Descriptive statistics
In this study, fertility intention was measured as a desire to have a second child or ideal number of children. Table 1 shows the descriptive statistics for the analytical sample. The full distributions of variables used in the analysis are reported. For fertility intentions for a second child under the universal two-child policy, 3362 women were sampled, of which 84% (n=2855) did not intend to have another child; 13.2% (n=443) intended to have another child, and 1.9% (n=64) would like to have two or more children (shown in Figure 1). Among women aged 25–44 years that already had one child, 82.1% (n=1841) did not intend to have another child and 17.9% (n=401) intended to have another child.
Participants’ fertility intentions with no conditional constraints were measured by the question, “If you had no need to consider the fertility policy, socioeconomic status, or physical situation, how many children in a family do you think would be perfect?” Four response options were provided: “zero (no intention/childless),” “one,” “two,” and “more than two.” The distributions of answers were 1.1%, 7.5%, 78.4%, and 13% for the four options, respectively. The average educational attainment (years) of the women, their father, and their mother were 8.88, 6.79, and 5.47 years, respectively.
Table 1. Descriptive analysis of the sample
Variables
|
average
|
St.dev
|
Min
|
Max
|
Samples
|
Female ideal number of children Fer_ideal
|
2.033
|
0.498
|
0
|
10
|
3388
|
Female fertility intention for second child Fer_contr
|
0.171
|
0.427
|
0
|
5
|
3362
|
Parent’s education Pedu
|
12.577
|
6.700
|
3
|
16
|
3388
|
Parent’s organizational type when participant at 14 age Porgan_type
|
2.645
|
0.727
|
1
|
3
|
3388
|
Parent’s marital status when participant at 14 age Pmarit_stat
|
0.030
|
0.169
|
0
|
1
|
3388
|
Siblings size SibSize
|
6.853
|
7.254
|
0
|
14
|
3388
|
Hukou when participant born Hukou_born
|
0.883
|
0.322
|
0
|
1
|
3388
|
Participant’s education Dedu
|
12.973
|
6.468
|
3
|
22
|
3388
|
Participant’s marital status Dmarit_stat
|
0.082
|
0.274
|
0
|
1
|
3388
|
Participant’s organizational type
Dorgan_type
|
2.475
|
0.684
|
1
|
3
|
3388
|
Full time job Fulltime
|
0.643
|
0.479
|
0
|
1
|
3388
|
Participant’s political status Polit_stat
|
0.046
|
0.210
|
0
|
1
|
3388
|
Participant’s Hukou Hukou_growup
|
0.736
|
0.441
|
0
|
1
|
3388
|
Participant’s age Age
|
38.697
|
7.799
|
15
|
49
|
3388
|
Participant’s birth cohort Cohort
|
1977.303
|
7.799
|
1967
|
2001
|
3388
|
Notes: N=3388. Source: China Labor-force Dynamics Survey 2016.
Figure 1. Women’s fertility intentions for another child under the universal two-child policy in the 2016 China Labor-force Dynamics Survey
Fertility intentions for a second child under the universal two-child policy among women who already had one child
Table 2 shows the ZIP model estimates including the zero and count parts of models. I ran four models: Model 1 included family of origin characteristics; Model 2 was Model 1 controlled for daughter’s education and organizational type; Model 3 was Model 2 further controlled for daughters’ workload and marital status; and Model 4 considered Model 3 and daughters’ political status. The number of siblings (family structure in childhood) was positively and significantly associated with women’s fertility intentions for a second/another child. Women with more siblings were likely to plan a larger family size compared with their counterparts with none or one sibling. This meant there was intergenerational transmission of childbearing cultures in the context of contemporary China.
Daughters’ and parents’ education were significantly negatively associated with the no more children intention, and significantly positively associated with the intention to have a second/another child. Parents’ educational level contributed to their daughters’ fertility intentions, perhaps by communication, observation, or imitation. However, the influence of daughters’ education was much stronger than that of parents: odds ratio 0.060 (for zero or no child) and −0.595 (for a second/another child) compared with parents (0.045 and −0.348, respectively). Although the effect of a daughter’s educational attainment on her own fertility intentions was complex, women who invested more in education had less desire to have another child than their less-educated counterparts. The bidirectional time squeeze between family life and occupational career may mean that more skills are needed to deal with this relationship for women with higher education.
In China, working for the government, public institutions, or state-owned enterprises represents high social status, a stable income, and good working conditions. To some extent, organizational job types can be used to measure family conditions. Under those conditions, parental influence was significantly and negatively associated with daughters’ no-child intention, whereas for daughters, it was significantly and positively associated with another-child intention. This suggested women who were born in a good family and had good SES conditions were much more likely to have another child than other women. From the perspective of birth environments (Hukou, registered residence), it was also confirmed that women from a rural origin (probably born in a relatively poor conditions) were more likely to prefer no more children compared with their urban counterparts. Daughters’ and parents’ marital status, daughters’ political status, and workload (full-time or not) did not influence fertility intentions regarding no more children or a second/another child.
Table 2. Fertility intentions for a second/another child under universal two-child policy among women who already had a child
|
(1)
|
(2)
|
(3)
|
(4)
|
|
0
|
1+
|
0
|
1+
|
0
|
1+
|
0
|
1+
|
Family Background Characteristics
|
SibSize
|
-0.084*
(0.050)
|
0.549**
(0.239)
|
-0.039
(0.047)
|
0.550**
(0.220)
|
-0.034
(0.047)
|
0.589***
(0.227)
|
-0.035
(0.048)
|
0.590**
(0.239)
|
Pedu
|
0.065**
(0.026)
|
-0.649***
(0.190)
|
0.046*
(0.025)
|
-0.322**
(0.164)
|
0.045*
(0.025)
|
-0.343**
(0.171)
|
0.045*
(0.027)
|
-0.348*
(0.210)
|
Porgan_type
|
|
|
|
|
|
|
|
|
government, public institutions or state-owned enterprise
|
-0.961***
(0.203)
|
0.409
(1.167)
|
-0.865***
(0.204)
|
1.104
(0.994)
|
-0.857***
(0.204)
|
1.22
(1.036)
|
-0.855***
(0.214)
|
1.245
(1.206)
|
Others(ref.)
|
|
|
|
|
|
|
|
|
Hukou_born
|
|
|
|
|
|
|
|
|
rural
|
0.293
(0.202)
|
-2.068**(1.029)
|
0.446**
(0.190)
|
-1.512
(1.015)
|
0.472**
(0.188)
|
-1.584
(1.052)
|
0.472**
(0.189)
|
-1.592
(1.085
|
Unrural/Urban(ref.)
|
|
|
|
|
|
|
|
|
Pmarit_stat
|
|
|
|
|
|
|
|
|
First married(ref.)
|
|
|
|
|
|
|
|
|
others
|
0.342
(0.209)
|
-2.793
(2.717
|
0.341
(0.221
|
-2.400
(2.037)
|
0.362
(0.225
|
-2.076
(2.021)
|
0.367
(0.229)
|
-2.038
(2.056)
|
Daughter’s Characteristics
|
Dedu
|
|
|
0.049**
(0.024
|
-0.619**
(0.177)
|
0.058**
(0.025)
|
-0.599***
(0.176)
|
0.060**
(0.026)
|
-0.595***
(0.201)
|
Dorgan_type
|
|
|
|
|
|
|
|
|
government, public institutions or state-owned enterprise
|
|
|
-0.324*
(0.185)
|
2.594**
(1.059)
|
-0.272
(0.187)
|
2.542**
(1.081)
|
-0.254
(0.210)
|
2.587*
(1.354)
|
Others(ref.)
|
|
|
|
|
|
|
|
|
Fulltime
|
|
|
|
|
|
|
|
|
Yes(ref.)
|
|
|
|
|
|
|
|
|
No
|
|
|
|
|
0.190
(0.133)
|
-0.251
(0.732)
|
0.188
(0.133)
|
-0.256
(0.741)
|
Dmarit_stat
|
|
|
|
|
|
|
|
|
First married(ref.)
|
|
|
|
|
|
|
|
|
others
|
|
|
|
|
-0.031
(0.235)
|
-0.653
(1.646)
|
-0.033
(0.237)
|
-0.68
(1.770)
|
Polit_stat
|
|
|
|
|
|
|
|
|
Independents(ref.)
|
|
|
|
|
|
|
|
|
member of Chinese Communist Party
|
|
|
|
|
|
|
-0.083
(0.273)
|
-0.061
(3.259)
|
Note: a. Dependent variables were women’s fertility intentions under the universal two-child policy (Fer_contr) and all estimated by zero-inflated Poisson models. b. standard errors in parentheses. c. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively.
Fertility intentions for a second child tested by Hukou, age, and interactions between education and sibsize
In Table 3, daughters’ Hukou in adulthood, age, and the interaction effect between sibling size and daughter’s educational attainment were added to Model 4 to form Models 5, 6, and 7, respectively. Age had a significant negative effect on the no more children intention and a significant positive effect on another child intention. It may be that women had to take their last chance to give birth to another child before biological limits occurred. Compared with their rural counterparts, urban women who already had one child were more likely to have a second/another child, which was significant at a 10% critical value. Interactions between women’s education and the number of siblings were added to test if family size affected children’s educational attainment. The results were non-significant. Previous research in different cultures and regions found a negative relationship between sibsize and educational attainment (Blake 1981; Hanushek 1992; Knodel and Wongsith 1991). However, some different conclusions were reported after parents’ preferences and family characteristics were considered (e.g., Angrist et al. 2010), although there was no causal link between family structure and child’s educational attainment. There was no obvious evidence for the existence of gender inequality in family resource allocation. In China, women’s socioeconomic position and reputation have greatly improved in recent decades.
Table 3. Regressions added by Hukou, age, and interactions between education and sibsize
|
(5)
|
(6)
|
(7)
|
|
0
|
1+
|
0
|
1+
|
0
|
1+
|
Hukou_growup
Non-rural (rural as ref.)
|
-0.500**
(0.230)
|
2.005*
(1.044)
|
|
|
|
|
Age
|
|
|
-0.079***
(0.015)
|
0.465***
(0.068)
|
|
|
SibSize*Dedu
|
|
|
|
|
0.011
(0.014)
|
0.153
(0.101)
|
Other variables
|
Controlled
|
Controlled
|
Controlled
|
Note: a. Dependent variables were women’s fertility intentions under the universal two-child policy (Fer_contr); other variables were the same as Table 2 column 4. All were estimated by zero-inflated Poisson models, but have been condensed because of limited space. b. Standard errors in parentheses. c. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively.
Fertility intentions without situational or institutional constraints
Results from the multinomial logistic regression analyses are presented in Table 4. I estimated fertility intentions under no conditional constraints (i.e., ideal childbearing expectation). Model 8 captured the association between family of origin background and women’s fertility intentions. This basic model served as a starting point to determine if and how the association changed when variables of interest were added. In Model 9, I added women’s educational attainment as an explanatory variable to see whether fertility intentions differed by level of education in different cohorts. This considered fertility intentions without situational constraints, so women’s SES characteristics were not controlled, except for education.
Women’s fertility intentions were classified in four categories: zero (no fertility intention/no birth expectation), one child, two children (as the reference group), and three or more children. The number of siblings was positively and statistically significantly associated with women’s intended number of children. Women with two or more siblings were more likely to plan a larger family size than their counterparts with only one brother/sister.
This paper considered the differentials of educational effect in the inter-generational context. Parents’ education had a negative effect on the possibility of a daughter having three or more children. If parents had attained a medium or high educational level, their daughter would have a lower probability of having three or more children. After daughters’ education was added, the effect of parents’ education decreased. The higher the daughters’ education (e.g., higher than junior high school), the lower the intended number of children. This result was consistent with those reported by Testa et al. (2017). With increased educational attainment, women may learn more about contraceptive techniques, and may also perceive a higher opportunity cost than their counterparts with a low education.
The possibility of no fertility intention for women with a primary school education was much lower than for those with a junior high school education at a significance level above 10%. The possibility of expecting one child for women with a senior high school education was much higher than those with a junior high school education at a significance level above 5%. Women with a lower education level (illiterate, primary school) were more likely to expect three or more children than their higher-educated counterparts (higher than junior high school), especially in situations where there was a wide gap in educational attainment between parents and daughters; for example, parents were lower-educated (e.g., illiterate) but daughters were higher-educated (higher than senior high school), where the daughters’ fertility intention was much lower.
Women may readjust their intended fertility plan over their life course (Liefbroer 2009), so a cohort effect was considered. For the sampled women, various historical events experienced in childhood might have affected their fertility intentions in adulthood. For example, the Cultural Revolution (1966–1976), Reform and Opening (since 1978), and one-child policy (preliminary introduction period 1977–1981, initial stage of implementation 1982–1986) were taken as historical events. Women born in 1977–1986 had stronger fertility intentions than their counterparts born in 1967–1971, and even 1972–1976, who were inclined to have one child. The fertility policy changed from encouraging fertility after the People’s Republic of China was established to controlling the population in the 1980s, with the enforcement of the one-child policy. Although the one-child policy had been implemented which would be a hard startpoint for the former(born in 1977–1986), the former then benefited from Reform and Opening and enjoyed a better socio-economic condition during the process of their growup, compared to the later(born in 1967–1971) who encountered Cultural Revolution, suffered a nationwide starvation and had painful memories about childhood. A possible reason for this was that women’s fertility intentions were affected by their childhood experience and increased with the improvement of their living situation, which was consistent with Easterlin’s relative income hypothesis. The fertility control policy suppressed women’s fertility expectations in adulthood, which may relate to the improvement in living standards, although resistance to birth control occurred in the same time. In addition, daughters whose parents were divorced, cohabitating, remarried, or widowed would prefer to be childless or had a higher desire to reproduce compared with their counterparts whose parents were in their first marriage, possibly stemming from psychological insecurity or compensation.
Table 4. Influence of childhood family background on women’s ideal fertility intentions estimated by multinomial logistic regression
|
|
(8)
|
|
|
(9)
|
|
|
0
|
1
|
3 or more
|
0
|
1
|
3 or more
|
Pedu
|
illiterate
|
0.457
(0.431)
|
-0.102
(.196)
|
-0.091
(0.132)
|
0.478
(0.471)
|
-0.011
(0.207)
|
-0.387***
(0.145)
|
primary school completed(ref.)
|
|
|
|
|
|
|
junior high school completed
|
-0.148
(0.459)
|
0.173
(0.171)
|
-0.290**
(0.148)
|
-0.225
(0.462)
|
0.163
(0.174)
|
-0.193
(0.150)
|
senior high school completed or higher level
|
-0.240
(0.610)
|
-0.084
(0.224)
|
-0.374*
(0.201)
|
-0.316
(0.631)
|
-0.137
(0.228)
|
-0.181
(0.207)
|
SibSize
|
0
|
0.391
(0.731)
|
0.288
(0.283)
|
-0.285
(0.347)
|
0.329
(0.748)
|
0.269
(0.286)
|
-0.214
(0.354)
|
1
|
0.503
(0.485)
|
0.284
(0.195)
|
-0.501**
(0.216)
|
0.515
(0.486)
|
0.268
(0.196)
|
-0.461**
(0.217)
|
2(ref.)
|
|
|
|
|
|
|
3
|
0.024
(0.526)
|
-0.004
(0.196)
|
0.394**
(0.161)
|
-0.031
(0.526)
|
0.023
(0.197)
|
0.370**
(0.163)
|
4
|
-1.552
(1.071)
|
-0.418*
(0.242)
|
0.354**
(0.178)
|
-1.561
(1.071)
|
-0.406*
(0.243)
|
0.324*
(0.180)
|
5+
|
0.551
(0.531)
|
-0.491**
(0.249)
|
0.730***
(0.167)
|
0.542
(0.536)
|
-0.469*
(0.250)
|
0.665***
(0.168)
|
Dedu
|
illiterate
|
|
|
|
-0.174
(0.639)
|
-0.268
(0.345)
|
1.000***
(0.185)
|
primary school completed
|
|
|
|
-0.903*
(0.532)
|
0.156
(0.187)
|
0.435***
(0.135)
|
junior high school completed(ref.)
|
|
|
|
|
|
|
senior high school completed
|
|
|
|
-1.142
(0.764)
|
0.466**
(0.202)
|
-0.542**
(0.234)
|
junior college or higher level completed
|
|
|
|
0.214
(0.556)
|
0.302
(0.229)
|
-0.569**
(0.262)
|
Porgan_type
|
|
|
|
|
|
|
Selfemployer(main peasant)(ref.)
|
|
|
|
|
|
|
government, public institutions or state-owned enterprise
|
-0.343
(0.709)
|
-0.080
(0.236)
|
-0.444*
(0.229)
|
-0.461
(0.730)
|
-0.136
(0.240)
|
-0.257
(0.232)
|
others
|
-18.159
(4875.770)
|
0.239
(0.265)
|
0.205
(0.252)
|
-18.234
(4776.797)
|
0.204
(0.269)
|
0.370
(0.259)
|
Hukou_born
|
|
|
|
|
|
|
Rural(ref.)
|
|
|
|
|
|
|
Non-rural
|
0.499
(0.688)
|
0.441*
(0.235)
|
-0.408
(0.275)
|
0.420
(0.720)
|
0.351
(0.241)
|
-0.135
(0.286)
|
Pmarit_stat
|
|
|
|
|
|
|
First married(ref.)
|
|
|
|
|
|
|
others
|
1.439**
(0.636)
|
0.390
(0.390)
|
1.330***
(0.236)
|
1.463**
(0.645)
|
0.400
(0.393)
|
1.289***
(0.238)
|
Cohort
|
1967-1971(ref.)
|
|
|
|
|
|
|
1972-1976
|
0.762
(0.581)
|
-0.109
(0.187)
|
-0.114
(0.145)
|
0.727
(0.584)
|
-0.110
(0.188)
|
-0.106
(0.146)
|
1977-1981
|
1.244**
(0.590)
|
-0.361
(0.221)
|
0.210
(0.162)
|
1.143*
(0.599)
|
-0.388*
(0.225)
|
0.338**
(0.165)
|
1982-1986
|
1.029
(0.652)
|
-0.420*
(0.239)
|
0.339*
(0.174)
|
0.851
(0.662)
|
-0.434*
(0.244)
|
0.532***
(0.180)
|
1987-1991
|
1.174*
(0.669)
|
-0.224
(0.238)
|
-0.474**
(0.235)
|
0.992
(0.683)
|
-0.268
(0.246)
|
-0.201
(0.241)
|
1992-1996
|
1.718**
(0.789)
|
0.119
(0.326)
|
-0.131
(0.326)
|
1.600**
(0.809)
|
0.077
(0.332)
|
0.178
(0.332)
|
1997-2001
|
-16.915***
(0.000)
|
-0.811
(1.060)
|
0.242
(0.792)
|
-16.816***
(0.000)
|
-0.923
(1.066)
|
0.484
(0.803)
|
intercept
|
-5.316***
(0.639)
|
-2.226***
(0.203)
|
-1.914***
(0.168)
|
-4.903***
(0.668)
|
-2.342***
(0.228)
|
-2.173***
(0.190)
|
Nagelkerke R2
|
|
0.079
|
|
|
0.102
|
|
Note: a. Dependent variables were fertility intentions without any situational or institutional constraints (Fer_ideal). b. Standard errors in parentheses. c. *, **, and *** represent significance at the 10%, 5%, and 1 % level, respectively.