A Profile of the Background of India’s Young by Raghbendra Jha

When the results of the 2011 Census of India were announced two factors were most noticed: (i) a reduction in the total fertility rate from 2.9 in 2001 to 2.62 in 2011, and (ii) deterioration in the gender balance, i.e., the number of girls per 1000 boys between the censuses of 2001 and 2011.  For 0-4, 5-9, and 0-6 year olds this fell from 939 to 891, 920 to 889, and 927 to respectively.  

While the first effect is usually taken as an indicator of the demographic transition associated with rising per capita incomes, the second is often cited as evidence of widespread gender bias in the Indian population.  Considerable evidence (Jha et.al. 2011) exists of sex selection tests and follow-up abortions if the fetus is found to be female. However, since the Census does not include household level characteristics, identifying such characteristics that increase the chance of feticide is difficult with this database.  Chaudhri and Jha (2013) used household level data for the National Sample Survey (NSS) rounds of 1993-94 and 2004-05 to identify characteristics of households determining the number of girl children relative to boys and found, ironically enough, that higher education of mothers and higher prosperity are each associated with increasing gender bias.  Only when the product of the two reaches a relatively high level does gender bias start coming down. 

It behooves us to ask the complementary question: what are the determinants of the number of children in the household, i.e., the household’s fertility?  For the same data set as Chaudhri and Jha (2013) I identify the determinants of the number of children (aged 0-14) at the household level. This helps us build up a profile of the background of India’s young: from those aged 9 (born in 2004-05) to those who are 34 (those who were 14 in 1993-94). 

My econometric analysis reveals the following. In both years (1993-94 and 2004-05) the number of children is significantly higher the greater the number of females in the childbearing age (15-49) in the household.  Indeed this effect peaks with the age group 26-35 of potential mothers and is lower for younger women (15-25) and older women (36-49).   The (instrumented) number of girls (5-14) not in school is also a strongly significant determinant of the number of children.  Average education of adult females in the household, computed as the total number of years of schooling of these women divided by the number of such women, lowers the number of children in the household.  Similarly, the higher the log of Monthly Per Capita Income  (MPCE) (figures for 2004-05 are deflated to make them comparable with 1993-94 figures) the lower is the number of children. (Instrumented) shares of health and education in the household budget have a positive impact.  Scheduled Caste households have fewer children in both years whereas Scheduled Tribe households had fewer children in 1993-94 but more children in 2004-05.  Muslim households had more children in both years with the coefficients for this dummy weighty and strongly significant for both years.  Households in the BIMARU states (Bihar, Madhya Pradesh, Rajasthan and Uttar Pradesh) had more children than those in non-BIMARU states whereas rural households had fewer children than urban households.

It is interesting also to note the differences in coefficients across the two time periods. Compared to 1993-94 women in the child-bearing ages had fewer children in 2004-05 with the drop being highest for women in the age group 26-35.  The impact of (instrumented) number of girls not in school went up over time as did that of the (instrumented) shares of education and health expenditures.  Higher MPCE led to a further drop in the number of children. SC and ST households had significantly higher number of children as did households in the BIMARU states.  The change in the number of children in Muslim households was insignificant.   The impact of average education of females went up over time. 

Thus, growing incomes, social and household decision factors of households and regional factors help determine the patterns of household level fertility in India. In this particular case they help determine the demographic composition of India’s youth (aged 9 to 34) today. The drop in India’s total fertility rate (TFR) from 6.1 in 1961 to 2.62 in 2011 masks widespread variation in TFR among various groups. Demographic transition is well underway in India with rising incomes associated with fewer children and smaller family size.   The number of women in the child-bearing age group has a significant effect on the number of children.  This effect is particularly strong for women aged 26-35.  Higher average education of females lowers household size whereas (instrumented) shares of expenditure on education and health have varying effects.  The impact of a household being SC or ST varies by year and by the regression model chosen. Over both time periods Muslim households have more children and are more likely than the general population to have larger family sizes.  Households in BIMARU states have more children and have larger family sizes as do urban households.  The full demographic transition is, therefore, yet to set in over Muslim households and those in BIMARU states.



Further Reading

Chaudhri, D.P. and R. Jha (2013) “India’s Gender Bias in Child Population, female education and growing prosperity” International Review of Applied Economics, vol. 27, no.1, pp. 23-43. 

Jha, R. (2013) ‘The determinants of household level fertility in India’ mimeo ASARC,

            Australian National University, Canberra.


Jha, P., Kesler, M., Kumar, R., Ram, F., Ram, U., Aleksandrowicz, L., Bassani, D., 

Chandra, S. and J. Banthia (2011) “Trends in selective abortions of girls in India: analysis of nationally representative birth histories from 1990 to 2005 and census data from 1991 to 2011” Lancet, vol. 377, Issue 9781,


More posts by this author:

Please follow and like us:

Co Authors :

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.