China’s family planning policies and their labor market consequences

Abstract

China initiated its family planning policy in 1962 and its one-child policy in 1980, and it allowed all couples to have two children as of 1 January 2016. This paper systematically examines the labor market consequences of China’s family planning policies. First, we briefly review the historical evolution of China’s family planning policies and the existing literature. Second, we investigate the effects of these policies on the labor market, focusing on the size and quality of the working-age population and its age and gender composition. We give special attention to regional differences in the demographic structure resulting from the interaction of the family planning policies and internal migration. Finally, we discuss ongoing and prospective policy changes and their potential consequences. Although urban areas and coastal provinces have implemented stricter family planning policies, our analysis shows that because of internal migration, the aging problem is more severe in rural areas and in inland provinces. Our simulation results further indicate that the new two-child policy might fall short of pulling China out of its aging situation.

Appendices

Appendix 1: Data sources and variable definitions

1. Data sources:

The 1 % sample of the 1982 Population Census data of China (observations 10,037,508) was conducted by China’s National Bureau of Statistics on 1 July 1982.

The 1 % sample of the 1990 Population Census data of China (observations 11,475,065) was conducted by China’s National Bureau of Statistics on 1 July 1990.

The 0.1 % sample of the 2000 Population Census data of China (observations: 1,180,111) was conducted by China’s National Bureau of Statistics on 1 November 2000.

The 20 % sample of the 2005 1 % National Population Sample Survey of China (Mini-Census, observations: 2,585,481), part of the China census program, is an inter-census survey administered by the National Bureau of Statistics in China on 1 November 2005.

Tabulation on the 2010 Population Census of the People’s Republic of China, compiling the results of the 2010 Population Census conducted by China’s National Bureau of Statistics on 1 November 2010.

China city statistical yearbook 1991 is an annual statistical publication that comprehensively reflects the development of the economy and society of China’s cities. The following variables used for analysis are from the yearbook: log GNI in 1990, log fixed capital investment in 1990, the number of college students per 10,000 people in 1990, and the number of medical doctors per 10,000 people in 1990.

2. Definition of Variables:

Birth rates in 1981–1985 are defined as the number of new births in the corresponding years per 1000 people.

Log population in 1990 is the log number of observations at the prefecture city level.

Average education in 1990 refers to the average years of schooling at the prefecture city level.

Percentage of employment in manufacturing in 1990 is the ratio of the number of workers employed in the manufacturing industry to the total employment at the prefecture city level.

Unemployment rate in 1990 is defined as the ratio of the number of non-working persons seeking jobs to total labor force for urban hukou population at the prefecture city level.

Proportion of Han Chinese in 1990 is the share of Han Chinese in the total population at the prefecture city level.

A migrant is defined as a person whose place of hukou registration is different from his or her place of living.

Migration flow is the number of migrants between pairs of prefecture cities.

In-migration rate in 2005 is the ratio of the number of migrants who move into a prefecture city to the total hukou population of the prefecture city.

Out-migration rate in 2005 is the ratio of the number of migrants who move out of their registered prefecture city to the total hukou population of the prefecture city.

Percent of population aged 65+ refers to the percentage of the population aged 65 and above at the prefecture city level.

Sex ratio in Table 4 is the sex ratio in a prefecture region, that is, the ratio of sampled males to females who had rural hukou in the prefecture region and who were born before November 2004.

Self-reported good health equals one if respondents report that they are healthy or equals zero if respondents report that they have some or many health issues in their work and daily lives.

Appendix 2: Supplementary figures and tables

Fig. 14

One-child policy regulatory fines in 1980–2000, by province. (Ebenstein 2010)

Table 5 Initial years of being allowed to have a second birth, by province and specific population group

Table 6 Demographic and economic characteristics of prefecture cities

Table 7 Descriptive statistics of control variables in Table 4

Table 8 China’s mortality rates by gender and age group, deaths per 1000 people

Appendix 3: Procedures of simulations

1. Mortality rates used for simulations

Table 8 shows China’s mortality rates by gender and age group in 1973–1975, 1981, 1990, 2000, and 2010. The 1973–1975 rates are the average rates of the 3 years.

Mortality rates of a year of age are assumed to be identical to the rates of the age group which the year of age is in. We assume the mortality rates in 1970–1972 are the same with the average mortality rates in 1973–1975. We interpolate the mortality rates in 1976–1980 by assuming the rates change linearly from 1975 to 1981. Similarly, we interpolate the mortality rates in 1982–1989, 1991–1999, and 2001–2009, assuming that rates change linearly over years. The mortality rates after 2010 are assumed to be identical to the 2010 rates.

2. Simulations in Fig. 3

• Step 1: Extrapolate backward the population size of 1970 by gender and age (0–99 years) according to the gender-age population structure in the 1 % sample of the 1982 Population Census data, assuming that mortality rates from 1970 to 1982 are those described in Table 8.

• Step 2: Calculate the lifetime number of births for each birth cohort of women, beginning with the cohort of 1918, based on actual birth records from the samples of the 1982, 1990, and 2000 Population Census and the 2005 Mini-Census. Starting from the cohort of 1971, the lifetime number of births is assumed to decrease by 0.1 for every 10 years of cohorts (e.g., 1.6 for birth cohorts 1971–1980, 1.5 for birth cohorts of 1981–1990, etc.).

• Step 3: Had there been no family planning policies in history, the lifetime number of births for each cohort of women is assumed to increase by 1, and the sex ratio at birth is assumed to be 106 boys to 100 girls. The lifetime number of births of a woman is distributed to each year of age of the woman based on the probability of childbearing by age of women described in Wang (2015). The probability of childbearing by age is derived from detailed birth records of the China Health and Nutrition Survey, and is created separately for four cohort groups: cohorts of 1950 and older, 1951–1960, 1961–1970, and 1971 and younger. All the four probability distributions show zero chance of childbearing below 15 or above 49, and bell-shaped probability curves between 15 and 49 with peaks around 22–24.

• Step 4: Starting from the population structure in 1970, we first derive the population size by gender and age in 1971, assuming that a part of the population die between 1970 and 1971 following the mortality rates in Table 8. Particularly, the people aged 99 in 1970 are completely removed in 1971, so that the upper bound of age in 1971 remains 99. Then, we calculate the number of newborn boys and girls in 1971, based on the number of women in 1971, the number of births of each woman in 1971, and the sex ratio at birth. Similarly, we simulate the population size by gender and age in a year based on the population structure in the previous year, and finally, we create the population pyramids in 1990, 2000, and 2010 in Fig. 3.

3. Simulations in Figs. 10 and 11

• Step 1: Predict the population structure in 2015 based on the population structure in the 2010 population census, using the actual fertility rates and the same procedure described in Section 2 of this appendix.

• Step 2: Assume the lifetime number of births is added by 0.3, 0.5, or 0.7. With the new fertility rates and the same procedure as in Section 2 of this appendix, we start from the population structure of 2015 and derive the population size and the elderly dependency ratio every 5 years until 2050.

4. Simulations in Fig. 13

All steps are the same with Section 3 of this appendix, except that women’s lifetime births are assumed to increase by 1.

Appendix 4: A conceptual model of migration

Suppose there are two cities in the economy: A and B. Each city has a representative firm producing a homogeneous good with the same production function g(z _i , L _i ), i = A, B, where L _i is the labor input in city i and z _i incorporates all technologies that affect labor productivity in city i. Let $\frac{\partial g\left({z}_i,{L}_i\right)}{\partial {z}_i}>0,\;\frac{\partial g\left({z}_i,{L}_i\right)}{\partial {L}_i}>0,\kern0.24em \frac{\partial^2g\left({z}_i,{L}_i\right)}{\partial^2{L}_i}<0$. Assume that each worker offers one unit of labor, and therefore, L _i also represents the working-age population in city i. The L _i working-age people in the current period were born in the last period by the last-period working-age population, L _i,− 1. For simplicity, we assume that (1) the L _i population will not become labor forces until the current period, and (2) the L _i,− 1 population completely exit the labor markets in the current period. In other words, in each period, only one generation of population serves as the labor forces. We assume L _i  = L _i,− 1 f _i , where f _i is the last-period fertility rates in city i.

Trade in products is costless, which equalizes prices across regions. We normalize the price of products to be 1. Firms maximize profits as price takers. Without migration between the two cities, the equilibrium real wage in city i satisfies ${w}_i=\frac{\partial g\left({z}_i,{L}_i\right)}{\partial {L}_i}$. Assume that migration imposes a cost c > 0 on each migrant. There will be migration in the equilibrium if and only if $\left|\frac{\partial g\left({z}_A,{L}_A\right)}{\partial {L}_A}-\frac{\partial g\left({z}_B,{L}_B\right)}{\partial {L}_B}\right|>c$. Under such a condition, the equilibrium real wages with migration satisfies |w ^m _A  − w ^m _B | = c. Without loss of generality, we assume $\frac{\partial g\left({z}_A,{L}_A\right)}{\partial {L}_A}-\frac{\partial g\left({z}_B,{L}_B\right)}{\partial {L}_B}>c$, and then people migrate from city B to city A. This condition implies that technologies are more advanced in city A or/and that the size of labor is smaller in city A.

Assume the size of migrants is M. In the equilibrium, we have

$$ \frac{\partial g\left({z}_A,{L}_A+M\right)}{\partial \left({L}_A+M\right)}-\frac{\partial g\left({z}_B,{L}_B-M\right)}{\partial \left({L}_A-M\right)}=c $$

Therefore, $M$ can be expressed as the following function of other factors.

$$ M=m\left({z}_A,{z}_B,{L}_A,{L}_B,c\right)=m\left({z}_A,{z}_B,{L}_{A,-1},{L}_{B,-1},{f}_A,{f}_B,c\right) $$

The migration flow between city A and B depends on technologies affecting labor productivity, the initial working-age population, birth rates in both cities and migration costs.

Suppose that the birth rate in city B is f _B  = f. City $A$ has a negative fertility shock, f _A  = f − d. Then, we have

$$ \frac{\partial g\left({z}_A,{L}_{A,-1}\left(f-d\right)+M\right)}{\partial \left({L}_{A,-1}\left(f-d\right)+M\right)}-\frac{\partial g\left({z}_B,{L}_{B,-1}f-M\right)}{\partial \left({L}_{B,-1}f-M\right)}=c $$

Taking derivative with respect to d on both sides, we can get

$$ \frac{\partial M}{\partial d}=\frac{\frac{\partial^2g\left({z}_A,{L}_A+M\right)}{\partial^2\left({L}_A+M\right)}}{\frac{\partial^2g\left({z}_A,{L}_A+M\right)}{\partial^2\left({L}_A+M\right)}+\frac{\partial^2g\left({z}_B,{L}_B-M\right)}{\partial^2\left({L}_B-M\right)}}{L}_{A,-1}>0, $$

which implies that, given other factors being constant, a negative fertility shock in city $A$ increases the marginal product of labor in city A and thus induces a higher volume of migration flow from city B to city A.