Does Agglomeration Discourage Fertility? Evidence from the Japanese General Social Survey 2000–2010

This study empirically investigates whether agglomeration discourages married couples' fertility decisions. Exploiting Japanese social survey data to control for economic and social factors underlying fertility, it examines whether agglomeration affects completed fertility and the timing of childbirth. Results indicate that agglomeration impedes completed fertility. In addition, this study finds that agglomeration delays young married couples’ fertility decision, and that they bear children later in life.


RIETI Discussion
Papers Series aims at widely disseminating research results in the form of professional papers, thereby stimulating lively discussion.The views expressed in the papers are solely those of the author(s), and neither represent those of the organization to which the author(s) belong(s) nor the Research Institute of Economy, Trade and Industry.

Introduction
Recent literature in economic geography has emphasized great benefits of agglomeration, such as higher productivity and faster human capital accumulation in denser areas (e.g., Combes et al., 2008Combes et al., , 2010Combes et al., , 2012;;Glaeser and Maré, 2001; Glaeser and Resseger, 2010; de la Roca and Puga, 2012).The economics of agglomeration attracts economists and policymakers hoping to understand spatial inequalities in wage and firm productivity.However, we still lack an adequate understanding whether the costs of agglomeration affect demographic structures, although Sato (2007) mentions a strong connection between agglomeration and fertility rate.Does agglomeration discourage fertility?This study empirically answers this question by examining the agglomeration effects on the completed fertility (i.e., the total number of children that married couples have during their lifetimes) and the timing of childbirth.
Fertility rates in most developed countries have rapidly declined along with economic growth, and raising fertility rates has become a policy priority.Figure 1 presents trends of total fertility rates in Panel (a) and share of population aged 65 and above in Panel (b) for selective OECD countries.France, Germany, the United Kingdom, and the United States have experienced sharp declines in total fertility rates after the 1960s, as have Italy and Japan after the 1970s.Although recent total fertility rates have steadily remained at approximately 2 births per woman in France, the United Kingdom, and the United States, the rates are less in Germany, Italy, and Japan.Rapidly declining fertility rates accelerates the graying of society, as shown in Figure 1(b) for Germany, Italy, and Japan.Whether increasingly unbalanced demographic structure can support a social security system is a concern as government seek policies to recover total fertility rates (Grant et al., 2004).
[ Figure 1] This study takes spatial rather than temporal views of fertility.Figure 2 illustrates maps of total fertility rates in Panel (a) and population density in Panel (b) for Japanese municipalities.
These maps show the negative relationship between total fertility rates and population density, as shown in Panel (c).This relationship might be explained by a model of Sato (2007), in which an agglomerated region attracts workers, intensifying population density and wage rate, while reducing fertility rate through agglomeration diseconomy.Goto and Minamimura (2015) note that rising real wage in denser areas increases the opportunity costs of childrearing.The rise of real wage in denser areas increases the opportunity costs of rearing children, while attracting workers, again elevating population density.This circularity engenders lower fertility rates in denser areas. 1   [Figure 2] Although theoretical studies predict that agglomeration discourages fertility, it remains unclear how it affects married couples' fertility decisions, such as the timing of childbearing.Figure 3 presents geographical distributions of fertility rates by age group and shows regional heterogeneity among different age groups.Note that fertility rates among ages 35-39 are relatively high in denser areas, especially Greater Tokyo and Osaka, although fertility rates among ages 25-29 are lower.Figure 3 therefore implies that households residing in denser areas postpone parenthood.This point is the basis of this this study for investigating whether and how agglomeration affects married couples' fertility decisions.
[ Figure 3] This study exploits Japanese social survey data to control for customarily unobservable household characteristics, which also enables us to control for spatial sorting.For example, city dwellers might have different opinions about fertility compared to rural residents, or the desire for security in old age motives having children, particularly in rural area (Nugent and Gillaspy, 1983;Nugent, 1985;Rendall and Bahchieva, 1998). 2 In addition, denser areas offer numerous job opportunities, and might attract people more intent on careers than parenthood.A social survey dataset provides empirical insights into spatial sorting of household characteristics and minimizes sorting bias.
Migration also influences households' fertility decisions through higher financial and nonfinancial costs (e.g., difficulties in building and maintaining social relationships).Thus, migrants are expected to have fewer children than non-migrants.Households' endogenous location choices will feature prominently regional variations in fertility rates.Accordingly, this study does a straightforward robustness check to evaluate how agglomeration affects fertility, focusing on non-migrant married couples.
This study offers new evidence about married couples' fertility, particularly the fertility process during their lifetimes.It finds that agglomeration discourages completed fertility; however, urban 1 Becker (1960) primarily developed the economic analysis of fertility.As explained in Becker and Lewis (1973), the interaction between quantity and quality of children is important in economic models of fertility.Willis (1973) extended the fertility model to incorporate opportunity costs of rearing children to additional working.See Becker (1992), Browning (1992), and Hotz et al. (1997) for details of fertility analysis. 2See also Fenge and Meier (2005) This study relates strongly to recent literature on fertility and housing prices.Simon and Tamura (2009) investigate the effects of housing rents on age at first marriage, age at birth of the first child, and number of children.They find that higher rents delay marriage and childbirth and reduce the number of children in households.Given that rents are higher in denser areas, this study complements their findings.In addition, Lovenheim and Mumford (2013) and Dettling and Kearney (2014) show that rising housing prices has price and wealth effects.More specifically, a rise in housing prices heterogeneously affects fertility decisions between homeowners and renters.
They find that an increase in house prices positively affects fertility for homeowners through the wealth effects.This study uses population density, instead of using housing prices, to capture the aggregate effects of agglomeration costs because, as shown later, population density captures numerous childrearing costs, such as educational costs.
The remainder of this paper is organized as follows.Section 2 discusses relationships between low Japanese fertility rates and agglomeration costs.Section 3 explains theoretical backgrounds of household fertility.Section 4 describes the empirical framework.Section 5 presents the social survey data.Section 6 discusses the estimation results.Section 7 presents the conclusions.

Low Japanese Fertility Rates and Agglomeration Costs
Japan's National Institute of Population and Social Security Research (2012) regularly conducts fertility surveys.As shown in Figure 4(a), major reasons why households do not have ideal numbers of children include the high childrearing and education costs.Figure 4(b) shows that both reasons provoke decreases in the planned number of children, compared to the ideal number of children.Thus, high childrearing costs drive lower fertility rates.
[ Figure 4] To explore those costs, we rely on surveys by Japan's Ministry of Education, Culture, Sports, Science and Technology (2009).Figure 5 shows annual costs of extramural activities for public school students by city size.Clearly, households who live in bigger cities pay more for extramural activities.In this background, there are two possibilities: extramural activities cost more in bigger cities and/or students there consume more.Panels (a) and (b) of Figure 6 show regional difference indices of school and tutorial fees, respectively.A stylized fact is that price indices in denser areas exceed those in less dense areas.Thus, even if consumption of educational services is identical across areas, educational costs differ through regional price differences.
[Figures 5-6] Summing up, a large number of Japanese households acknowledge that general costs of childrearing and education are higher in denser areas.This paper considers the aggregate effects of agglomeration costs.

Theoretical Background
To explain why households residing in denser areas have fewer children than those in rural areas, this study relies on a standard model by Becker and Lewis (1973) and Willis (1973).The utility function of the representative household residing in region r is given by where x r denotes the consumption of goods, y r denotes the number of children, and q r denotes their quality per child, assumed to be equal among children in each household.It is also assumed that each household is endowed with one unit of time allocated between working and rearing children.Households must spend time b r y r rearing children y r , where b r is a positive constant.
The household budget constraint becomes p xr x r + (p yr + p qr q r )y r = w r (1 − b r y r ), where p xr denotes the price of consumption goods in region r, p yr denotes the cost of rearing children, p qr denotes the price related with quality of children (e.g., education, training, health), and w r denotes the wage rate.In the budget constraint, p yr + p qr q r denotes the direct marginal cost of rearing children and b r w r denotes the opportunity cost of rearing children to working (i.e., loss of earnings) Utility maximization yields demand functions for consumption goods, children, and quality of children, respectively, as follows: , and q r = ξ μ The following relationships can be obtained: ∂y r ∂p yr < 0, ∂q r ∂p yr > 0, ∂y r ∂b r < 0, and ∂q r ∂b r > 0, which suggest that increases in childrearing costs reduce numbers of children per household and simultaneously increases the quality of children.In the empirical analysis, this study uses regional variations in p yr and b r , which are assumed to increase with economic concentration.
This analysis draws the following inference about the income effect on the number of children as follows: suggesting that a rise in wages increases the number of children and their quality simultaneously.
In the empirical part, this study uses cross-sectional variation between households to analyze the income effect on demand for children.
Although this study does not explore general equilibrium, its theoretical predictions are that the number of children per household is lower when full costs of childrearing are higher and that the number of children per household is higher when household income is larger. 3This study examines to what extent this prediction mirrors reality.

Measuring Agglomeration Effects on Fertility
This study estimates the demand function for children y r in (1) for married couples in which the wife is of childbearing age (wife is aged 15 to 49, by the definition of total fertility rate).
It is assumed that regional variations in childrearing costs p yr and b r are largely derived from population density, as shown earlier.Thus, the concern of our empirical analysis is to examine agglomeration impacts on fertility.
A standard approach is to linearly regress the number of children on population density and 3 Households endogenous location choices that respond disparities in real income may affect regional differences in fertility rates at spatial equilibrium; however its analysis exceeds the scope of this study.other control variables.However, it becomes an issue whether the dependent variable takes a discrete number.If so, the Poisson regression is more appropriate than linear regression.The

Poisson regression model to be estimated is given by
where y ir is the number of children of household i residing in region r; Dens r(i) is the population density of region r where married couple i lives during the study period; α is a parameter of interest that captures the aggregate effect of agglomeration costs on fertility and is expected to be negative; M i is a dummy variable that takes the value 1 if either spouse in household i has emigrated and 0 otherwise; X i is a vector of variables denoting household characteristics (age, gender, cohort dummies, employment status, condition of health, education, years of working experience, and household income); Xi is a vector of variables on household's social characteristics that affect fertility decision; D Reg r is a vector of region dummy variables; D Year t is a vector of year dummy variables; θ is a vector of parameters (α, γ, β , δ , η , ψ ).Thus, the parameter vector that maximizes log-likelihood function (y, θ) is estimated as follows: where n is the number of observations.
Note that the regression includes customarily unobservable household characteristics Xi .Using a social survey dataset mitigates estimation bias arising from spatial sorting driven by households' qualitative factor.
Our question is whether agglomeration affects completed fertility.Focusing on married couples among whom the wife is aged 50 or older, the outer age for childbearing as defined by total fertility rate, this paper estimates the Poisson regression model modified as follows: where Dens50 r(i)t denotes the population density of cities where the married couples lived when the wife was aged 50, and Z i is limited to the vector of variables on husband's and wife's education levels of X i in Model (2) because no historical information itemizes the income, work experience, and health status for married couples aged 50 and over. 4nterpretations of parameter α in Models ( 2) and (3) may be ambiguous when the sample includes migrants, even if migration status is controlled for.The Poisson models are estimated for two samples that include migrants with a migration dummy and that exclude migrants to distinguish effects of migration and agglomeration costs.

Testing Catch-Up Process on Fertility
where Age wife i denotes the wife's age for married couple i, and φ measures the catch-up process on fertility decision: positive value of φ suggests that married couples residing in denser areas delay their fertility decision.This regression is estimated for non-migrants aged 50 or below.
Another aspect of the catch-up process on fertility is whether agglomeration affects the timing of marriage and birth of the first child, respectively (e.g., Simon and Tamura, 2009).The sample is not divided by wife's age in this regression.Effects of agglomeration are estimated by this OLS regression: where Age wife i,k denotes the wife's age for married couple i at the time of marriage (k = 1) and birth of the first child (k = 2), respectively; Dens All r(i)t denotes the population density which takes the value of Dens r(i)t if married couple i is aged 50 or younger and the value of Dens 50 r(i)t if married couple i is aged 50 and above; and u i,k is the error term.Parameter α k captures agglomeration effects on the timing of marriage and birth of the first child, respectively.

Data
We exploit social survey data to control for customarily unobservable household characteristics affecting fertility decision.Our cumulative dataset (i.e., pooled cross section) is constructed from the Japan General Social Surveys (JGSS), which covers 2000,2001,2002,2005,2006,2008, and 2010. 5The sample is limited to married couples.Definitions of some variables are in Appendix A.  Figure 9(c) shows households' opinions on the need for children in marriage.Some households believe that children are not necessary.Thus, a dummy for non-necessity of children in a marriage takes the value 1 for households that agree or somewhat agree with this opinion and 0 otherwise 5 We discarded the JGSS 2003 dataset because it omits questions about number of siblings.
in Panel (c) of Figure 9.
[Table 1 and Figures 7-9] We include the number of siblings because people who have relatively many siblings might have more children in a marriage.JGSS polls the number of siblings for both spouses, but often only one of them answers this question.Our question merges spousal responses.If both answer, the average number of siblings is used; if one answers the question, the number that he or she provided is used.2).Agglomeration measured by population density has a significantly negative effect on the number of children.In Column (1), the semi-elasticity of population density on the number of children is −0.134 (average marginal effect).Compared to Column (3), the magnitude of population density is overestimated by approximately 0.010 if the customarily unobservable household characteristics and migration are not controlled for.However, bias arising from unobservable household characteristics is considerably low.

Agglomeration Discourages Fertility
After excluding migrants from the sample and controlling for unobservable household characteristics, the average marginal effect of population density on the number of children is −0.127 in Column (5).This magnitude means that the 10% increase in population density on average decreases fertility by approximately 13 births per 1,000 married couples aged 15-49 (≈ −0.127 × 0.1 × 1, 000).Results show that agglomeration certainly discourages fertility.
Results indicate that both spouses' educations and wives' working hours negatively affect demand for children.The income effect positively correlates with the number of children: other things being equal, high-income married couples tend to have more children, as predicted theoretically.In Columns (2), (3), and (5), a dummy variable for non-necessity of children and the number of siblings demonstrate significantly negative and positive effects on the number of children at the 1% level, respectively.However, neither the old-age security motive score nor the migration dummy is statistically significant at the 10% level in Column (3).

Agglomeration Affects Completed Fertility
Table 3 presents estimation results for married couples among who do not have fertility decisions due to the end of maternal age (wife of married couple is aged 50 and above by the definition of total fertility rate).
From Columns (1)-( 3) in which the sample includes migrants, it is clear that agglomeration discourages completed fertility.However, the magnitudes of population density diminish compared to those in Table 2.More interesting, the effect of education is not significant even at the 10% level.In other words, higher education discourages fertility for young married couples, but does not necessarily affect completed fertility.In contrast, migration discourages completed fertility.
The sample is limited to married couples of non-migrants in Columns ( 4) and ( 5) to identify the agglomeration effect on fertility.The main results do not change: agglomeration discourages completed fertility.Comparing samples including and excluding migrants suggests that migration significantly influences completed fertility.
Thus far, empirical results have suggested that agglomeration discourages fertility through higher childrearing costs in denser areas.Furthermore, married couples residing in denser areas on average have fewer children as measured by completed fertility.However, the average gap in the number of children shrinks between denser and less dense areas as married couples get older.Therefore, we next uncover how agglomeration affects the timing of marriage and the birth of the first child.

Agglomeration Delays Birth of First Child
Table 5 presents estimation results of agglomeration impacts on the wife's age at marriage.The coefficient estimates of population density take positive values in Columns ( 1)-( 5), but they are not significant at the 10% level.In contrast, higher education level significantly delays the timing of marriage at the 1% level.It is not necessarily evident that agglomeration discourages marriage.
Table 6 provides evidence on whether agglomeration delays birth of first child, as shown in Figure 8(b).Unlike the estimation results for marriage, the coefficient estimates of population density are significantly positive at the 5% level.Similar to the timing of marriage, higher education level significantly delays the timing of the first child's birth at the 1% level, and migration does so significantly, as shown in Column (3).
In sum, agglomeration strongly discourages fertility among younger married couples; however married couples in denser areas tend to have their children later in life.In other words, agglomeration delays the birth of the first child, but does not necessarily delay the timing of marriage.Married couples in less dense areas tend to have children early in life and stop after approximately two children.Consequently, the catch-up process on fertility in denser areas can be observed.Note that slight gap in completed fertility remains between denser and less dense areas.

Conclusion
This paper has investigated whether agglomeration discourages fertility.Using a social survey dataset that inquired into households' fertility decisions enabled us to control for economic factors alongside customarily unobservable household characteristics.
We found that agglomeration discourages married couples' fertility decision, suggesting that the price effects in denser areas strongly influence demand for children.A more interesting finding is that agglomeration delays young married couples' fertility decision.Other things being equal, the average gap in the number of children is relatively large when couples are younger and shrinks as they age.Although the gap never disappears, married couples in denser areas tend to have children later in life.
The findings of this study have important implications for lifting fertility rates in developed economies.Agglomeration hampers fertility rates through higher childrearing costs, implying that economic growth policies centered upon agglomeration economies simultaneously discourage married couples' fertility decision and delay the timing of birth of the first child.Policymakers who advocate agglomeration in graying societies should consider how to minimize "by-products" of these economic growth policies.
Future research needs to address two limitations in this research.We simply focus on married couples, but decisions to marry affect the total fertility rates.Thus, it should be noted that low total fertility rates in denser areas also originate from their high proportions of unmarried people.
Self-selected migration also needs to be addressed.Denser areas are likely to attract single people who will work long term and displace married couples with children due to high cost-of-living.
Clarifying these mechanisms remains for future research.

Dummy for University or Higher Education
Takes the value 1 if a respondent graduated from university or graduate school and 0 otherwise.

Dummy for Not Healthy
Takes the value 1 if answers are 4 or 5 on a one-to-five scale (1=good, 5=bad).

Dummies for Survey Years
Take the value 1 if married couple i answers the questionnaire either in the 2000,2001,2002,2005,2006,2008, or 2010 survey and 0 otherwise.

Figure 3
Figure 3 intimates that households residing in denser areas simply might delay parenthood due to high associated costs.This section examines how married couples residing in denser areas bear children later in life.Thus, the cross term of population density and wife's age is introduced into the Poisson regression model (2) as follows: Figure 7. Panel (a) of Figure 7 displays differences in the number of children by dividing 75th percentile point for population density (Dens All r(i)t ).Households averaging aged 20-24 in denser areas (exceeding the 75th percentile of population density, 4,100 persons/km 2 ) have half as many children as households in less dense areas.The gap between the two groups narrows, but a slight gap remains.Panel (b) of Figure 7 presents average ideal numbers of children by cohort group.We see the younger generation tending toward lower ideal numbers, but no geographical gap relates to population density.People in the same generation desire the same number of children regardless of where they live.Panels (a) and (b) of Figure 8 present distributions by city size for the wife's age at marriage and at birth of the first child, respectively.Approximately, couples residing in denser areas tend to get married and have their children later in life than couples in less dense areas.This finding associates agglomeration with delayed childbearing among married couples.Panels (a) and (b) of Figure 9 depict differences in how households perceive the roles of government and family in providing old-age security.Panels (a) and (b) indicate a trend for households to want government instead of families to assist their old-age.Old-age security score in Table 1 indicates the sum of values in Panel (a) and (b) of Figure 9. Minimum and maximum values are 2 and 10.Greater values indicate how households are responsible within the families in their old age.The motive to be secure in old age predicts that such households have more children.

Figure 10
Figure 10 shows the average gap in the number of children predicted from the Poisson regression model (4).The baseline is Tsumagoi-mura in Gunma Prefecture (approximately 114 persons/km 2 in 1980).Figure 10 depicts four cases in Kanto (Utsunomiya-shi, Odawara-shi, Figure 10 depicts four cases in Kanto (Utsunomiya-shi, Odawara-shi, Saitama-shi, and Musashino-shi).As mentioned, the catch-up process appears among couples after age 30 in denser areas, and the average gap in the number of children between the baseline area and each area shrinks.The average gap at age 25 between Tsumagoi-mura and Musashinoshi is approximately −0.25; it becomes approximately −0.11 at age 50.Comparing Musashino-shi to Utsunomiya-shi, the gaps are approximately −0.10 at age 25 and −0.05 at age 50.Although the average gap in the number of children between denser and less dense areas remains slightly, married couples tend to have children later in life.

2 .
numbers of observations for full sample, sample (wife's age < 50), and sample (wife's age ≥ 50) are 4,102, 2,282, and 1,820, respectively.The numbers of observations for wife's age at marriage are 1,683, 1,036, and 647, respectively.The numbers of observations for wife's age at birth of first child are 3,667, 1,964, and 1,703, respectively.The household who has the maximum number of children and the uppermost 1 percentile of the distribution of working hours for husband and wife are excluded from the full sample as extreme outliers.Population density is expressed in persons/km Working hours are expressed in 10-hour units.

Figure 1 :Figure 2 :
Figure 1: Total Fertility Rates and Population Aging Rate of Selective Developed Countries Note: Created by author.Japan's fertility data are obtained from the Vital Statistics of the Ministry of Health, Labour and Welfare.Other data are obtained from the World DataBank of the World Bank.

2 )
Very high costs of rearing children and education Small living space Dislike giving birth at a later stage in life Want children but cannot have Very high costs of rearing children and education Small living space Dislike giving birth at a later stage in life Want children but cannot have (b) Compared with Planned Number of Children

Figure 4 :
Figure 4: Reasons Why Households Do Not Have Ideal Number of Children (Multiple answers allowed, %) Note: Created by author based on 2010 Japanese National Fertility Survey Volume I, National Institute of Population and Social Security Research

Figure 5 :
Figure 5: Annual Costs of Extramural Activities for Public School Students by City Size Note: Created by author based on 2012 Survey on Household Expenditures on Education per Student, Ministry of Education, Culture, Sports, Science and Technology.

Figure 8 :
Figure 8: Wife's Age at Marriage and Birth of First Child

Table 1
presents descriptive statistics of variables.The average number of children in the sample is approximately 1.98.Average numbers of children by age group appear in

Table 2
presents estimation results of average marginal effects obtained from Poisson regression model (

Table 4
(4)sents estimation results of the Poisson regression model(4).The important point is that the cross term of population density and wife's age significantly and positively affects the number of children in Columns (1) and (2) of Table4.Although married coupled residing in denser areas tend to have children later in life, creating a big gap in the number of children between denser and less dense areas in the early life-stage, this gap gradually shrinks as they get older.

Table 1 :
Descriptive Statistics of Variables for Regression Analysis

Table 5 :
Wife's Age at Marriage and Agglomeration Heteroskedasticity-consistent standard errors clustered at cohort groups are in parentheses.Constant is not reported.* denotes statistical significance at the 10% level, ** at the 5% level, and *** at the 1% level.

Table 6 :
Wife's Age at Birth of First Child and Agglomeration Heteroskedasticity-consistent standard errors clustered at cohort groups are in parentheses.Constant is not reported.* denotes statistical significance at the 10% level, ** at the 5% level, and *** at the 1% level.