Abstract
Selected Issues
Japan’s Fertility Rate1
Japan’s population is projected to decline by more than 25 percent in the next 40 years, if the current low fertility rate persists. This overwhelming demographic challenge ahead of Japan leads to consideration as to whether there are policies the authorities can implement to affect the fertility rate. This chapter takes stock of recent developments with respect to Japan’s fertility rate, and argues that a combination of policies could raise the fertility rate if implemented in a coordinated and sustained manner.
A. Introduction
1. Japan’s population is projected to decline by more than 25 percent in the next 40 years, if the current low fertility rate persists. Japan’s total fertility rate (sum of the age-specific birth rates of women aged 15 to 49) stood at 1.4 children per woman in 2018, which is well below the population replacement level of 2.1.2 In addition, a large gap remains between the actual fertility rate and the “desired” rate of 1.8—the latter being the rate expected if people had their desired number of children. This gap raises the question as to whether public policies can help increase Japan’s fertility rate by removing potential obstacles. A higher fertility rate could have positive impacts on aggregate GDP over the long-run, by increasing the size of the labor force and helping stabilize public debt (as a share of GDP). At the same time, this policy objective needs to be mutually consistent and reinforcing with other public policy objectives. For example, the IMF has stressed the need to promote female labor force participation—see IMF (2018), Colacelli and Fernandez-Corugedo (2018). However, if not accompanied by sufficient support for working couples with small children, higher FLFP could potentially increase the opportunity cost of giving birth and thereby lower the fertility rate. The Japanese government has stepped up its support for households with children, with notable measures being increasing the availability of childcare facilities and making preschool and childcare free. This chapter takes stock of recent developments in Japan’s fertility rate, and discusses whether public policy can help increase the fertility rate.
Japan: Population Shrinkage by Prefecture, 2015–2045
(In percent)
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
Source: Statistics Bureau of Japan. Prefecture data.Japan: Population Shrinkage by Prefecture, 2015–2045
(In percent)
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
Source: Statistics Bureau of Japan. Prefecture data.Japan: Population Shrinkage by Prefecture, 2015–2045
(In percent)
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
Source: Statistics Bureau of Japan. Prefecture data.B. Overview of Japan’s Fertility Rate
2. Japan’s fertility rate has improved in recent years but remains low at 1.4 children per woman. After the second baby boom in the early 1970s, Japan’s fertility rate declined until it bottomed out at 1.26 in 2005 (see top-left chart of Figure 1). This declining trend of fertility mirrors the trend increase in the percentage of late-married and non-married (see the bottom two charts of Figure 1). The marriage rate (the number of marriages per population of 1,000 persons) in Japan halved from around 10 in the early 1970s to around 5 in 2015. In more recent years, Japan’s fertility rate has marginally recovered, but remains low at around 1.4 children per woman—well below the population replacement rate of 2.1, and lower than all other G7 countries except Italy (see the top-right chart of Figure 1).
Japan’s Fertility Rate–Stylized Facts
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
Japan’s Fertility Rate–Stylized Facts
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
Japan’s Fertility Rate–Stylized Facts
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
3. A gap exists between the actual and desired number of children per couple. According to a survey by the National Institute of Population and Social Security Research (the NIPSSR), an ideal number of children for a couple is 2.32 on average.3 However, those couples plan to have on average 2.01 children—the sum of the number of children already born (1.68) and the number of additional children the couple plans to have (0.33). On the other hand, while the unmarried rate of females aged 30–34 has risen to about 35 percent, 90 percent of unmarried females aged 18–34 intend to marry in the future. Based on the NIPSSR survey and other data, the government estimates the “desired” fertility rate of 1.84 that could be achieved through closing the gap between the actual and desired number of children, and removing obstacles to higher fertility rates.
4. The direct cost of childbearing seems to be the most important factor which prevents couples from attaining their ideal number of children. The NIPSSR 2015 survey summarizes the reasons why couples do not have their ideal number of children. The top reason cited by 56 percent of couples is the cost of childbearing; this is followed by the reluctance to bear a child at an advanced age (40 percent).
Japan: Reasons Why Couples Do Not Have Their Ideal Number of Children
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
Source; The 15th Japunest; National Fertility Survey {Survey on Married Couples), National Institute of Population and Social Security Research (2015).Note: Subjects were first-married couples with fewer planned children than their ideal number. The percentage of married couples with fewer planned children than their ideal number is 30.3 percent.Japan: Reasons Why Couples Do Not Have Their Ideal Number of Children
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
Source; The 15th Japunest; National Fertility Survey {Survey on Married Couples), National Institute of Population and Social Security Research (2015).Note: Subjects were first-married couples with fewer planned children than their ideal number. The percentage of married couples with fewer planned children than their ideal number is 30.3 percent.Japan: Reasons Why Couples Do Not Have Their Ideal Number of Children
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
Source; The 15th Japunest; National Fertility Survey {Survey on Married Couples), National Institute of Population and Social Security Research (2015).Note: Subjects were first-married couples with fewer planned children than their ideal number. The percentage of married couples with fewer planned children than their ideal number is 30.3 percent.5. The opportunity cost of childbearing also matters. During the period of Abenomics, Japan’s female labor force participation rates (FLFP) have substantially increased from 63 percent in 2012 to 71 percent in 2018 (for those aged 15 to 64), according to OECD Employment and Labour Market Statistics. A higher FLFP could increase household incomes and make the cost of childbearing affordable (i.e., the income effect). On the other hand, higher FLFP could increase the opportunity cost of having children as females typically need to leave their workplaces for a certain period of time (i.e., the substitution effect). In fact, according to the NIPSSR (2015) survey, about 15 percent of couples responded that their jobs/businesses are one of the main reasons for the gap between the actual and desired number of children, which can be seen as a measure of the opportunity cost of having children.
6. The opportunity cost of having children is high for full-time workers. According to a longitudinal survey of Japanese newborns in 2010, the total share of female workers sharply dropped after giving birth from 62 percent to 35 percent (see Ministry of Health, Labour and Welfare (2019)). The total share of female workers gradually recovered to the level before childbirth (61 percent) about 5.5 years after giving birth. However, as illustrated in the text chart on the right, most of the female workers re-entered the job market as part-time workers. The share of full-time workers dropped from 38 percent to 25 percent after childbirth and remained low at around 26 percent even 7.5 years after giving birth. The MHLW survey suggests that a number of full-time female workers give up full-time status after childbirth, which has enduring implications for their lifetime income and employment benefits.
Japan: Female’s Employment Status after Giving Birth
(In percent)
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
Source: The 8th Longitudinal Survey of Newborns In the 21st Century, Ministry of Health, Labour and WelfareJapan: Female’s Employment Status after Giving Birth
(In percent)
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
Source: The 8th Longitudinal Survey of Newborns In the 21st Century, Ministry of Health, Labour and WelfareJapan: Female’s Employment Status after Giving Birth
(In percent)
Citation: IMF Staff Country Reports 2020, 040; 10.5089/9781513529424.002.A003
Source: The 8th Longitudinal Survey of Newborns In the 21st Century, Ministry of Health, Labour and WelfareC. Empirical Analysis: Data and Methodology
7. This section provides a simple empirical analysis to study the possible drivers of Japan’s fertility rate. D’Addio and Mira d’Ercole (2005) discussed the historical development of fertility rates across OECD countries and showed that, based on a static cross-section analysis, fertility rates are lower when the direct costs of childbearing are higher. Regarding the impact of increased FLFP, they calculated that the correlation between female employment rates and total fertility rates in OECD countries turned from negative to positive in the late 1980s. Using a dynamic panel data model, they observed that the fertility rate is higher when (i) female employment rates are higher, and/or (ii) the ratio of female to male wages is lower. In this section, an approach similar to D’Addio and Mira d’Ercole (2005) is applied to Japan’s prefecture-level panel data to study Japan specific features.5
8. A panel dataset was constructed that covers data for 47 prefectures for the period 2001 to 2015. In the panel regression analysis, the dependent variable is the prefecture-level total fertility rate, and five variables that might affect the total fertility rate are included as explanatory variables:
Education costs. The share of education expenses in total expenditures for non-single households is derived from the Family Income and Expenditure Survey conducted by the Ministry of Internal Affairs and Communications. As this could be viewed as a proxy for the direct cost of child-rearing, the expected sign is negative for the relationship between the total fertility rate and education costs.
Female labor force participation rate. The prefectural labor force participation rates (ratios of labor force over the population) for women between 15 and 64 years old are derived from data of the National Census.6 Based on the findings by D’Addio and Mira d’Ercole (2005), the expected sign is positive for the relationship between the FLFP rate and the total fertility rate.
Wage gap. Ratios of female to male monthly wages (a higher ratio indicates a smaller wage gap between male and female) are used to measure the gap. The wage is average monthly contractual cash earnings in the MHLW’s Basic Survey on Wage Structure. D’Addio and Mira d’Ercole (2005) find that a smaller wage gap (i.e., a higher ratio of female to male wages) implies a larger opportunity cost due to foregone income during maternity leave, leading to lower fertility rates. However, in the case of Japan, female workers often give up their full-time status after childbirth, as discussed in the previous section. Hence, a large wage gap could reflect changes in women’s employment status after childbirth in Japan. Provided that the observed large wage gap (i.e., a low ratio of female to male wages) is attributable to a large share of female part-time workers after childbirth, the opportunity cost of giving birth would increase over the long-run, which could lower the fertility rate. Furthermore, a smaller wage gap is expected to yield better income prospects after childbirth, which increases the affordability of childbearing. In light of these factors, the expected sign of the wage gap-total fertility rate relationship could be positive, which would run contrary to the findings of D’Addio and Mira d’Ercole (2005).
Availability of childcare facilities. The capacity (sum of authorized quotas of children) of childcare facilities for each prefecture is obtained from the MHLW’s Survey of Social Welfare Institutions. Following Unayama (2009), the capacity is divided by the female population between 20 and 44 years old to make it comparable across prefectures. This variable can be viewed as an indication of commitment to child-friendly policies by prefectural governments. The expected sign of the relationship between total fertility rate and childcare facilities is positive.
Unemployment rate. Prefecture-level unemployment rates are obtained from the Statistics Bureau’s model-based estimates, based on data from the Labor Force Survey. This variable is introduced to factor in macroeconomic conditions, and the expected sign of the unemployment rate-total fertility rate relationship is negative.
9. The Pooled-Mean Group estimator is the preferred model. The simple pooled OLS regression model does not allow for prefecture-specific effects. A commonly-used alternative, the fixed effects estimator, takes account of prefecture-specific effects, but fails to deal with the issue of the endogeneity of explanatory variables with respect to the total fertility rate. In order to deal with the endogeneity issue, the PMG (Pooled Mean Group) estimator proposed by Pesaran et al. (1999) is used here, following the approach of D’Addio and Mira d’Ercole (2005). The PMG estimator distinguishes long-run and short-run dynamics. Coefficients for long-run effects are assumed to be identical across prefectures, while those for short-run effects are allowed to differ. The other model used in D’Addio and Mira d’Ercole (2005), a GMM (Generalized Method of Moments)-System estimator developed by Arellano and Bover (1995) and Blundell and Bond (1998), is a less preferred option here, because the post-estimation Sargan test rejected the null hypothesis that over-identifying restrictions are valid, indicating a potential misspecification (see Table 1). Therefore, the discussion hereafter is based on the PMG estimates, with a focus on the long-run dynamics, while turning to the GMM estimates as a complementary reference.
Japan: Estimation Results of Total Fertility Rate by Prefecture, 2001–15
Japan: Estimation Results of Total Fertility Rate by Prefecture, 2001–15
| VARIABLES | (1) PMG (long-run coefficients) | (2) PMG (short-run coefficients) | (3) GMM-SYS |
|---|---|---|---|
| Wage gap | 1.039*** (0.198) | 0.402*** (0.0815) | |
| FLFP | 4.003*** (0.583) | 0.787*** (0.214) | |
| Education costs | -1.170** (0.507) | -0.0518 (0.226) | |
| Childcare facilities | 3.153*** (0.811) | 1.151*** (0.211) | |
| Unemployment rate | 4.442*** (0.655) | -0.116 (0.266) | |
| Lag of total fertility rate | 0.438*** (0.0340) | ||
| Error correction term | -0.385*** (0.0353) | ||
| Change in wage gap | -0.355*** (0.0835) | ||
| Change in FLFP | -10.29*** (2.774) | ||
| Change in education costs | 0.135 (0.172) | ||
| Change in childcare facilities | 1.015* (0.564) | ||
| Change in unemployment rate | -0.888* (0.465) | ||
| Constant | -0.873*** (0.0814) | -0.116 (0.0999) | |
| Observations | 658 | 658 | 658 |
| Sargan test (p-value) | 0.000 |
Japan: Estimation Results of Total Fertility Rate by Prefecture, 2001–15
| VARIABLES | (1) PMG (long-run coefficients) | (2) PMG (short-run coefficients) | (3) GMM-SYS |
|---|---|---|---|
| Wage gap | 1.039*** (0.198) | 0.402*** (0.0815) | |
| FLFP | 4.003*** (0.583) | 0.787*** (0.214) | |
| Education costs | -1.170** (0.507) | -0.0518 (0.226) | |
| Childcare facilities | 3.153*** (0.811) | 1.151*** (0.211) | |
| Unemployment rate | 4.442*** (0.655) | -0.116 (0.266) | |
| Lag of total fertility rate | 0.438*** (0.0340) | ||
| Error correction term | -0.385*** (0.0353) | ||
| Change in wage gap | -0.355*** (0.0835) | ||
| Change in FLFP | -10.29*** (2.774) | ||
| Change in education costs | 0.135 (0.172) | ||
| Change in childcare facilities | 1.015* (0.564) | ||
| Change in unemployment rate | -0.888* (0.465) | ||
| Constant | -0.873*** (0.0814) | -0.116 (0.0999) | |
| Observations | 658 | 658 | 658 |
| Sargan test (p-value) | 0.000 |
D. Results
10. Every explanatory variable has a statistically-significant impact on the prefectural fertility rate in the long-run. Table 1 shows the results:
Wage gap. A rise in the female wage relative to the male wage (a smaller gap between male and female wages) has a positive impact on fertility rates in the long-run. This supports the argument that a smaller wage gap could indicate lower opportunity costs of childbirth over the long-run, and make childbearing more affordable. On the other hand, the sign is negative in the short-run. A possible interpretation of this negative short-run effect is that a smaller wage gap could lower the fertility rate in the short-run due to a larger foregone income during maternity leave.
Female labor force participation rate. A higher female labor force participation rate has a positive impact on the fertility rate in the long-run, in line with the findings of D’Addio and Mira d’Ercole (2005). The long-run result indicates that a one-percentage point increase in the FLFP rate is associated with a 0.04 increase in the fertility rate (see Table 1). Meanwhile, the sign is negative in the short-run, possibly pointing to a negative impact of opportunity costs on the fertility rate, as discussed above.
Education costs. A reduction in education costs has a positive impact on the fertility rate in the long run, though this was not confirmed in the GMM estimates. The short-run coefficient is not statistically significant.
Childcare facilities. An increase in childcare facilities has a positive impact on the fertility rate both in the short run and long run, demonstrating the potential effectiveness of child-friendly policies in raising the total fertility rate.
Unemployment rate. The sign of the long-run coefficient is positive, while it is negative in the short run. The coefficient is statistically insignificant in the GMM estimates. This positive correlation could imply a low opportunity cost of childbearing when unemployment rates are high, though further analysis is warranted on this point.
E. Policy Implications and Conclusions
11. The Japanese government’s Work Style Reform, which has intensified since 2016, could have a positive impact on the fertility rate.7 Introduction of public policies to reduce the unwilling exclusion of female workers from the labor market following childbirth could raise the fertility rate over the long-term. It is of particular importance to nurture a working and social environment where female regular workers can retain regular-worker status after childbirth, if they wish to. Potential measures the authorities could undertake to achieve this goal include: (i) further increasing childcare availability; (ii) rewarding firms with high retention rates of female employees after childbirth; and (iii) eliminating disincentives to regular and full-time work embedded in the tax and social security systems (see IMF (2019)). The impact on the fertility rate could be reinforced by measures to alleviate the direct cost of childbearing, such as lowering education and childcare costs.
12. Public policies supporting fertility should be implemented in a coordinated and sustained manner. Since the impact of each policy on the fertility rate is relatively small, a wide array of policies needs to be put in place in a coordinated, mutually-reinforcing manner in order to make a meaningful impact on the fertility rate. Lastly, the negative short-run effects of female labor force participation and the wage gap on the total fertility rate point to a need for the authorities to be persistent—sustaining public policies to support fertility even if the fertility rate is negatively affected in the short-run.
References
Abe, K. and Y. Harada, 2008, “Effects of Policies to Support Child-Rearing on Fertility Rate: Analysis of Municipalities’ Data”, Kaikei-kensa Kenkyu, Vol. 38 (in Japanese).
Asano, T. and N. Yamagata, 2010, “Measures to Support the Fertility Rate and their Policy Impacts”, Graduate School of Public Policy, The University of Tokyo.
D’Addio, A.C. and M. Mira d’Ercole, 2005, “Trends and Determinants of Fertility Rates in OECD Countries: The Role of Policies”, OECD Social, Employment and Migration Working Papers No. 27, OECD: Paris.
Arellano, M. and O. Bover, 1995, “Another Look at the Instrumental-Variable Estimation of Error Component Models”, Journal of Econometrics, Vol. 68, pp. 29–52.
Blundell, R. and S. Bond, 1998, “Initial Conditions and Moment Restrictions in Dynamic Panel Data Models”, Journal of Econometrics, Vol. 87, pp. 115–43.
Cabinet Office of Japan, 2019, Annual Report on the Declining Birthrate 2019, Tokyo.
Colacelli, M. and E. Fernandez-Corugedo, 2018, “Macroeconomic Effects of Japan’s Demographics: Can Structural Reforms Reverse Them?”, IMF Working Paper 18/248.
International Monetary Fund, 2018, “Japan: Article IV Consultation—Staff Report,” IMF Country Report 18/333.
International Monetary Fund, 2019, “Japan: Article IV Consultation—Staff Report,” IMF Country Report.
Kato, H., 2018, “The Analysis on Disparities of Fertility Rate of Japanese Municipalities”, Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.14, No. 1.
Ministry of Health, Labour and Welfare, 2019, “The Eighth Longitudinal Survey of Newborns in the 21st Century (2010 Cohort)” (in Japanese), Tokyo.
National Institute of Population and Social Security Research, 2015, “The Fifteenth Japanese National Fertility Survey,” Tokyo.
OECD, 2019, Social and Welfare Statistics, OECD: Paris.
Pesaran, M. H., Y. Shin and R.P. Smith, 1999, “Pooled Mean Group Estimation of Dynamic Heterogeneous Panels”, Journal of the American Statistical Association, Vol. 94, pp. 621–34.
Pesaran, M.H. and R.P. Smith, 1995, “Estimating Long-run Relationships from Dynamic Heterogeneous Panels”, Journal of Econometrics, Vol. 68, pp. 79–113.
Toda, A., 2007, “Empirical Analysis of Fertility Rates – Focus on its Relationship with the State of Economic Activity and Family Policy”, RIETI Discussion Paper Series 07-J-007 (in Japanese), Tokyo.
Unayama, S., 2009, “Policy Responses to the Low Fertility and Aging and Female Labor Participation – Analysis of Prefectures’ Data”, RIETI Discussion Paper Series 10-J004 (in Japanese), Tokyo.
Prepared by Takuma Hisanaga (APD).
The total fertility rate in a specific year is defined as the total number of children that would be born to each woman if she were to live to the end of her child-bearing years and give birth to children in alignment with the prevailing age-specific fertility rates. This indicator is measured in children per woman. See OECD (2019).
See NIPSSR (2015).
This rate is based on the planned number of children for a couple, as well as other statistics including the share of single people who hope to marry and the desired number of children of single people.
Among the previous studies of the fertility rate in Japan, Abe and Harada (2008) observed that, based on a cross-section analysis of municipal-level data, a rise in female wages had a negative impact on the fertility rate, and that the impact of improved availability of childcare facilities on the fertility rate was positive. Kato (2018) also analyzed municipal-level data to find a positive impact of the female labor participation rate on the total fertility rate.
As the National Census is a quinquennial survey, values for gap years are filled by linear interpolation.
See also Box 2 of International Monetary Fund (2019) Japan: Article IV Consultation—Staff Report, for a case study of the experience of Nagi-town in Okayama Prefecture.
