Update on Trends in Sex Ratio at Birth in India
With additional data available at more regular intervals on sex ratio at birth, it is easier to track changes without depending on the decennial censuses. However, the diversity of sources also means differences in trends across sources and regions. A brief analysis of the latest available data is presented to identify the most recent trends and concerns in states that need attention, like north-western and eastern states of India.
In Rajan et al (2015) we analysed changes in the 0–1 year sex ratio (sex ratio at birth or SRB) and the 0–6 year sex ratio in India. This analysis was based on census data collected in 2001 and 2011 which contains information on about 20 to 21 million live births which occurred in the years preceding the censuses. Our analysis showed that the SRB declined over the 10-year period from 905 (female births per 1,000 male births) to 899 and that there has been no let-up in daughter deficit. A striking aspect of the analysis was that in the north-western states, which have a long history of high levels of daughter deficit, there was an increase in the SRB between 2001 and 2011 while southern and eastern states experienced declines between the two censuses.
Sources of Data
Since the release of Census 2011 data on SRB, a number of other available sources yield more recent information on SRB. These include various editions of the Sample Registration System (SRS), the fourth edition of the National Family Health Survey (NFHS) and information from the Civil Registration System (CRS). Information on the SRB contained in these sources is presented in Table 1 (p 15). The table contains information for India and its major states.
As a benchmark, columns 2 and 3 present SRB from the 2001 and 2011 Censuses. NFHS-based SRB is presented in columns 4 and 5 for 2005–06 (NFHS–3) and 2015–16 (NFHS–4). The NFHS provides information on SRB based on children born in the five years preceding the survey.1 At the time of writing this note, the NFHS–4 data were not available for all states and union territories, and so there are no all-India figures. Columns 6, 7 and 8 contain information from the SRS on SRB, based on three-year moving averages, for the periods 2010–12, 2011–13, and 2012–14. Columns 9, 10, 11 and 12 present SRB from the CRS for 2011, 2012, 2013 and 2014, and columns 13 to 16 present the extent (percentages) of birth registration annually for 2011 to 2014. Despite the missing information, what is the picture emerging from these different sources?
At the all-India level, in 2011 the census—which is based on about 21 million births—yields an SRB of 899; the SRS which is based on about 1,57,626 births per year yields an SRB of 908 (with a 95% confidence interval of 899 to 917) for the period 2010–12;2 and the CRS which is based on about 21.8 million births (with an estimated registration of 83.6% of all births) yields an SRB of 909 in 2011. While the SRB figures are not entirely consistent across data sources, they lie in a relatively narrow range. The SRS-based SRB declined from 908 in 2010–12 to 906 in 2012–14. However, given the small number of births on which this statistic is based and the wide confidence interval, essentially there is no change in the SRB over time based on the SRS data. Even if we work with a longer time span, say 2007–09 till 2012–14, it does not alter the conclusion.
In contrast, the CRS data—which is based on increasing coverage of registered births from 83.6% in 2011 to 88.8% in 2014—shows a sharp decline in SRB from 909 in 2011 to 887 in 2014. While the numbers from the SRS and the CRS are not directly comparable, as the SRS is based on a three-year average, and the CRS on annual data, the contrast is striking. While it is possible that there is a systematic tendency to not register female births as compared to male births, unless underreporting of female births has increased over time, there is no reason to expect that the temporal changes in the SRB figures based on the CRS are driven by underreporting.
What about state-specific patterns? In Rajan et al (2015) based on Censuses 2001 and 2011 we found that the bulk of the north-western states (Himachal Pradesh, Punjab, Chandigarh, Uttarakhand, Haryana, Delhi and Rajasthan) experienced an increase in the SRB between 2001 and 2011 while states in the eastern region (Bihar, Sikkim, Nagaland, Mizoram, Tripura, Meghalaya, Assam, West Bengal, Jharkhand and Odisha), that is, traditionally a region of the country with limited son preference, the SRB declined between 2001 and 2011.
An examination of state-level SRB based on SRS data over the years 2010–12 and 2012–14 shows that the changes are relatively small, ranging from a decline of eight points in Delhi to an increase of nine points in Haryana. Given the short time period under consideration, this is perhaps not unexpected. Examination of the data over a longer period, that is, between 2007–09 and 2012–14—although the pattern is not as clear as in the census—does show an increase in SRB in the north-western states (Gujarat, Haryana, Punjab and Rajasthan) and a decline in the ratio in some of the eastern states (Assam and Bihar). Analysis of SRB based on NFHS data mirror some of the patterns found in the census. The NFHS data show an increase in the SRB in north-western states (Haryana, Punjab and Rajasthan) and a decline in the SRB in the eastern states (Assam, Odisha and West Bengal).
In contrast to the relatively stable picture emerging from the SRS, the CRS shows that between 2011 and 2014 several states—even if we focus only on those states that have an (estimated) birth registration rate of more than 80%—have witnessed a sharp decline in SRB. For instance, Gujarat, Himachal Pradesh, Karnataka, West Bengal and Tamil Nadu have recorded a decline in SRB in the post-Census 2011 period. While not unusual, the source-dependent difference in SRB for Tamil Nadu, for example, is perturbing. According to the CRS, Tamil Nadu has experienced a decline in SRB from 905 to 834 between 2011 and 2014 on a 100% birth registration rate between 2011 and 2014. However, other sources such as the SRS peg Tamil Nadu’s SRB at 929 in 2007–09 and at 921 in 2012–14 in contrast to the NFHS figures of 896 in 2004–05 and 954 in 2015–16.
While it is a welcome change to have a large number of data sources that offer the SRB as opposed to a situation where one has to wait for 10 years to obtain figures on sex ratios, these different sources need to be used carefully. The census and the CRS have the advantage of working with a large number of births. However, in a number of large states (Bihar and Uttar Pradesh) the share of registered births remains well below 80%. Furthermore, if households systematically do not register female births, then the figures based on the CRS are likely to exaggerate the female deficit. Surveys such as the SRS and NFHS do away with the problems of birth registration, but the number of births on which they are based, especially the state-level figures, is just too small to be reliable, that is, the SRB has wide confidence intervals. At the moment, there is perhaps no data source that may be considered uniformly (for all states) superior in terms of computing the SRB.
Indeed, the differing picture emerging from the various data sources invites cherry-picking. This is unwise. Given the continuous increase in the level of birth registration, the large number of births on which the CRS is based, and its far more regular availability, it is important to continue strengthening the CRS (see also Rajan and Mohanachandran 1999) so that important policy-relevant information may be readily and reliably gleaned.
1 NFHS–3 was based on interviews with 1,24,385 women in the age group 15–49 years and 74,369 men in the age group 15–54 years and about 55,000 births in the five years preceding the survey. NFHS–4 is based on substantially larger numbers and interviews were conducted with 6,28,826 women and 94,324 men and the statistics are based on 2,65,653 children born over five years prior to the survey (IIPS 2016).
2 In 2014, the SRS covered a population of 7.5 million and the birth rate was 21.0 which translate into about 1,57,626 live births (Office of Registrar General 2016a).
IIPS (2007): National Family Health Survey (NFHS 3), 2005–06, India, Volume 1, Mumbai: International Institute for Population Sciences.
— (2016): National Family Health Survey (NFHS 4), 2015–16, State Fact Sheets, Mumbai: International Institute for Population Sciences.
Kishor, Sunita and Kamla Gupta (2009): Gender Equality and Women’s Empowerment in India, National Family Health Survey (NFHS-3), India, 2005–06. Mumbai: International Institute for Population Sciences; Calverton, Maryland, USA: ICF Macro.
Office of the Registrar General (2012): Census 2011, Government of India, viewed on 4 April 2011, http://censusindia.gov.in/2011-prov-results/census2011_PPT_paper1.html.
— (2014): SRS Statistical Report Detailed Tables, New Delhi, viewed on 9 December 2016, http://www.censusindia.gov.in/vital_statistics/SRS_Report_2014/9.%20SRS%....
— (2016a): SRS Bulletin, Vol 50, No 1, July, New Delhi.
— (2016b): Vital Statistics of India Based on the Civil Registration System 2014, New Delhi.
Rajan, Irudaya S and P Mohanachandran (1999): “Estimating Infant Mortality in Kerala,” Economic & Political Weekly, Vol 34, No 12, pp 713–16.
Rajan, Irudaya S, Sharada Srinivasan and Arjun S Bedi (2015): “Coming Back to Normal? Census 2011 and Sex Ratios in India,” Economic & Political Weekly, Vol 50, No 52, pp 33–36.
EPW looks forward to your comments. Please note that comments are moderated as per our comments policy. They may take some time to appear. A comment, if suitable, may be selected for publication in the Letters pages of EPW.