In our previous article The arrival of the girl child we had presented evidence to suggest that the decline in the sex ratio at birth may have been arrested. Further, that amongst the many issues facing the Indian economy and polity, gender imbalance is a problem that may be on its way to getting resolved.

This optimistic conclusion has turned out to be controversial. As expected. For the question of missing girls has haunted many sections of academia, civil society and government during the last 2 decades. The child (0-6 years) sex ratio (CSR) had dropped alarmingly, especially in north-west India. Large-scale interventions to have a son (read murder of the girl child) had meant that approximately 4-5% of all boys born each year were extra. There was hope that all the efforts made to reverse the dismal decline in daughters would bear fruit and positive results would be seen in the 2011 Census. This hope was based on evidence from the Sample Registration System (SRS) that the sex ratio at birth had begun to improve since its trough in 2004. (The SRS, a division of the census, collects data on the sex ratio at birth and publishes these results for an average of 3 years. The child sex ratio is inferred by the census?and all household surveys?from the age and sex distribution of the surveyed population.)

These hopes were dashed when Census 2011 revealed that the CSR had declined from 927 in 2001 to 914 in 2011. Note that the biological norm is 950 girls born for every 1,000 boys born. However, while important for research and policy, this decline tells us relatively little about recent changes in the sex ratio at birth (SRB). Census 2011 has yet to release the information on a closely related variable, the sex ratio for children of zero years of age (0-364 days, SR0). The next census is not due for another 10 years.

There is considerable evidence that the pace of change in India has accelerated in the last decade. The census CSR is an average of the SRB for children born since February 2004. Since then, per capita income in the country has increased by close to 60%. There are alternative sources of information for the sex ratio?namely, the SRS and the periodic data compiled by the National Fertility Health Surveys (NFHS). The last information published by the SRS is for 2006-08; the last NFHS survey is for 2005-06. The NFHS is much relied upon by social scientists for its trustworthiness.

There is yet another source of information on the sex ratio?the NSS. These surveys are used for calculations about poverty, and standards of living. Their sample size (around 1,25,000 households) is slightly more than the NFHS. And for the sex ratio at zero years (less than 365 days) these data are just as good as the NFHS?or other surveys of equivalent size. The NSS data show a steep upturn in the sex ratio, 0-364 days, for the last decade: a level of 977 in 2009-10 compared to 901 in 1999-2000. This is where the controversy, and doubts, begin. The popular belief is that such a number is impossible for India?perhaps in another era, not in our lifetime. No, not really; turn back to circa 1961 and 1971 and the numbers for CSR given by the census were 976 and 964.

The question that comes to mind is the seeming discrepancies in the table. Why does the NSS SR0 (the sex ratio for the 0-364 day old population) rise from 901 in 2001 to 924 in 2004-05 and 977 in 2009-10 when the same NSS reports a 0-6 child sex ratio falling from 927 in 1999-2000 to 916 in both 2004-05 and 2009-10? This has to do with the fact that the CSR (0-6) of the NSS is nothing but the average of the SRO?s (derived from the age-sex distribution in the survey year) for the previous 7 years. So the NSS 2009-10 CSR (reported in the year row 2011) is actually the average of the NSS SR0 from 2002-09; correspondingly, the NSS CSR for 2004-05 is actually the average of the NSS SR0 from 1997-2004. Note that when the NSS SR0 fell from 938 in 1991 to 901 in 2001, the same NSS CRR (0-6) rose from 922 to 927!

Consistent with the improving trend in the sex ratio is conclusion of a recent Lancet study on sex-selective abortions; the rate of increase in sex-selective abortions had declined from a 260% increase, 1991-2001 to a 170% increase, 2001-2011. Census 2011 also reports an improvement in the CSR for the bad sex ratio states of Punjab and Haryana.

Hence, it is important to start looking for explanations for why the sex ratio at birth is improving. We attempt to do that in a forthcoming paper*. Here are some preliminary thoughts/insights/ explanations. Tentative, but not infirm.

The steep declines that took place in 1991, and especially in 2001, peaking around 2005, were probably linked to a confluence of factors. With rapid growth, large numbers of Indians emerged out of poverty and entered the lower tiers of the middle class; and sex determination technologies came within reach of this emerging middle class. A third factor was rapid fertility decline. This emerging middle class has the most at stake in consolidating its new status and uses the family as a vehicle of upward mobility. And yes, it is present both in rural and urban areas.

It employs several strategies to move up. It shapes the family by reducing the number of children and engineering its gender composition?making sure more boys than girls are born. It educates the boys and sends them to urban areas or into salaried professional occupations. It demands handsome dowries at the marriage of sons. Daughters have little or no place in their grand design (they may be needed as daughters-in-law somewhere along the way?but there are always other families who will fulfil this need).

But what happens when the size of this middle class begins to shrink and people begin to move into the higher income brackets? What happens when girls have the same education as boys? What happens when the boys do not turn out as expected?witness Punjab where the huge improvement in CSR could partially be attributed to disaffection with a generation of sons lost to drug addiction and a dawning affection for girls who ?care?, and offer the elusive old-age support that sons were wanted in the first place for. Interestingly, both in upper class India and China a mother with two sons is today pitied, if not openly derided!

One final comment. India is a large, and diverse country. The data presented are an aggregate for the nation. Regional patterns will remain of immense importance in determining where daughters are missing. Even within a picture of plenty there could be places with abysmal sex ratios. Thus Salem, Theni and Dindigul districts of Tamil Nadu were long standing pools of female infanticide within an ocean of normal CSR for the state as a whole: 962 in 1981, 945 in 1991, 942 in 2001 and now back to 946 in 2011. Interestingly, it is these very districts that have improved.

Even when India had better CSR until 1981, the north and north-west of the country presented an abysmal anti-female, anti-daughter picture telling us that this show will still travel around the country yet. The dark shadow cast by the emerging middle class on girls is fortunately declining in size and, therefore, importance. That maybe the most important message of the back-to-normal sex ratios in India.

* Ravinder Kaur and Surjit S Bhalla, Changing Contours of Gender Imbalance in India: The Role of the Middle Class, forthcoming.

Surjit S Bhalla is chairman of Oxus Investments, an emerging market advisory firm; Ravinder Kaur is professor of Sociology, IIT Delhi