1 Introduction

China's sex ratio at birth (SRB) has been skewed since the early 1980s and reached a peak of 121.18 in 2004 before it gradually dropped to 111.3 in 2020 and eventually 108.3 in 2021, yet it is still above the normal range of 103–107. Regional disparities remain one of the key attributes of China’s SRB (Gietel-Basten, 2012; Gu & Roy, 1995; Poston, et al, 1997; Zhu et al., 2009).

An urgency imposed by reduced family fertility, a strong preference for sons and the spread of sex selection technology are three indispensable preconditions, which together determine the emergence of skewed SRB and the extent to which it is skewed (Guilmoto, 2009). Among the three main preconditions, son preference is the most essential and fundamental. Male-oriented patriarchal family systems lead to the persistence of son preference in China, Korea and India, which generate strong incentives to raise sons (Das Gupta et.al., 2003). Particularly in rural China, the patrilineal group (clan) composition of villages and family participation in practices such as building ancestral halls and updating genealogies, significantly shape son preference (Feldman et. al., 2007; Murphy et. al., 2011). The one-child policy implemented in China from 1980 to 2013 deteriorated the loss of daughters (Fred & Liu, 1986; Banister, 2004). In the context of a restrictive fertility regime, couples adapt by seeking fetal sex determination and sex selection to realise their preference for sons.

Evidences from many countries shows the expression of son preference changes over the transition from high to replacement-level fertility rates and skewed SRBs are expected to decline in the post-transitional stage (Bongaarts, 2013). The case of South Korea is often presented as a success in normalisation of skewed SRB. As SRBs go through the three stages of “rise, plateau, and decline” (Guilmoto, 2009), South Korea has undergone a transition from ‘son preference’ to ‘daughter preference’ and ‘no preference’. The shift in gender preference in South Korea can be attributed to conceptual changes driven by socio-economic growth, the ever-improving social security system, the rising educational level of women and the ever-increasing urbanization, which combined have spurred a rudimentary shift of the role of children from providing financial security to providing emotional support (Chung & Das Gupta, 2007). These changes are likely to further dilute son preference in the longer term (Chun & Das Gupta, 2021; Chung & Das Gupta., 2007; Murphy et. al, 2011).

However, a recent study finds that the underlying preferences for sons in South Korea still substantially persist, although they have declined to a large extent (Yoo et.al, 2017). It is argued that changes in the cultural, economic and institutional environment rather than changes in preferences are driving the normalisation of SRB in South Korea. The process and pace through which preferences change and persist is not yet well understood.

Is China the same story as Korea? China began to implement the three-child policy in June 2021 in response to its low fertility since the end of the one-child policy in 2016 (Zhai & Jin, 2023). External policies no longer impose fertility constraints on families. China's rapid modernisation, increasing urbanisation and social transformation in recent decades have affected people's fertility desires. China’s SRB is declining and what is happening to son preference over time? Variation in SRB across Chinese provinces is strongly related to variation across provinces in the degree of son preference (Poston et. al., 1997). Are there also significant provincial differences in gender preferences for children? Is son preference still dominant in some provinces?

Given the complex interactions between family systems, development and social norms that generate son preference, the factors that shape the change in son preference have not been well studied. We therefore undertook this research to explore in depth the temporal changes and regional differences in gender preference and to interpret why son preference persists and how it is related to persistently high SRB in the China’s context.

2 Measurement and data

This study has adopted a mixed research strategy including both quantitative and qualitative research. Descriptive statistics is the main method of quantitative research. The quantitative data comes from the five representative national sample surveys: the 1997 National Population and Reproductive Health Sample Survey, the 2001 National Family Planning/Reproductive Health Sample Survey, the 2006 National Population and Family Planning Sample Survey, the 2017 National Fertility Survey and the 2019 National Population and Family Dynamics Survey. These five surveys were organised and conducted nationwide by the government agency responsible for population and family planning and covered a wide range of areas, providing a solid data support for analysing gender preference at the national and regional levels.

The ‘ideal number of children’ which reflects an individual's gender preference for children, is currently the most widely used indicator of fertility intentions. All five surveys included questions about the ideal number of children, albeit with slightly different question designs (Appendix 2). In this study, women between the ages of 15 and 44 were selected for data analysis in these surveys. The ‘ideal number of children’, ‘the ideal number of sons’, ‘the ideal number of daughters’ and the number of responses ‘doesn't matter what sex’ were averaged and the proportions of the latter three to the average ideal number of children were calculated. Appendix 3 shows the recoding rule.

Corresponding to SRB, the ideal sex ratio among children was also calculated as the gender preference measurement indicator, i.e., the number of sons per 100 daughters in the ideal number of children. An ideal sex ratio of 100 among children means no gender preference, whilst a value greater than 100 suggests a son preference.

It is important to note that despite efforts have been taken to avoid compromising the preference information during our value assignment, the differences in question and choice design may still have led to certain deviations. The 1997 survey data only provided limited quantitative information because the single-choice question could not cover all gender combinations of children. Therefore, the response ‘let nature take its course’ was assigned the value of ‘I don’t mind gender’. The 2001 and 2006 surveys required respondents to provide specific numbers or select ‘I don’t mind’ when asked about son and daughter preference, rather than allowing for a more open-ended response. This might lead to the underestimation of son preference and daughter preference and the findings of no preference might be overstated in these three surveys.

In contrast, the survey designs of 2017 and 2019 offered the option of selecting ‘at least one’, which allows for a wider range of number and gender combinations, resulting in more accurate measurement of gender preferences. This choice is likely to make preferences for sons or daughters more apparent when compared to the previous surveys.

This study involved qualitative research through in-depth interviews to understand gender preferences and their persistence in certain provinces. The researchers aimed to learn about the interviewees' life histories and how they developed and practiced their gender preferences. The research team obtained ethical approval and conducted online semi-structured interviews with the interviewees' informed consent.

The questionnaire was designed to cover two main topics: gender preference and socio-demographic background. The main questions were: ‘What is the current gender preference for having children?’, ‘Is there still a significant difference in the value placed on giving birth to boys versus girls in these regions?’, ‘Which groups of people have a relatively strong preference for sons?’, ‘What behavioral strategies do they adopt to help achieve their preferences?’, ‘What are the reasons for the preference for male offspring in these provinces?’, ‘How do these regions differ from others in terms of gender preference?’. The prompts included in this study allowed for the inclusion of important areas related to family and community, which are considered relevant in contemporary China.

The study participants were married couples residing in provinces with high and persistent sex ratio imbalances such as Guangdong, Jiangxi, Guangxi, Hainan, Fujian, and Hunan. The research team recruited some interviewees through acquaintances’ introduction, while others were sourced from popular social media platforms in China, including WeChat, Moments and Douban.com. The team conducted a total of 49 interviews between October and November 2022, guided by the saturation theory. For some interviewees, a second or even third interview was conducted to ensure comprehensive data collection. The interviews, which lasted between 40 and 75 min, were recorded with the interviewee’s consent and transcribed by the interviewer. Anonymity and confidentiality were guaranteed to all interviewees prior to the interview.

The researchers aimed to reflect a diversity of opinion and experience by including a cross-section of the population across various socio-economic and educational profiles. The interviewees, aged between 21 and 42 years old, consisted of 6 males (including couples interviewed together) and 43 females (Appendix 4). Most of the females were in their peak reproductive years (25–39) and 73% of them had a college degree or higher. Occupations varied greatly. The majority of residents lived in urban areas of Guangdong Province. However, there were also rural residents in Fujian, Jiangxi and Guangxi provinces. Approximately half of the interviewees currently residing in urban areas migrated from rural areas, where rural life has had a significant impact on their current lifestyles and beliefs. The majority of interviewees were first married, while some were divorced or remarried. Most of the participants had one or two children, while a few had three children. Additionally, some participants were pregnant or currently without children.

The research team also virtually interviewed officials from Guangdong Provincial Health Commission and three county/city health commissions with intentions to learn about the measures taken by the government to rein in the skewed SRB.

3 SRB and changes in gender preferences

3.1 Skewed SRB turned from a national problem to a regional problem

Since 1982, China's sex ratio at birth has been skewed and continued to climb until peaking in 2004. It gradually dropped to 108.3 in 2021 and is expected to continue declining without any significant sign of rebound (Fig. 1). Regarding urban and rural areas, the sex ratio at birth decreased in villages, townships and cities in 2020. The decline in SRB was more rapid in townships and villages, returning to the range of 110–111, compared to cities, thus gradually narrowing the urban–rural gap (Fig. 2).

Fig. 1
figure 1

Trend of Sex ratio at birth in China. Sources: population censuses; population sample surveys; Statistical Data on Women and Children in China 2018; Statistical Bulletin on the Development of Health Undertaking in China 2019; website of National Bureau of Statistics

Full size image
Fig. 2
figure 2

Source: China’s population census

National Sex ratio at birth by urban/rural area (1990–2020).

Full size image

In regard to birth order, SRB presents a U-shaped distribution with the progression of birth order. The SRB for first births remains high at 113.2, while that for second births drops to 106.8 in 2020 and that is especially severe in births subsequent to the second child.

Despite a general waning of SRB, some provinces still exhibit a robust inclination towards male offspring among the reproductive-age population. In 2020, there were still 20 provinces with an abnormally high SRB, though their SRBs had all gone down. Table 1 shows that in terms of provinces and except for Xinjiang and Tibet, SRBs in 29 provinces have all gone down by 2020 compared with 2010. Whilst 9 provinces have seen their SRBs return to the normal range, 20 provinces were still haunted by a high SRB. Hainan, Jiangxi, Fujian, Guangdong, Guangxi, Hubei and Hunan were the top 7 provinces with the highest SRB, among which Hainan and Jiangxi still had a SRB above 120. These provinces have long been haunted by their persistently high SRBs (National Bureau of Statistics of China, 2022). In 2020, the combined number of births in these seven provinces accounted for 35.8% of all babies born in China that year (Appendix 1). These provinces with imbalanced sex ratio at birth are concentrated in southeastern China.

Table 1 SRB by province (1990-2020). Sources: China’s population census
Full size table

3.2 Temporal changes of gender preferences and regional variation at the provincial level

3.2.1 A growing trend toward daughter preference

Like many other countries, most respondents desire to have both a son and a daughter. According to the data from the five surveys, a balanced preference prevailed overall at the national level, remaining stable at around 50% from 1997 to 2006 and reaching a peak of 75% after 2017 (Tables 2 and 3).

In these surveys, while son preference showed a changing declining trend, daughter preference exhibited a fluctuating upward trend. Both proportions went up in 2001 and 2017 and both went down in 2006 and 2019. Such fluctuations were affected by the different survey questionnaires with the 1997 survey of one question type, the 2001 and 2006 surveys remaining the same and the 2017 and 2019 surveys adopting similar questions (marked by different colors in Table 2). Only identical questions and choices would aid in obtaining comparable results. By examining the changes in survey data between 2001 and 2006, as well as between 2017 and 2019, it is evident that both the proportion of son preference and the proportion of daughter preference have decreased considerably.

Table 2 Changes in gender preferences in the five surveys (%, person)
Full size table

Examining gender preference differences in the same year is more informative. In 1997, the proportion of sons was slightly greater than that of daughter, the gap widened further in 2001 and narrowed in 2006. After 2017, the proportion of sons preferred (8.64%) was significantly lower than that of daughter (13.12%). In 2019, both proportions decreased significantly, but son preference declined faster than daughter preference, further widening the gap between the two. Therefore, it can be inferred that between 2006 and 2017, the difference in son preference and daughter preference continued to narrow and eventually daughter preference prevailed, marking a reversal of gender preferences for children.

Due to variations in survey questions and the influence of fertility policy, the average desired number of sons and daughters fluctuated from year to year.Footnote 1 Changes in absolute value could be influenced by changes in the ideal number of children or indicate an increase or decrease in gender preference intensity. In 1997, 2001 and 2006, the average ideal number of sons stabilised around 0.65, accounting for approximately 37%. This number then sharply increased to 0.91 in 2017, accounting for 47%. The average desired number of females fluctuated and eventually surpassed the average number of sons after 2017. Both numbers and proportions declined in 2019, with the desired number of males falling quicker than the ideal number of daughters. Since then, the proportion of daughters has progressively surpassed that of sons.

Women of varying ages tend to have different gender preferences. To account for potential age structure differences in surveys, this study standardised the age structure of all five surveys to that of the 15–44 age group in the 2010 census. The results did not significantly differ from the unstandardised results (Table 3). In other words, the proportion of the ideal number of sons remains unaffected by structural differences. The changes observed are genuine and not caused by changes in the age structure of the surveys.

Table 3 Changes in the average ideal number of children in the five surveys
Full size table

It should be noted that gender preference reflects people's attitudes and beliefs regarding gender. The different directions and speeds of change between sons and daughters may lead to changes in the relative relationship between the two (Hou et. al., 2018). The relative link between the ideal number of sons and the ideal number of daughters or the ideal sex ratio of children, is more useful in predicting the development direction of gender preference. Despite a slight decline in recent years, daughter preference has weakened at a slower rate than son preference. Prior to 2006, the desired sex ratio among children was above 100, indicating a strong preference for sons. The changes in sex ratio between the two have caused a decline in the ideal sex ratio among children, 93.07 in 2017 and further 92.31 in 2019, indicating a shift towards a preference for daughters.

Women with a daughter preference are more likely to outnumber those with a son preference. Among different groups of women, those with higher levels of education are less likely to have a gender preference for their children. This finding is consistent with previous research (Chung & Das Gupta, 2007; Lin, 2009; Murphy, et al, 2011). Additionally, women residing in urban areas tend to have a stronger preference for daughters than those living in rural areas. Compared to older cohorts, younger cohorts exhibited significantly weaker son preference and stronger daughter preference.

3.2.2 Son preference dominant in some southeastern provinces

Gender preferences also featured obvious spatial agglomeration. In five surveys, daughter preference had dominated Jiangsu, Zhejiang, Shanghai and the three northeastern provinces, whilst son preference had dominated in a few southeastern provinces like Fujian, Jiangxi, Guangdong and Guangxi provinces.Footnote 2 In 2019, in most provinces, the proportion of the ideal number of daughters was greater than that of the ideal number of sons, with their ideal sex ratios among children signaling a salient inclination towards daughter preference. The ideal sex ratio among children was below 100 in most provinces in 2019, suggesting a consistent daughter preference among provinces. Guangxi and Guangdong were the only two provinces where the ideal sex ratio among children was higher than 100.

Figure 3 displays the changes in the ideal sex ratio among children in the years 2001, 2006, 2017 and 2019 in the 7 provinces with the highest SRB in 2020. These provinces exhibited significant differences in the ideal sex ratio among children in 2001, with Guangdong and Guangxi maintaining ratios above 120. From 2001 to 2019, the ideal sex ratio among children decreased in all provinces, with Guangdong and Fujian falling by more than 15%, while Hubei, Jiangxi, and Guangxi dropping by more than 12%. Despite the initial variations in ratios and the differing rates of change, gender preferences consistently converged across all provinces.

Fig. 3
figure 3

Changes in the ideal sex ratio among children in the seven provinces with the highest SRB in 2020

Full size image

4 Why son preferences decline but still persist in certain provinces

4.1 Decline of son preference and rise of daughter preference and gender indifference

The findings of the qualitative research confirm the results of the quantitative analysis, indicating a decline in son preference and a rise in daughter preference and gender indifference. Most interviewees expect to have both a son and a daughter. The trend of son preference is decreasing, while both daughter preference and gender indifference are becoming more prevalent.

Qualitative research shows the bottom line of having a son is no longer the family norm. The majority of interviewees, regardless of their educational attainment, place of residence or occupation, expressed a willingness to accept either a son or a daughter. As one interviewee, LKX3, a 30-year-old pregnant urban female government employee, stated, ‘Having a baby is like opening a gift box. Either having a son or having a daughter, each has its benefits.’

Higher educational attainment and occupational status are associated with a decreased likelihood of gender preference among women. Urban female interviewees exhibit more gender indifference or daughter preference. These findings align with those of quantitative research.

After giving birth to their first child, many interviewees expressed a desire for a second child of the opposite sex in order to have both a son and a daughter. However, some who had already had a son were concerned about having another son and hoped for a daughter instead (Shi & Yang, 2021). It was common for them to stop having children after having a son, rather than continuing to try for more sons. Having a daughter as the first child increases the probability of having a second child, as having two daughters or one son and one daughter are both acceptable.

Postnatal discrimination against girls has almost disappeared. In sympathy with the improvement of material conditions, young parents basically have no palpable preference for son over daughter in parenting their children, with some parents even being harder on their son(s) than their daughter(s) in a bid to help their son(s) develop a different temperament from their daughter(s). "Whether it's a son or a daughter, the expenses are the same—they drink the same brand milk, go to hospital in case of illness, and receive high education."(LHQ1, a 26-year-old male kindergarten teacher with a son and a daughter). This is a significant improvement compared with excess mortality rate of baby girls in the 1980s and 1990s (Attané, 2009; Li & Zhu, 1996). Parents would generally accept the fact of having a daughter and raise her indiscriminately. Some individuals have come to appreciate the unique and excellent qualities of daughters and have abandoned the notion of desiring a son.

However, there are significant intergenerational differences in gender preference. According to the interviews conducted, most parents born in the 1950s and 1960s preferred baby boys, often requesting at least one grandson or even more. This preference is particularly strong among those who live in rural areas and have less education. “My father-in-law has said in front of us many times that a certain family is so lucky and blessed to have ten grandsons. What he means is that only a family with all sons is blessed—that is, the more grandsons the better.” (ZCX2, a 41-year-old urban woman with a pigeon pair).

Son preference of the previous generation has been internalised as a belief. "My mother's generation was keen on having a son. As a matter of fact, they probably didn't think that much. Sometimes they didn't know why they wanted to have a son. They just knew that they had to have a son as they lived in the countryside."(LKX06, a 31-year-old social worker with two sons).

4.2 Changing but still prominent patrilineal practices, familism culture and gender stereotypes

4.2.1 Patrilocal coresidence and continuing the family lineage

In China, patrilineal descent, patrilineal succession, and patrilocal residence are all concrete manifestations of patriarchy in China (Li, 2012). According to Xu (2013), the tradition of patrilocal coresidence has eased in society due to modernisation, fertility decline and increased mobility. With the increasing population migration, most young people do not live with their parents or parents-in-law after settling down in cities. A new phenomenon is emerging where young couples sometimes live with both sets of parents. “After marriage, the young couple can live in the husband's family or in the wife's family, whichever has better financial conditions. One of the two children takes their father's surname and the other takes their mother's surname."(LHQ3, a remarried woman age 37). The high volume of circular migration in China not only accelerates the diffusion of new norms but also decreases patrilocal coresidence, which contributes to weakening son preferences (Das Gupta et al., 2009).

However, patrilocal coresidence is much more prevalent than other residence arrangements in contemporary China (Xu, 2013). Patrilocality remains the main organising principle for social and economic life for both male and female migrants (Eklund, 2015).

In the past two decades, it has become more common to include the names of daughters and their children in genealogies and for women to inherit family land and wealth even in rural areas. However, it is still considered natural for sons to carry on the family lineage and be given priority in inheriting the family's property. “My father said to me ‘I will leave most of my property to my son. Giving it to my daughter is like giving it to someone else’” (ZCX8, a 36-year-old housewife with a son). Southeastern provinces such as Guangdong and Fujian have a higher concentration of private enterprises and family businesses. This, coupled with the need for inheritance of family assets, contributes to the perpetuation of son preference (Shen & Chen, 2020).

In these provinces with higher SRB, the notion that only the son is considered a 'family member' is still prevalent. This is because daughters are expected to be married off sooner or later and therefore excluded from their natal family. “If people in their fifties are arguing with each other outside the home, they will say, 'I don't want to quarrel with you. Just call your son over.' Even if the son is useless, it is always the son who comes over to solve the dispute. Daughters are useless. Once married, they belong to someone else's family”. (LHQ2, a 21-year-old woman with a son).

4.2.2 Traditional customs and pressure of families without a son

The provinces with higher SRB have a deeply rooted familism culture. These areas have historically been less affected by wars, allowing local people to retain the custom of living together with clan members in rural areas. Within the clan, members have a strong sense of belonging and bonding and hold to the ‘big family’ tradition. Son preference is stronger in patrilineal families comprising three or more generations (Attané, 2009). The proportion of households consisting of three generations is higher in these provinces than the national average (Appendix 1). Even when two generations live separately, the older generation typically has a greater influence on the reproductive decisions of younger adults due to close familial ties, indicating a commitment to consolidated patrilineal values.

There are still many traces of son preference in folk customs. In rural areas in these provinces, activities such as updating genealogies, building ancestral halls and worshiping ancestors remain prevalent, with a strong emphasis on males. In weddings, certain fruits such as dates, lotus seeds, melon seeds, oranges, peanuts, pomegranates and longan, which imply fertility, are considered essential. At funerals, male offspring traditionally play important roles such as carrying the banner, leading the spirit to heaven, holding the incense burner or spirit seat and continuing the incense. In Chinese culture, a family without a son is considered to be extinct and the family lineage is therefore ended. This is viewed as a great unfilial deed and a source of shame (Huang & Huang, 2007).

In these provinces, ‘having a son is preferable to having a daughter’ has become an unspoken consensus. There are many hidden manifestations of son preference. For example, when guessing at the sex of fetus, people always say it must be a boy. This is because everyone believes that the answer of a boy will please the pregnant woman and her whole family. If the fetus is found to be a boy through fetal sex determination, the family of pregnant woman will also give the doctor a thicker red packet.

Patriarchal customs that prioritise having a son can put tremendous pressure on families without a male child, particularly in rural areas. “In villages like ours, if you give birth to a son, you will be freed from all the gossip; if you have only one child and it happens to be a girl, people will tirelessly persuade you to have a son”. (LKX06, a 31-year-old woman living in a village and with two sons). In this social environment, a woman's value is often measured by her ability to bear a son (Li & Lavely, 2009).

Interviews reveal that the stress of not having a son, particularly prevalent in rural communities, diminishes in urban areas. Additionally, autonomy and familial support have been found to positively impact women experiencing such stress. “My husband also don't want me to be totally drained for having a second child, so he helps me bear all the pressure. To be honest, I won't be able to handle it without my husband”. (TJ6, 35 -year-old woman living in Guangxi and with a daughter).

4.2.3 Gender stereotypes

Social norms and folk customs define the different roles played by men and women in public spaces and private domains, determining their respective family and social positions (Sangwha, 1999). Women often face disadvantages in accessing resources due to persistent stereotypes that portray them as less competent, particularly in these provinces. ZCX2 (a 41-year-old woman with a pigeon pair) mentioned she once drove a car to his husband's hometown in Jiangxi Province and surprised the locals. “‘Wow, you women from Hunan can drive?' I was surprised too. Isn't it normal for women to drive a car? So, I observed carefully and didn't see any women driving a car there. In my hometown, more than half of women should be able to drive a car.” It is unfortunate that driving was once considered a male-only trait. The act of women driving a car is often viewed as a challenge to traditional gender roles. The locals' surprise at this phenomenon reflects their adherence to gender stereotypes.

In regard to traditional gender roles in the family, the husband is responsible for feeding the family, whilst the wife is responsible for doing housework and taking care of children. Gener stereotyping has decreased significantly in youner generations. However, older generations still hold onto traditional gender roles and view anything different as shameful. Even if the wife has a higher income than the husband and the husband is better at taking care of children, other people probably never know this fact. “It is generally accepted that the wife must do the laundry for her husband and I had a dispute with my mother-in-law on this matter. I earn more than my husband. She said that I don't have to do the laundry, but I must never tell other people. She doesn't want me to convey the idea of ‘my husband being homemaker and I being breadwinner’. They're afraid that I might embarrass their son…They think it is natural for a man to earn more than his wife.” (LHQ2, a 21-year-old woman with a son).

4.3 Differences in the rearing costs and benefits of son and daughter

In contemporary times, parents tend to show no difference in raising and educating their sons and daughters, resulting in almost equal childrearing and education costs for both genders. However, the most significant disparity lies in the expenses associated with marriage. Typically, when a son gets married, he is expected to prepare the housing and pay a substantial amount of bride price, which is primarily covered by his parents. As a result, a son’s marriage has become a crucial method of intergenerational resource transfer. Due to the long-standing skewed SRB, the marriage squeeze has resulted in a continuous rise in bride prices over the past few decades. In fact, the marriage of two sons can consume all of their parents’ savings (Wei & Zhang, 2011). This greater burden of having a son contributes to the fear of having two sons, which undermines son preference to some extent.

Many interviewees believe that parents have a duty to assist in their son's marriage, but only need to support their daughter within their abilities. This tradition is linked to patriarchy because the son is seen as the primary family member and investing in the son is seen as equivalent to investing in the family or in the parents' old age. TJ4 (a 29-year-old woman with a daughter) said: “If you have a son, you will need to help him out when he grows up. You will have to prepare a house for him and help him look after his children after his marriage. It costs more to have a son than a daughter and you will probably have more worries. Still, you can't completely ignore your daughter. If conditions allow, you should also give your daughter some necessary support.” In the past decade, a popular saying has emerged that suggests ‘having a daughter brings wealth, while having a son requires spending a fortune’.

The expectation that a son will provide financial support in old age is strongly associated with son preference especially when there is no pension or social security system in place (Li & Lavely, 2009). Upon the universalisation of endowment insurance, parents’ high expectation for their son's support on old-age care are waning gradually (Ebenstein & Leung, 2010). In both urban and rural areas, interviewees expressed a reluctance to rely on their children for old-age care. “In the past, people in their 60 s had no fixed income when they got old and had to count on the support from their children. Now we all have social security, so we won't have to worry about how to feed ourselves in the future.” (LKX3, a 30-year-old pregnant woman). The dwindling demand for son's support on old-age care is conducive to weaken son preference.

The decreasing price-performance ratio of sons is accompanied by the increasing cost-performance ratio of daughters. This implies that raising a son is a less cost-effective investment than raising a daughter. According to interviews on elderly care needs, respondents expressed a preference for daughters due to their perceived higher levels of empathy and caregiving ability, which can provide more emotional and caring support to their parents (Chun & Das Gupta, 2021; Shi, 2009). "In rural areas, there is a saying like 'having a son makes you proud, but having a daughter makes your life way easier."(LKX6, a 31-year-old woman with two sons and living in a village). Interviewee LKX8, who was a social worker, shared her experience of looking after elderly parents. "When looking after the elderly, I found that they turned to their daughters more often than to their sons for help. Although some elderly people have many sons, they tended to seek assistance from their daughters, because daughters are more attentive, while sons are often busy with work. They feel that a daughter is more reliable than a son."

The increase in women's educational attainment has led to greater social participation, and the income gap between men and women is narrowing in non-agricultural industries. This trend is also likely to weaken son preference (Qian, 2008). However, it should be noted that women's educational attainment is still relatively low in these provinces with higher SRB. With the exception of Guangdong Province, where the percentage of women aged 15 and over with a college diploma or higher is higher than the national average, the other six provinces fall below the national average (Appendix 1). In Guangdong and Fujian provinces, where the market economy is thriving, couples enter the market through family-based operations and production. In family-based operations, the husband typically assumes the role of a leader while the wife often serves as an assistant. This can result in the undervaluation of the wife's labor.

5 Translating preferences into reproductive behaviors

5.1 Having a son is a responsibility to parents

Historically, China was governed and held together by families and clans, leading to a prevalent ‘family-centric’ familism culture. China's intergenerational culture emphasises the ‘harmonious coexistence’ between parents and children, as well as the traditional ‘filial piety’ culture.

Due to the deep-seated traditional family ethics and filial piety in these provinces, reproduction has become a matter of great importance for the family lineage. Not only does the previous generation attach importance to the reproductive behaviors of their offspring, but the current generation also values the opinions of the previous generation. Intergenerational differences in gender preference have led to conflicts between parents' desire for grandsons and their adult children's reproductive autonomy. Reproduction has become a source of tension between the two generations, resulting in varying outcomes based on their interactions.

For example, after LKX9 (a 37-year-old woman with two daughters and a son) gave birth to two daughters, her parents-in-laws adopted a carrot-and-stick approach. At first, her parents-in-laws threatened not to help take care of her daughters and LKX9 angrily fought back by saying they should not expect her to take care of them when they get old. LKX9 chose to be a stay-at-home mom. Her parents-in-laws later changed their tactic and expressed their willingness to help her take care of her second child, so that she could be relieved from the childcare burden and prepare for new pregnancy as soon as possible.

The conflict between the two generations regarding reproduction ended with the daughter-in-law’s compromise. Due to her parents-in-law's modest approach, she relented and was recommended an effective folk remedy for conceiving a son. As a result, she began preparing for pregnancy and eventually gave birth to a son, making her parents-in-law happy.

Most interviewees expressed a strong desire for a specific ideal scenario: giving birth to a son first, in order to avoid being pressured into having a second child and to have a say in the decision to have another child, when to have it and how many more to have. This partially explains why China's sex ratio at birth for first-borns remains high (Jiang et al., 2016). The interviewees viewed having children as a life task and a responsibility to their parents.

5.2 A dilemma to implement son preference: sex selection or having more children

Fetal sex determination and gender selection techniques are available in these provinces. For instance, in Guangdong Province, which is adjacent to Hong Kong, people send the blood samples of pregnant women to Hong Kong in order to determine the sex of the baby. It is important to note that these conduits have existed for a long time and do not charge much.

Ultrasound is also used in both private clinics and even public hospitals to determine the sex of fetuses (Chu, 2001). Some doctors may not disclose the information directly, instead, they tend to communicate the fetal sex cryptically. In some cases, the family of the pregnant woman may offer a 'red packet' to show their gratitude. The availability of ultrasound determination of fetal sex is even higher in relatively remote townships. Like fetal sex determination, sex selective abortion is not difficult to have either. Not only are there private clinics, but women can also travel to Hong Kong for abortion.

Some interviewees reported pressure from their parents to have a grandson or a desire for a son as reasons for resorting to such techniques. Among the 42 interviewees with children or being pregnant at the time of the interview, eight had undergone fetal sex determination and three couples had done so twice. Others cited the need to know the baby's gender for practical reasons, such as buying clothes.

Since the 1990s, the Chinese government has implemented a series of laws and regulations to prohibit the use of technology for fetal sex determination and sex-selective abortion and has launched severe crackdowns. However, there is still a discrepancy between the actual and anticipated effects (Tang, 2020). Local government officials acknowledged that gender-based discrimination is difficult to investigate due to covert tactics and lack of evidence. An official from local government said: "These are highly lucrative services. We definitely can't stop the illegal practices, but the crackdowns certainly help with our SRB governance".

Aside from sex selection, having more children has become another option to achieve desired gender preference. Reproduction has become an autonomous decision within the family due to the lifting of external policy restrictions since 2013. The conflict between son preference and the number of children still remains, but it has shifted from an external to an internal one. Couples are increasingly making rational decisions regarding reproduction due to the rising costs of childbearing, childrearing and education. Family income, time and childcare resources are significant constraints that lead to having fewer children. A recent study shows son preference no longer has a significant effect on fertility potential in China (Yu et. al., 2023). There is no need to have more children because of the quest for the child's gender preference, which makes no difference between urban and rural areas. Couples are often faced with a dilemma when trying to have a son within a limited number of births: sex-selective abortion or having more children.

However, some couples still feel motivated by son preference or parental pressure to have more children (Bongaarts, 2001; Li, et al., 2017). “This is the situation here in my hometown. I have seen many people with two daughters already bear their third or fourth child, because they are bent on having a son.” (LKX2, a 34-year-old housewife with two daughters and one son) "For some couples, achieving their ideal gender preference is more important than the stress of raising an additional child and sex-selective abortion is seen as a temporary inconvenience. LHQ2, a woman in her 30 s, shared her neighbor's story. “All she wanted is to have two children, one son and one daughter, so the second child must be a son. Since the second child was not a boy, she would not keep her. She thought the pain was temporary and there was no need to have another daughter.”

One positive trend is that some young couples are becoming less accepting of abortion, leading them to choose having more children over sex selection. The main reasons for this shift are respect for life, natural motherhood and the desire to protect women's health. Additionally, the physiological, economic and opportunity costs of pregnancy continue to rise for women. Most interviewees stated that abortion is more harmful to health and fertility than childbirth. “My mother said that aborting one child hurts a woman as much as having seven children. Therefore, no matter what, don't have an abortion, but give birth to the child”.(LKX9, a 37-year-old woman with two daughters and a son). The decision to not terminate pregnancies not only prevents gender-based selection but also results in unintended pregnancies resulting in live births. Many second children were born as a result of unintended pregnancies in the interviews.

6 Discussion and conclusion

6.1 The change in gender preference is a manifestation of the value shift from familism to individualism.

The advantages of having a son lie mainly in the continuity of the family lineage, the inheritance of family property and the respect of others. Therefore, having a son could bring the family social benefits such as respect and reputation, which are highly valued in the acquaintance society, thus building social relationships and affirming the social self. They are the embodiment of family interests, responsibility to ancestors and social judgment on the family, all of which are extrinsic and family-centered and have no direct impact on the parents’ current standard of living and emotional perception.

On the contrary, the practical, palpable and individual-centered emotional and caring support provided by the daughter can help improve the well-being of the parents. Therefore, the preference shift from son to daughter signifies the tug-of-war between familism and individualism values, as well as the transition from a family-centric culture to individual-centered considerations.

Socio-economic development has led to a change in values regarding reproduction. As people lead more affluent lives, their reproductive desires are shifting. The pursuit of self-actualisation and realistic interests is becoming more important than material and moral considerations, as well as the need for social respect. This shift is leading to a gradual erosion of utilitarianism in reproductive decision-making.

Clash with parental values between generations is particularly pronounced. Unlike their parents, who find meaning in life through fostering accomplished children, those born after the 1980s and 1990s hope to strike a balance between child-rearing and personal quality of life by doing their best to raise their children based on their financial conditions without sacrificing their own consumption or enjoyment. In this generation of young parents, the sense of responsibility to children has blended with individualism.

6.2 To fulfill their parents’ son preference is a strategy for young couples to combine filial piety with reduced parenting stress

Gender preference and its translation into reproductive behavior reflect the nature and dynamic of intergenerational relation in China. The study found that parents have an impact on the reproductive behaviors of young couples. Intergenerational relations are an important factor to consider when examining gender preferences and related behaviors in these provinces.

The interviews revealed the presence of transactional intergenerational relationships. The older generation exerts influence on the reproductive behaviors of the younger generation through financial and emotional support. This support symbolises the close intergenerational relationships and it is also a means for the older generation to receive filial piety, gratitude and obedience from the younger generation. Due to deficiencies in China’s fertility support policy, the burden of childbearing and childcare falls mainly on young couples. To fulfill their reproductive duties while maintaining a foothold in a highly competitive society, young couples have to rely on the support of their parents. They strive to meet their parents' wish to have a son in exchange for financial and childcare support, thereby maintaining harmony in the father-son relationship.

Their reproductive compromise is not just a matter of personal choice, but rather a coping strategy for dealing with life stress and inadequate parenting support. This reflects both the continued influence of traditional family culture in the face of China’s rapid modernisation and the practical and expedient choice made by families to manage limited childbearing and parenting resources. This is also a manifestation of neo-familism, a new kind of familism that emerges in a risk society such as contemporary China, where individuals are expected to face and solve these systemic risks but lack the resources to do so, forcing them to return to their parents and families for support (Yan, 2018).

6.3 Changes in gender preference are driven by a combination of period and cohort effects

Contemporary China features a society in transition. Socio-economic development, social norm and value change, advance of sex selective technology, together with the development of governance measurements, were period changes. Notably, the post-80 s and post-90 s generations have become the primary group for childbearing. They grew up in a new era of development and have different views towards family size and desirable sex composition of children compared to their parents. The disparities between the two cohorts highlight the differences in life background and personal experiences against the backdrop of social transition. Changes in their attitudes and preferences are typically driven by their cohort (Yoo et. al., 2017). When the younger cohorts gradually replaced the older ones as the main forces of childbearing in China, son preference decreased and gender indifference eventually increased at the aggregate level (Lin, 2009).

6.4 A shift from behavioral to conceptual expression of son preference

The resistance to induced abortion has effectively contributed to the decline in fetal sex selection. Although a preference for sons still exists, sex-selective abortion is no longer considered a desirable option for achieving an ideal gender composition of children. Therefore, even if a preference for sons exists, it may not lead to sex-selective abortion. Sex-based stopping patterns may be an alternative for couples who have the time and resources to care for more children. The resistance to induced abortion has also led to a trend of having more children in case of unwanted pregnancy, regardless of fetal sex concerns.

In this sense, the existence of gender preference may contribute to the fertility rebound. If people only have an emotional preference for sons without physically seeking fetal sex selection or discriminating against daughters after birth and if they have more children to practice their gender preferences, it can actually help boost fertility and bring the skewed sex ratio at birth back to the normal range. This is a win–win situation that the Chinese government hopes for in a low fertility regime. The ultimate objective should be to eliminate the persistent gender inequality, address the prevalent bias towards sons and encourage a shift towards 'daughter preference' or 'gender indifference'.

6.5 Implications for gender imbalance in the future

Although son preference is weakening, it remains pronounced in several provinces. This study reveals the challenges that slow down the weakening of son preference in these provinces. Son preferences are still rooted in the value of life and have a great influence over realistic rationality. Traditional gender roles have had a lasting impact on social and family life, with no indication of fundamental change. The remaining patriarchy and familism culture are the main causes that still carry tremendous clout in these provinces.

There are also positive signs. Women are not only victims of son preference and fetal sex selection, but they also make decisions about and bear the brunt of fetal sex selection, as well as take on childbearing and childcare responsibilities. The gender identity and autonomy of young women are important determinants of whether gender preference translates into sex selection. Women who are well-educated and economically independent tend to have weaker son preferences. They generally have a relatively high status in the family, dare to insist on their own viewpoints and are adept at tactically countering their parents-in-law (Wang et. al., 2013). Additionally, they prioritise their own health and are opposed to unprincipled sex-selective abortions. These findings suggest that efforts to empower women have been successful and should be strengthened.

The disparity between urban and rural areas in terms of patriarchy family structure and gender-specific customs and practices is significant. However, this gap is likely to narrow over time due to rapid urbanisation and the spread of modern lifestyles in China. According to this research, young couples in both urban and rural areas show little difference in gender preferences. As a result, the urban–rural difference in the sex ratio at birth has almost disappeared as of 2020.

Regional divergence will continue to be a prominent feature in many aspects of Chinese society for the foreseeable future due to uneven economic and social development, as well as differences in cultural backgrounds. As the change of gender preference has yet to filter through to behavior and there are a large number of newborns in these provinces with persisting son preference, they are holding back the speed with which the national-level SRB can return to normal. Nevertheless, according to Guilmoto’s (2009) theory of sex ratio transition, the sex ratio falls when the patriarchal system is undermined and son preference is weakened. This study clearly provides some reason for optimism about the future return of China’s SRB to normal. The demographic transition trajectories of China and South Korea are becoming increasingly similar in terms of changes in fertility and gender preferences.

6.6 Implications for future research

Our study has limitations and suggests directions for future research. Firstly, the quantitative data, although drawn from nationally representative samples, have significant shortcomings. Not only are the questions inconsistent, but there may be differences in reporting errors. This makes the quantitative results less intuitive and unambiguous. More data from different sources will need to be analysed in the future to further confirm the changes in gender preferences. Secondly, the number of qualitative interviews was relatively small and interviews were conducted online. Interviews conducted online do not allow for a deeper understanding of the real-life situation of the interviewee, who may try to present only the socially acceptable side of the story. Thirdly, the rise of SRB in these provinces was the earliest and has been the longest. More systematic efforts are needed to discuss the reasons specific to these provinces.

Despite these limitations, our study provides a unique insight into recent changes in gender preferences in China as it transitions from low to extremely low fertility. Cultural factors are increasingly important in determining gender preferences. Given that cultural change takes time, China’s SRB faces the risks of decelerating decline. Sustained government efforts are particularly needed in the areas of women's empowerment and social norm change.