Childlessness in India
Koyel Sarkar, Université catholique de Louvain
Thomas Baudin, IESEG Business School
We propose a new interpretation of thedynamics of childlessness in India over the last decades. Grounding on therecent decomposition of childlessness proposed by Baudin et al. (2016),we use micro level data to assert that a significant part of childlessness amongIndian couples is explained by the emergence of better educational and economicopportunities offered to women. This phenomenon exhibits a clear geographicalheterogeneity without contradicting the fact that many Indian women arechildless at the end of their reproductive cycle because of primary infertilityrelated to sterility, venereal diseases and poverty.
Childlessness is defined by the absence of any living children ina womans life; when a woman remains childless after the age of 40, it isusually called definitive childlessness. The latest census, conducted in 2011,has recorded the highest ever definitive childlessness rates in India: 7.89percent among women above 40 years of age. Although the rate is only aroundhalf of that measured in the US for the year 2014, it is nevertheless far fromnatural sterility rates. More importantly, given the Indian population size, inabsolute numbers, definitive childlessness affects more than twelve millionIndian women, far more than the 1.6 millions of definitively childless Americanwomen.
Using micro-level regressions, we show that the probability to endreproductive life without children exhibits a clear U-shaped relationship witheducational attainment of women. This is indicative that poverty (proxied bylow educational attainment) and sterility are not the unique gradients ofchildlessness, better economic opportunities and empowerment inside couples(proxied by high educational attainment) also determine the probability to be childless.We show that this result is robust to the introduction of important controlvariables such as the level of development of the state where women are living,education of the husband, age at marriage, religion and caste.
1.2. Data and Methodology:
We use secondary data from all the threerounds of District Level Household and Facility Survey (DLHS, 1998-99, 2004-05and 2007-08). We also complement the study using the Indian Census Data 2011for mapping childlessness at regional levels. We have a final sample of 158,112after successive filtering of women born between 1953 and 1968, who have beengrouped into cohorts of births, (we want to prevent any selection bias due tocohort-based mortality after 50). In our regression models, we study thedeterminants of the probability to end reproductive life as childless, forwhich we use information about completed fertility to build the variable childlessnesswhich is dichotomous. It takes value 1 if the respondent has no kids and 0otherwise. We consider two kinds of fixed effects, the first is cohortfixed-effect and the second is state fixed-effect. Eight cohorts have beencompiled for the study, the oldest being women born in 1953-54 and the youngestbeing women born in 1967-68. All the 35 states and union territories in Indiahave been considered under the state-fixed effects in the models.
1.3. Descriptive Findings:
As we see above at the country level (figure top left), childlessnessexhibits a U-shaped relationship with years of schooling. It indicates that inIndia above an education threshold (around 9 or 11 years of schooling), withincrease in years of education, childlessness among women tends to increase. Usingstate level data we show (figure top centre) that the correlation between statesaverage childlessness rates and states average fertility of women is stronglynegative. Again, when we consider the major states in India (figure top right),the correlation between states average education and states averagechildlessness is clearly positive. This suggests that the states having highlevels of education (supposedly the most developed ones) are also the oneshaving higher childlessness, and vice versa. These macro level findings favourof our hypothesis that India is experiencing a childlessness transition whichgoes beyond the simple explanation by poverty or infertility driven. We findsimilar regional patterns when mapped district level childlessness witheducation categories from the Census 2011 data.
1.4. Micro Level Analysis:
Our study goes well beyond deliveringsummary statistics and offer regression results (table below) for theprobability to end reproductive life childless. We evidence a U-shaperelationship between this probability and the educational attainment of women,which is robust to the introduction of economic and social covariates, culturalvariables, marriage history and environmental variables. This is a sign thatboth poverty driven and opportunity driven childlessness co-exist in India,which is indicative that this co-existence is valid in all cultural and social groupsof India, and is not confounded by matrimonial history or caste and religious partitionsof the country. Thus we show that childlessness does not concern poor anduneducated women only, many childless Indian women are highly educated.
Table: Determinants of Childlessness:
Notes: Odds-ratio reported. *** p<0.01, ** p<0.05, * p<0.1
Presented in Session 1234: Sexual and Reproductive Behaviour