner City Fund (ICF), 2017).

The Indian government launched the Integrated Child Development Services (ICDS) in 1975. It is the largest national programme in the world which targets long-term nutrition and holistic development of children by providing a range of services in one platform known as Anganwadi (or childcare) centres (henceforth, AWC). The package of services provided by the ICDS includes supplementary nutrition, pre-school education, immunisation, health check-up, nutrition and health education, and referral services for children aged 0-6 years, adolescent girls, pregnant women, and nursing mothers.

In Dhamija and Sen (2017), we evaluate the causal impact of ICDS exposure through AWC access in rural areas on children’s health outcomes in later years of their lives, when they are not eligible to be the direct beneficiaries of the services anymore.

Data and empirical strategy

We use the second wave of the nationally representative India Human Development Survey (IHDS) data, conducted in the year 2011-12. This survey captures information related to health, education, employment, economic status, marriage, fertility, gender relations, and social capital for surveyed households. Detailed information at the village level is available separately. This village questionnaire, which is available for rural areas only, asks every village head about the availability of AWC in their village. The specific question used is as follows: “Does this village have... Government Anganwadi or other Child Care Centre? How many years ago did it open?”

We combine the village-level information on the construction year of AWC with the birth year of 10-13 year-old adolescents to find the years of exposure of an adolescent to the ICDS services through the AWC in the initial three years of the child’s life. The implementation of the scheme or the ‘treatment’ through setting up of village-level AWC, combined with the exogenous variation of birth year is the basis of our identification strategy. Children born after the setting up of the AWC in the village are expected to gain from the exposure, as compared to the children born before the AWC came in. This variation helps us to devise our identification strategy in the treatment effect framework. This estimation strategy captures variation at both spatial level as well as temporal level. The spatial-level variation is exploited by the variation across villages whereas temporal level variation is exploited by the variation between birth year and construction year of AWC in every village.

This ‘intent-to-treat’ estimate may differ from the ‘average treatment effect’ on the treated if all the children who could access the ICDS services in their village would not have actually used it due to some unobservable reasons. This could be due to the difference in: (i) affordability across households, for example, some households may be able to provide better nutrition and care at the home; (ii) lack of awareness among the marginalised households that can make the malnourished children devoid of these services. In order to cater to these possible differences between target users and the actual users, we estimate our model conditional on several individual level and household-level variables, along with time and village fixed effects1 over different specifications. The outcome variables2 considered are: height (in cm), Z scores3 of height-for-age (ZHFA), weight (in kg), Z scores of weight-for-age (ZWFA), incidence of short-term morbidity (due to fever, cough or diarrhoea) in general, and fever and cough in particular. Incidence of short-term morbidity, in general, and specific ones are binary variables; whereas, all others are continuous variables. In the first model, the primary independent variable4 of interest is ‘years of exposure (YOEXP)’, ranging from 0 to 3 years. In the second model, we restrict the analysis to those adolescents only, who are exposed to ICDS for all three years of their initial life in comparison to the adolescents who are not at all exposed. Therefore, with all other specifications remaining same, the primary variable of our interest in the second model is a binary variable, called ‘Full exposure (FULEXP) ,’ which assumes a value 1 if years of exposure are three and 0 if there are no years of exposure.


Our results indicate a strong positive effect of early life ICDS exposure on health outcomes in later life. Specifically, we find that the exposure to ICDS scheme through AWC access by an extra year seems to increase height by 1.33 cm, ZHFA by 0.14 standard deviation5, and seems to reduce the likelihood of suffering from short-term morbidity and fever by 5 percentage points in each case. In the second model, when we restrict the sample to the children who are either fully exposed for all three years of initial life or not at all exposed, the former children are likely to have higher height (by 6.94 cm), higher ZHFA (by 0.80 standard deviation), more weight (by 3.63 kg), higher ZWFA (by 0.63 standard deviation), and lower likelihood of suffering from short-term morbidity and fever by 17 percentage points. Further, we show that the results are not driven by the cohort of children born in particular birth year. The estimates for weight are not always found to be significant in our study as it can be affected by deficiency in both current as well as past energy status (Sahu et al. 2015); whereas, height is a measure of long-term nutrition and deficiency that is very difficult to overcome beyond a specific age (Hoddinott and Kinsey 2001). Our estimates are robust to different age groups till adolescence, and treatment duration of the initial six years. We also find differential effects for the female cohort as compared to the males. These findings indicate that ICDS exposure does not only have immediate returns as found by Jain (2015) but also long-term returns. In a country like India where more than half of the children under the age of five years are moderately or severely malnourished, it is important to further strengthen the scheme by eliminating the hurdles (Gragnolati et al. 2006) in the successful operation of the scheme across all communities.


  1. Variables that are constant across individuals, like sex or ethnicity, and do not change or change at a constant rate over time.
  2. Outcome variable is the dependent variable which is explained in a regression model and whose value depends on the values of one or more other variables, called the independent or explanatory variables.
  3. The Z score expresses the anthropometric values such as height or weight as a number of standard deviations below or above median value of the reference population.
  4. In regression analysis, a variable that is used to explain variation in the dependent variable.
  5. Standard deviation is a measure that is used to quantify the amount of variation or dispersion of a set of values from the mean value (average) of that set.

Further Reading

  • Alderman, Harold, John Hoddinott and Bill Kinsey (2006), “Long term consequences of early childhood malnutrition”, Oxford Economic Papers,58(3): 450-474.
  • Bleakley, Hoyt (2010), “Malaria eradication in the Americas: A retrospective analysis of childhood exposure”, American Economic Journal: Applied Economics,2(2): 1-45. Available here.
  • Dhamija, G and G Sen (2017), ‘Benefits of Integrated Child Developmental Services: Later Life Evidence’, SSRN Working Paper. Available here.
  • Dhamija, G and G Sen (2018), ‘Lasting Impact on Health Through the Lenses of Severity in Earthquake’, SSRN Working Paper. Available here.
  • Gragnolati, M, C Bredenkamp, M Shekar, M Das Gupta and Y Lee (2006), India's undernourished children: a call for reform and action, The World Bank.
  • International Institute for Population Sciences (IIPS) and ICF (2017), ‘National Family Health Survey (NFHS-4), 2015–16: India’.
  • Jain, Monica (2015), “India’s Struggle Against Malnutrition—Is the ICDS Program the Answer?”, World Development,67: 72-89.
  • Jürges, Hendrik (2013), “Collateral damage: The German food crisis, educational attainment and labor market outcomes of German post-war cohorts”, Journal of Health Economics,32(1): 286-303.
  • Hoddinott, John and Bill Kinsey (2001), “Child growth in the time of drought”, Oxford Bulletin of Economics and statistics, 63(4): 409-436. Available here.
  • Morgane, Peter J, Robert Austin-LaFrance, Joseph Bronzino, John Tonkiss, Sofia Diaz-Cintra, L Cintra, Tom Kemper and Janina R Galler (1993), “Prenatal malnutrition and development of the brain’, Neuroscience & Biobehavioral Reviews,17(1): 91-128.
  • Sahu, Swaroop Kumar, S Ganesh Kumar, B Vishnu Bhat, KC Premarajan, Sonali Sarkar, Gautam Roy and Nitin Joseph (2015), “Malnutrition among under-five children in India and strategies for control”, Journal of natural science, biology, and medicine, 6(1): 18. Available here.
  • Victora, Cesar G, Linda Adair, Caroline Fall, Pedro C Hallal, Reynaldo Martorell, Linda Richter and Harshpal Singh Sachdev (2008), “Maternal and child undernutrition: consequences for adult health and human capital”, The Lancet, 371(9609): 340-357. Available here.
No comments yet
Join the conversation
Captcha Captcha Reload

Comments will be held for moderation. Your contact information will not be made public.

Related content

Sign up to our newsletter