Why are richer families in India still choosing sons?

When economists and demographers talk about sex selection, they usually point to son preference: the idea that parents want a boy to carry the family name, inherit property, or provide support in old age (Edlund 1999, Bhaskar 2011, Arokiasamy and Goli 2012).

But there is another side – daughter aversion (Alfano 2017, Bhalotra et al. 2020). In India, parents often pay large dowries when their daughters marry. In theory, these payments should simply clear the marriage market: wealthier grooms ‘cost’ more, poorer ones less. In practice, however, they make daughters less valuable than sons.

We provide new evidence and an explanation for how this happens. We build a simple model in which dowries serve as both the market-clearing price as well as the bequest from parents to their daughter. This theoretical model of the Indian marriage market captures three institutional features:

1. Marriage within caste: almost everyone marries within their caste. In our data, 98% of the parents and 95% of the children married within their caste.
2. Patrilocal residence: daughters leave home after marriage and live with their husband’s family.
3. Dowries as the norm: the bride’s family pays cash, gold, or gifts to the groom’s family across India. In our data, the dowry is three to four times the household’s annual income on average.

In our model, parents derive equal utility from what their son and daughter consume, and they want to share their wealth equally with sons and daughters. But because of patrilocal marriage, the bequest they give their daughters (through dowry) flows to the groom’s parents, who then decide how to divide resources within their household. If the daughter has little bargaining power in her new home, she ends up consuming less than what she received as dowry. Her parents, anticipating this, realise they cannot ensure her well-being even with a large dowry. As a result, they are worse off with daughters than with sons.

Three ingredients make this disadvantage stick:

1. No commitment from the groom’s family: once the dowry is paid, the bride’s parents cannot control how it is used.
2. Unequal power in the marital home: women’s weaker position means they get less benefit from the resources they bring.
3. The strong social norm that all girls must marry: parents cannot avoid these losses by keeping their daughters single.

This set-up helps us understand how institutional realities shape economic behaviour and generate sex selection even when parents genuinely care about their children, irrespective of their sex.

In Indian caste-based marriage markets, families match assortatively on wealth. The wealthiest girls match with the wealthiest boys. However, the shortage of girls due to sex selection implies that all other girls will ‘marry up’. This wealth-gap increases as we move down the wealth distribution, which implies, in turn, that less wealthy families are less disadvantaged by having a girl.

Based on the preceding discussion, the key prediction of our model is as follows: sex selection increases with relative wealth and wealthier families within each caste are more likely to sex select.

To test this prediction, we use data from the South India Community Health Study (SICHS) – a massive census and survey that we conducted, covering 298,000 households across 57 castes in Tamil Nadu’s Vellore district. The dataset includes detailed information on marriage payments, household wealth, and the sex of 80,000 young children.

Our main findings are as follows:

• Dowries are universal and substantial. The average payment is three to four times a household’s annual income. Even after decades of development, almost no marriage takes place without one.
• Daughters ‘marry up’. Parents consistently report that their daughters married into richer families, while their sons’ spouses came from households of equal or lower wealth. Relatedly, we can see that the amount given by the girls exceeds the amount received by the boys at each wealth level in Figure 1a.
• Wealthier families have more skewed sex ratios. Within each caste, the share of boys rises sharply with household wealth – from near-natural ratios (around 101 boys per 100 girls) at the bottom to highly distorted ratios (up to 118 boys per 100 girls) at the top. In Figure 1b, we can see the declining probability of having a girl as we move up the within-case wealth distribution.

Our findings also distinguish between relative and absolute wealth. Families who are richer relative to others in their caste are more likely to engage in sex selection. But families who are richer in absolute terms (when the whole economy grows) tend to show less bias.

This means that economic growth alone will not solve the problem. As long as inequality within castes widens, as it often does during growth, the pressure to find wealthy grooms and pay high dowries can intensify, reinforcing daughter aversion at the top.

First, that the marriage market channel is almost as important as the well-known son preference channel. Our estimates indicate that 52% of sex selection in the study area is due to the marriage market channel (see Figure 2a).

Second, that existing government programmes intended to tackle sex selection, such as cash transfer schemes that reward parents for having daughters, can potentially backfire (see Figure 2b). States like Haryana, Tamil Nadu, and Delhi offer payments or savings accounts to parents of girls, hoping to make daughters more ‘valuable’. When such transfers go to parents, they effectively increase household wealth, which in turn can raise dowry expectations and increase sex selection within castes. Targeting only the poorest families can worsen sex selection further up the wealth distribution by changing marriage prices, possibly exacerbating the problem on net.