Literature review

Factors affecting delayed age at first marriage

Factors affecting delayed age at first marriage are influenced by various economic and social factors [22]. Firstly, one key economic factor is the social class identity of individuals and families, which has been extensively researched in relation to marriage matching [23–27]. Over the past fifty years, scholars have increasingly focused on individuals rational decision-making in relation to marriage. They emphasize that individuals weigh the economic status of potential partners and consider whether marriage will result in a better quality of life than remaining single. If the post marriage quality of life is not favorable or offers no benefits, individuals may choose to delay or even forgo marriage [27]. Conversely, if economic resources are sufficient, both partners are more likely to marry early and have more children [28].

In Chinese society, marriage choices often adhere to traditional ideas, with considerations that involve interests exchanged between families of the same social class [29]. Although traditional arranged marriages are explicitly prohibited by Chinese marriage law, young people still seek suitable partners for marriage. “Suitability” is largely determined by factors such as income level, family wealth, and extended family connections [30,31]. An economically distinctive factor of Chinese marriage is the presence of a bride price. Typically, the groom's family pays a significant amount to the bride's family before the wedding, and the bride’s family reciprocates with a symbolic dowry that often serves as start-up capital for the newly formed family. However, in some rural areas, the bride price can be exorbitant, leading many rural mens families to be unable to afford it. Consequently, rural women often opt to marry urban men with higher economic statuses, resulting in an increasing proportion of single rural men [32–34]. In summary, the examination of economic factors in relation to delayed age at first marriage has focused primarily on the rational decision-making of individuals, without fully considering the influence of social policies, culture, and other regional factors on delaying marriage in different areas and time periods.

Secondly, among the social factors, the shift in the concept of marriage and the increase in educational attainment are the primary drivers behind the postponement of marriage. Studies pertaining to the second demographic transition (SDT) attribute the delayed age of marriage largely to changes in individuals and families perceptions of marriage, which serves as the most influential social force distinguishing the second demographic transition from the first [1]. Presently, the global trend entails a delay in the age of first marriage, with East Asian countries such as South Korea and Japan surpassing the rates reported in many Western countries. Notably, even certain African nations are witnessing a significant rise in the age of first marriage [35,36]. While Nordic societies present higher rates of lifelong non-marriage, Chinese residents in East Asian societies still place considerable value on the institution of marriage. Among Chinese urban youth, the transition in the concept of marriage is driven primarily by the pursuit of personal happiness, inspired by increased self-esteem and awareness. Additionally, negative attitudes toward traditional arranged marriages and a weakened authority of elders within families contribute to the ability of youth to make marital choices on the basis of personal preferences [37,38]. Chinese scholars have focused predominantly on the endogenous dynamics underlying the changing concept of marriage among contemporary youth. They analyze its impact on the postponement of marriage from the perspective of family authority and intergenerational transmission.

Moreover, the delay in educational attainment has led to shifts in gender roles within society. In traditional societies, men are the primary breadwinners, whereas women typically assume caregiving and child-rearing responsibilities [39]. As women gain greater economic independence in modern society, early marriage for financial stability is no longer necessary. This change is largely attributed to improvements in women's education levels and social status [40,41]. The “Action Plan for Revitalization of Education for the 21st Century [42]” issued by China in 1999 expanded access to nine-year basic education to over 80% of the population and significantly increased enrollment in higher education. This policy has influenced the age of first marriage, particularly among highly educated Chinese youth, who are more prone to remain unmarried. The impact is more pronounced for women, and data suggest that women over the age of 30 with higher education face greater challenges in establishing successful marriages [43–46]. In summary, the examination of social factors contributing to the postponement of marriage primarily focuses on the influence of specific indicators. However, it remains unclear whether other factors play a dominant role in shaping the delay in first marriage among Chinese youth in the 21st century.

Spatial differences in delayed age at first marriage

The delay in the age at first marriage among Chinese youth displays significant regional variations, driven by economic differentiation and social stratification. At the macro level, differences between urban and rural areas, as well as gender disparities, are evident. These differences are characterized mainly by the continual convergence of the probability of first marriage between males and females, alongside a prevalent trend of late marriage and non-marriage among rural males [47]. For example, over the past 30 years, the likelihood of first marriage for young women has decreased significantly, aligning with the age range typical for men. However, the proportion of lifelong unmarried men surpasses that of women. The gender imbalance in the birth cohort from 1984--2004, resulting in an 18 million shortfall of female babies compared with males in China—and the phenomenon of rural‒urban intermarriages further contributed to the marriage squeeze among rural males. Consequently, rural men have a lower probability and later expected duration of singlehood than their urban counterparts do [48]. This situation presents the most prominent issue in the urban‒rural disparity of first marriage patterns in China today. At the micro level, class differences accentuate the problem, specifically the conflict between the upper class's internal circulation of marriage resources and the aspirations of the middle and lower classes for upward mobility. In developed areas, some rural women find it challenging to attain upward mobility through marriage resources while being unwilling to accept similar or lower-class resources, resulting in consistent delays in their age at first marriage. On the other hand, disadvantaged males from the lower class seek partnerships with women from less developed or underdeveloped regions to alleviate their marital pressures. This imbalance in the transmission of marital pressure ultimately leads to the predicament faced by lower stratum males from less developed or underdeveloped areas, where they struggle to find marriage prospects or remain unmarried indefinitely. However, existing research on the age at first marriage among Chinese youth has focused predominantly on overall trends and urban‒rural and gender differences, failing to examine the differentiating characteristics and influential factors of delayed age at first marriage at the provincial level.

To summarize, existing research has concluded that the delay in the age of first marriage for young Chinese people is an inevitable process during the second demographic transition and that it will persist over the long term. There are many factors affecting the delay in age at first marriage, of which economic and social factors are the main ones. However, at present, there are obvious inter-provincial differences in the scene of first-marriage delays among youth groups brought about by China's unique dual urban-rural structure, the unevenness of regional demographic development, and the shift in the culture of marriage and childbearing. Under the reality of Chinese youth's more open, tolerant and autonomous view of marriage, it is necessary to explore whether the spatial and temporal differences in the completion of first marriage among youth groups are gradually widening from the provincial administrative scales with distinctive heterogeneity of marriage culture. Is there a similar trend in the change of age at first marriage between the whole and the parts of the country? To explore whether the factors and mechanisms influencing the delay in age at first marriage are more diversified and mutually reinforcing? These are the scientific questions that need to be addressed.

Therefore, this paper utilizes the data from China's three population censuses from 2000 to 2020 and the Western Region Life Tables published by the United Nations Population Division to analyze the gender, urban-rural, and regional differences in the pattern of first marriages in each province of China in 2020 by using the Multi-decreasing Table to compare and analyze the characteristics of the changing pattern of first marriages in China since the 21st century. We also utilize the 1996–2020 China Statistical Yearbook data to construct the panel data of 31 provinces in China for 25 years, to explore the influencing factors of the delayed age at first marriage through the dual fixed-effects model of region and time, and to explore the core driving mechanism of the delayed age at first marriage in China at this stage, with a view to making reasonable explanations and useful references for the special phenomenon of delayed age at first marriage in the second demographic transition of China which is different from that of the developed countries. This is to provide a reasonable explanation and useful reference for the special phenomenon of delayed first marriage in China's second demographic transition, which is different from that of developed countries.

Data sources

To construct the Net Nuptiality Table for different provinces, it is necessary to refer to the data on first marriages and the number of deaths among the unmarried population in each province. This study drew on established methodological approaches by leveraging mortality probabilities from the United Nations West Model Life Table for developing countries and used province-specific life expectancy by sex as the reference to derive proxy mortality rates for never-married populations [49–53]. We also leveraged urban-rural/age/sex/marriage-year-specific population data from the 2000–2020 population censuses to compute first-marriage probabilities, expected years in single state, and proportions remaining unmarried. It should be clarified that the number of age-specific first marriage was derived from marriage-year/age/sex tabulations in Volume 5 of China's census long-form micro-data (2000, 2010 and 2020). Since the 2020 census enumeration concluded in November 2020 and thus did not achieve full calendar-year coverage, we utilized 2019 data to ensure complete annual representation in nuptiality table construction.

To study the impact mechanism of delayed age at first marriage, this paper focuses on the period from 1996--2020, taking into account data availability and the concentrated period of China's second demographic transition. A total of 775 balanced panel data points is collected for the 31 Chinese provinces (excluding Hong Kong, Macao, and Taiwan). The base map used in Fig 3 was sourced from the Standard Map Service System of the National Geomatics Center of China [54].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The first marriage pattern of young people by gender in China from 2000 to 2020: (a) First Marriage probability, (b) Unmarried Remaining proportion curves and (c) the expected duration of singlehood since birth.

https://doi.org/10.1371/journal.pone.0338259.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. First marriage probability in different provinces of China: young men and young women in 2020.

https://doi.org/10.1371/journal.pone.0338259.g002

Variables selection

The dependent variable used was the average age at first marriage. This indicator reflected the average age at which first marriages occurred within a given year in a specific country or area. In this paper, the average age at first marriage by gender in 31 provinces from 1996 to 2020 was selected, with a sample size of 775. The average age at first marriage for the overall Chinese sample increased from 24.15 to 28.67 years old, a delay of nearly five years. It characterized the age structure of the population and the trend of age at first marriage during population development.

Independent variables included education and ethnic differences. The education variable was defined as the years of education, measured by the average number of years of education, which was calculated using the number of individuals in each year of education and the duration of each level, as indicated in the statistical yearbook. Ethnic differences were represented by the ethnic minority ratio, which indicated the proportion of ethnic minorities to the total population and reflected the cultural attributes of the region.

To address the issues of covariance and endogeneity in the measurement model, as well as to account for differences in economic development and population age structure across provinces and regions, three control variables were selected on the basis of the relevant literature [55–58]. The control variables chosen included regional development, living standards, and age structure. Specifically, regional development was measured by the proportion of the urban population to the total population, residents’ living standards were measured by per capita disposable income, and the population age structure was measured by the all-age dependency ratio, which reflected the burden of the population.

Net nuptiality table

Similar to life tables, marriage tables are a method used to calculate the expected years in single state by treating marital events as progressively decreasing forms within a population cohort. Net Nuptiality tables, on the other hand, are multi-decreasing life tables that consider both the probability of marrying and the probability of dying. This approach eliminates the influence of population age structure and mortality risk when calculating the probability of marriage, the proportion of unmarried individuals still alive, and the age-specific expected years remaining in the single state. In constructing the multi-decreasing first marriage table, reference was made to existing research findings [57], which identified that first marriages predominantly occur between exact ages 15 and 49. The process of constructing the multi-decreasing first marriage table is provided below:

First, the age-specific probability of first marriage and the age-specific probability of death in the multi-decreasing first marriage table need to be computed separately via the following formulas [53,59]:

(1)

(2)

where is the probability of first marriage at exact age x, converted from the rate of first marriages during the year during the year, and where is the probability of death at exact age x, estimated from the age-specific unmarried mortality rate . Because of the lack of statistics on the marital status of those who die, we usually use the probability of death in the year as a proxy when we assume that the occurrence of death is unrelated to the occurrence of marriage. To save space in the manuscript, the derivation process of the Details about the probability of first marriage is presented in Supporting Information S1.

Second, the number of unmarried survivors , the number of unmarried deaths , and the number of first marriages are further computed [60–62]. In a cohort of people born at the same time, since the number of first marriages and the number of unmarried deaths decreases with age, the academic community usually defines the initial number of people in the multi-decreasing first marriage table as 100,000 at the age of 0, i.e., , and the formula is as follows:

(3)

In Equation (3), the probability of a first marriage occurring before the exact age of 15 is extremely low; thus, at this stage, the number of unmarried survivors is also the number of remaining survivors, and only the diminishing effect of the number of deaths needs to be taken into account. It should be noted that the model specified in Equation (3) does not explicitly include two likelihoods: (a) to marry and then die within the same age interval, and (b) to die while about to get married. These were omitted following the common practice in nuptiality table construction, as the simultaneous estimation of these joint events requires detailed data that are typically unavailable. Given the very low mortality rates prevalent among the marriageable-age population in China, the exclusion of these joint events does not introduce substantial bias and is not expected to significantly affect the results for academic research purposes. This simplification is considered acceptable in demographic studies when focusing on marriage patterns rather than precise actuarial calculations [63]. Starting from exact age 15 up to exact age 49, the number of unmarried survivors needs to consider both the effects of death and first marriage, where is obtained from the product of the probability of the first marriage in Equation (1) and the number of unmarried survivors .

The average number of unmarried person-years lived by the marriageable cohort between exact ages 15 and 49 who remain unmarried until age is calculated as :

(4)

Notably, we assume that there are no more first marriages in the open age group from exact age 50 and above; thus, the average number of unmarried person-years in the corresponding first-marriage table is the number of deaths, , and by bringing the average number of unmarried person-years in each age group into Equation (5), the total number of unmarried person-years, , at age x can be further calculated:

(5)

Finally, we can find the average number of years that the unmarried surviving population at exact age x has been in the event of a first marriage, i.e., the expected years in the single state at age x, :

(6)

To show more clearly and visually the age at which the first marriage event occurs, we convert the expected years in single state at exact age 15 to the expected duration of singlehood since birth :

(7)

The complete Net nuptial table in Supporting Information S2.

Fixed effect model

Using balanced panel data for 31 provinces from 1996--2020, we conducted an analysis to identify the factors influencing the delay in the age of first marriage in China as a whole and by gender. The formulas used are shown below:

where represents the logarithm of the average age at first marriage in year t in province i, represents the logarithm of the average years of schooling in the corresponding year and province, represents the level of ethnocultural differences in the corresponding year and province, corresponds to the relevant control variable, represents the area fixed effect in province i, represents a dummy variable for year t, and denotes the randomized disturbance term. Compared with OLS regression estimation, the use of a fixed-effects model with time dummy variables allows for simultaneous control of the effects of time-varying regional factors that do not vary over time and macro factors that do not vary over time on the average age at first marriage.

Expected pre-marital singlehood duration

The overall probability of first marriage for females was generally higher than that for males (Fig 1a). The highest first marriage probability for both genders occurred in 2000, and the lowest in 2020. The peak first marriage probabilities for males and females were gradually converging, and the age at which the highest first marriage probability occurred was continuously being delayed. The proportion of unmarried individuals showed that in 2000, the proportion of unmarried females at age 49 was nearly 0%, while in 2020, it had increased to 3.9% (Fig 1b). In contrast, the proportion of unmarried males at age 49 was 2.3% in 2000, and by 2020, it had risen to 15.3%. The expected years in single state for both males and females in China had steadily increased (Fig 1c). In 2000, the expected of duration singlehood since birth for males was 26.85 years, and for females, it was 23.75 years, with a gap of 3.1 years. By 2020, the expected duration for males had risen to 33.63 years, and for females, it had reached 29.56 years, with the gap widening to 4.07 years.

Using cluster analysis, we categorized first-marriage probabilities by gender into low, medium, and high regional groupings (Fig 2). Females consistently showed higher first-marriage probabilities than males across age groups. The peak probability for males occurred at ages 26–30, with lower-probability provinces concentrated in central and southwestern China, while higher-probability clusters predominated in ethnic minority regions. For females, high-probability ages began at 24, declining after 30. Female low-probability provinces (predominantly in northeast China and eastern municipalities) were fewer than male counterparts, whereas high-probability areas extended beyond ethnic regions to include coastal (e.g., Shandong) and central provinces (e.g., Shanxi, Anhui).

In 2020, the expected duration of singlehood since birth was generally higher for males than for females (Fig 3). In the southern region, males were the latest (greater than 34.7 years). In the northern region, except for Heilongjiang, Jilin, and Liaoning, males were generally lower than 31.9 years. However, females in the municipalities were the highest (greater than 31.1 years). Provinces with a lower expected years in the single state for females were concentrated in the western of Xinjiang, Gansu, Qinghai, Ningxia, and the central of Inner Mongolia.

Our analysis of urban-rural disparities in expected years in the single state since birth by gender revealed distinct patterns: urban males exhibited fewer expected unmarried years than rural males nationwide, except in Beijing and Shanghai where this trend reversed (Fig 4). Conversely, urban females demonstrated greater unmarried expectancy than rural females in over half of provinces, with the most pronounced gaps in affluent eastern metropolises (Beijing, Shanghai, Tianjin). These spatial patterns reflect prolonged singlehood years among China's 2020 cohort with widening gender gaps, particularly concentrated in narrowing age windows for males. The persistence of fewer singlehood years among urban males and rural females contrasts sharply with delayed transitions among urban females in developed regions and rural males in disadvantaged areas, necessitating investigation of underlying mechanisms and key factors shaping these differentials.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Number of unmarried years experienced by young people in 2020.

Note: The base map used in Fig 3 was sourced from the Standard Map Service System of the National Geomatics Center of China (accessed January 19, 2025, available at: http://bzdt.ch.mnr.gov.cn/). Cartographic software: ArcGis 10.8.

https://doi.org/10.1371/journal.pone.0338259.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Urban–rural comparison of the number of years of unemployment experienced at aged 15.

https://doi.org/10.1371/journal.pone.0338259.g004

Descriptive statistics and regression results

Table 1 showed that the descriptive statistics results for the selected variables. The standard deviation of the average age at first marriage, both for the overall sample and by gender, is small. The mean age at first marriage for the total sample is 25.33, 26.32 for males, and 24.44 for females. The average years of education for the total sample is 8.26 years, 9.57 years for males, and 7.82 years for females.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Variable definitions and descriptive statistics in 1997–2020 (n = 775).

https://doi.org/10.1371/journal.pone.0338259.t001

After controlling for regional variations, each additional year of educational attainment delayed the age at first marriage by 0.22 years (β = 0.219, p < 0.01; Table 2), indicating that higher education prolongs singlehood through extended schooling and career establishment. Conversely, although the effect of ethnic minority proportion was statistically significant at both national (β = −0.041, p < 0.05) and female-specific levels (β = −0.037, p < 0.05), its demographic impact was negligible.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Area fixed effects model regression results: total sample, male sample and female sample in 1997-2020 (n = 775).

https://doi.org/10.1371/journal.pone.0338259.t002

After incorporating temporal effects via regional fixed effects (Table 3), although educational attainment and ethnic minority composition remained significant predictors of first-marriage timing, the effect size of education on age at first marriage was significantly reduced (β = 0.066, p < 0.01 for males; β = 0.062, p < 0.01 for females). Similarly, despite statistical significance, the impact of ethnic minority proportion was demographically negligible at the national level (β = −0.022, p < 0.05 for females).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Two-factor fixed effects model regression results: total sample, male sample and female sample in 1997-2020 (n = 775).

https://doi.org/10.1371/journal.pone.0338259.t003

The Global Moran's Index was no significant spatial autocorrelation of the age at first marriage at the provincial level in China (see Supporting Information S3). Considering the differences in education levels and ethnic cultures in China, we divided three regions (i.e., Eastern, Central, and Western, See Supporting Information S4) for the analysis of how the average years of education and ethnic culture influenced the age at first marriage. The Eastern included Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, and Hainan. The Central included Shanxi, Inner Mongolia, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan, and Guangxi. The Western included Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang. The results of the influence of education on the age at first marriage (Table 4) showed that the delaying effect of average years of education on the age at first marriage was weaker in the Eastern compared to other regions (−0.052, P ≤ 0.01), while in the Central, this effect was stronger (0.07, P ≤ 0.01). No significant correlation was found in the Western. Additionally, the regional differences in the age at first marriage were consistent for young males and young females.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. The analysis of regional differences between education and age at first marriage.

https://doi.org/10.1371/journal.pone.0338259.t004

The results of the influence of ethnic culture on the age at first marriage (Table 5) showed that in the Eastern and Central, an increase in the proportion of ethnic minorities would weaken the delaying effect on the age at first marriage (East: −0.957, P ≤ 0.01; Central: −0.586, P ≤ 0.01). In the Western, however, an increase in the proportion of ethnic minorities would strengthen the delaying effect on the age at first marriage (0.07, P ≤ 0.01). Additionally, both males and females exhibited characteristics consistent with the national overall trend.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. The analysis of regional differences between ethnic and age at first marriage.

https://doi.org/10.1371/journal.pone.0338259.t005

On the other hand, when the minority ratio is considered, an increase in the proportion of minorities in eastern and central China has a strengthening effect on the reduction in the age at first marriage, whereas the western region has a weakening trend. This can be explained by similar characteristics observed in the average years of schooling. Specifically, the western region is home to a large number of China's ethnic minorities, which constitute a greater proportion of the total population than other regions do. Additionally, the western region includes the most diverse range of ethnic minorities. Consequently, when focusing solely on the impact of the ethnic minority ratio on the delayed age at first marriage in the western region, the similarity in their ethnic minority marriage cultures diminishes the effect it produces.

Further analysis of the regional heterogeneity of the influence mechanism on age at first marriage reveals consistent characteristics in the Chinese male and female cohorts across the geographic gradient. The average years of schooling have a weakening effect on delaying first marriage in the eastern region but a strengthening effect in the central region for both males and females. This can be attributed to the higher level of economic development and social and cultural openness in the eastern region, which provides more opportunities for young people to receive higher education and vocational training. In addition to individual education level, other social factors, such as competitive preferences and high real estate prices, also contribute to weakening the effect of education on age at first marriage postponement, to some extent. Conversely, the central region has weaker educational resources, more conservative economic development, and cultural openness, which makes educational attainment more crucial for the social status and development expectations of the different group, resulting in a stronger effect of education on age at first marriage postponement than in the eastern region. Moreover, the ethnic minority ratio weakens the delay in the age at first marriage in the western region but strengthens it in both the eastern and central regions.