Chapter 2 cover photo credits: Mark Henley / Panos Picture

---

Son bias

2

---

Chapter 2 cover photo credits: 

© Mark Henley / Panos Picture 

Cambodia, Battambang province. Young peasant girl helping during the rice harvest.

---

1. Son bias and poverty dynamics

The Social Institutions and Gender Index (SIGI) Son Preference 

Sub-Index draws on Amartya Sen’s 1990 work on ‘missing 

women,’ or the number of women who could have been 

expected had girls received equal health care, medicine and 

nutrition. Sen hypothesised that international distortions in 

sex ratios equate to as many as 100 million missing women 

and can be explained by female foeticide and ‘gendercide’ – 

the systematic and often lethal neglect of and underinvestment 

in girls and women. Klasen and Wink (2003) further developed 

this approach to estimate sex ratio

1

 differences over time and 

space, and it is their methodology that underpins the country 

assessments in the SIGI rankings. 

In this chapter, however, we conceptualise intra-household 

gender  biases  more  broadly.  We  include  differential 

investments in, and care and nurture of, boys and girls from 

conception, with implications across a spectrum of negative 

developmental outcomes, from mortality through to human 

capital  development  deficits,  time  poverty  and  psychosocial 

ill-being. To signal this broader understanding of unequal 

treatment between sons and daughters, we refer to the social 

institution as ‘son bias.’ It is important to note from the 

outset that imbalanced sex ratios tend to be limited to certain 

geographical regions (Asia and the Middle East and North 

Africa region). However, our argument is that general intra-

household differentials between sons and daughters are more 

widespread and that there is good evidence on a number of 

indicators of this gender bias across regions. 

The chapter begins by reviewing the factors that underpin 

son bias, and then turns to a discussion of the multidimensional 

impacts of such practices on girls’ experiences of poverty 

and vulnerability. We recognise that son bias is not shaped 

by poverty alone, and indeed is found among upper wealth 

quintile groups in some countries and communities.

2

 

However, we focus our discussion here on linkages between 

poverty, vulnerability and son bias over the life-course and in 

intergenerational terms. The second half of the chapter focuses 

on promising initiatives aimed at challenging the norms and 

practices that underpin son bias.

Son bias

The patterning of son bias 

The most typical manifestation of son preference is the relative 

neglect of daughters, although the most extreme form is 

female infanticide

3

 – the intentional killing of baby girls. In 

many cases, however, female infanticide has been supplanted 

by  sex  identification  testing  and  sex-selective  abortion, 

shifting postnatal discrimination to prenatal discrimination 

(Klasen and Wink, 2003). A 2008 World Bank study drawing 

on Demographic and Health Survey (DHS) data estimated son 

preference based on the likelihood of families having another 

child if they have only daughters (Filmer et al., 2008). It found 

that, in Europe and Central Asia, families are 9.4 percentage 

points more likely to have an additional child if they have only 

daughters. In South Asia, they are 7.8 percentage points more 

likely, in the Middle East and North Africa 5.8 percentage 

points  more  likely  and  in  East  Asia/Pacific  3.7  percentage 

points more likely. They found no evidence of son preference 

by  this  measure  in  sub-Saharan  Africa  (surveys  did  find  a 

subjective preference for sons but this did not translate into 

demographic ratios) or Latin America (where there seems to 

be a preference expressed for daughters). 

This  is  largely  borne  out  by  SIGI  findings  (with  the 

exception of Europe and Central Asia), as Table 1 indicates. 

Latin America and the Caribbean, and sub-Saharan Africa  

have considerably smaller scores, indicating lower prevalence 

of son preference. These regional trends do, however, hide 

significantly  higher  ratios  in  a  small  subset  of  countries, 

especially India and China.

4

 Presence of other siblings and 

sibling order also have a strong effect on measures taken to 

ensure that future children are girls. For example, in India the 

first child is much less likely to be aborted for being a girl than 

subsequent children (Jackson, 2010). Overall, neglect of girls 

is generally more severe for later-born girls and for girls with 

elder sisters, and this is particularly the case in rural areas 

(Klasen and Wink, 2003). In India, the sex ratio of second-

born children has been estimated at 716 to 1,000 boys in the 

incidence  of  the  first  child  being  a  girl,  compared  with  an 

excess of girls – 1,140 girls to 1,000 boys – if the first-born is a 

boy and 910 girls to 1,000 boys for first born children (Sahni et 

al., 2008).

---

32

2 | Son bias

2. Accounting for son bias 

A substantial body of evidence shows that son bias is shaped 

by a complex interplay of economic, socio-cultural and 

demographic factors. Adding to this complexity is the fact 

that  intra-household  attitudes and  behaviours  intersect  with 

societal-level gender biases and in turn perpetuate both ‘private’ 

and ‘public’ sphere discriminatory norms and practices. In this 

section, we provide an overview of the key explanations for 

the survival and malleability of this social institution, and the 

ways in which these intersect with poverty dynamics.

Economic factors

The Economist noted in its March 2010 Leader on the perils of 

son preference that gendercide affects rich and poor alike, but 

that there is a substantial body of evidence highlighting the 

economic rationale for son bias. Arguments centre around the 

economic contributions that sons are able to make over their 

lifetime to the family on the one hand, and the costs daughters 

exact on the other (see Box 9). Sons are expected to maintain 

financial and social ties to the household throughout their lives 

and, in developing country contexts, where social security 

systems are underdeveloped, many parents rely on their 

sons’ future earnings for their old-age security (Jayaraman 

et al., 2009; Wang, 2005). Indeed, 51 percent of respondents 

in a fertility survey in Hubei province identified the primary 

motivation for a son as the desire for old-age support, with 

continuation of the family line a distant second (20 percent) 

(Ding and Zhang, 2009). Moreover, county-level pension 

programmes in rural China have been found to lower the 

sex ratio at birth by 9 percent (Ebenstein and Leung, 2010).

5

 

Patrilineal inheritance systems spanning a wide range of 

cultural and religious traditions (from Confucianism to Islam, 

from Hindu law to Kenyan inheritance customs) also mean that 

sons inherit property,

6

 exacerbating discrimination against 

girls and women and motivating prioritised investments in 

boys (Jackson, 2010; Quisumbing, 2007) (see also Chapter 1 on 

Discriminatory Family Codes). 

The economic ‘rationality’ of these practices is often 

reinforced by the fact that daughters are only transitory 

members of their natal families before their marriage, upon 

which they move to and contribute to the families of parents-

in-laws, typically becoming physically and psychologically 

isolated from their birth home (Chu et al., 2006). Moreover, 

female employment is often undervalued, making men 

potentially more productive future ‘assets.’ This is especially 

the case in rural areas, if, as discussed in Chapter 3 on Limited 

Resource Rights and Entitlements, women are not involved 

in commercial agriculture and/or do not have property 

rights (Gupta and Dubey, 2006). In other contexts, however, 

parents may seek to mobilise resources from older unmarried 

daughters to improve the family budget in general and the 

educational outcomes of sons in particular. This can be paid 

work (often unskilled or semi-skilled factory work in urban 

areas) or household work, which frees parents up to work 

longer hours (Chu et al., 2006). 

In cultures which practise dowry payments, daughters 

are often also seen as an economic liability on account of the 

high cost of weddings, as highlighted by adverts for mobile 

abortion clinics in India which cry ‘Pay 50 rupees now to save 

50,000 rupees later’ (Basu and Jong, 1999). Diamond-Smith 

et al. (2008) note that one-fifth of women surveyed identified 

dowry payments as the reason they did not want daughters. 

Rather than declining with the onset of modernity, these costs 

are escalating over time, and dowry payments may equate to 

as much as two-thirds of a household’s assets (Nolan, 2009) 

or several times more than total annual household income 

(Anderson, 2007). This owes in part to expectations that girls 

will be educated, with associated costs; the increasing demands 

that a consumer-oriented culture exerts; new economic trends, 

especially  increasing  international  remittances,  which  are 

inflating dowry demands; as well as the potential opportunity 

for social mobility which marriage represents, especially for 

poor low-caste families (Diamond-Smith et al., 2008; Pande and 

Astone, 2007). It is important to note that this is particularly 

burdensome in households where, because of parental desire 

for a son, there are multiple daughters, as parents continue 

to  have  children  in  an  effort  to  have  sons  (Bélanger,  2010; 

Lundberg, 2005). 

Finally, it is important to point out the linkages between 

female income and education (Fuse, 2008). Qian (2006) 

found that increasing female income, holding male income 

 

Table 1: SIGI son preference scores by geographical region

Overall average

Latin America and Caribbean

0.01

Europe and Central Asia

0.03

Sub-Saharan Africa

0.04

East Asia and Pacific

0.19

Middle East and North Africa

0.38

 

Source: http://stats.oecd.org/Index.aspx?DatasetCode=GID2

---