Do Suburbs Have Higher Fertility than Central Cities? Diversity of Regional Differences in Population Reproduction Within Metropolitan Areas in Japan

Abstract

Observing the distribution of fertility and clarifying its mechanisms are important for discussing sustainability of population reproduction. This study investigated regional fertility differences within Japanese metropolitan areas to test the validity of the hypothesis that fertility is lower in central cities and higher in suburban areas. Additionally, the relationship between urban life cycles and fertility trends was explored. Data from the “Specified Report of Vital Statistics” between 2000 and 2015 was used. The findings reveal that, though large metropolitan areas such as Tokyo, Osaka, and Nagoya exhibit significantly lower fertility in specific central wards, this pattern is not applicable to all Japanese metropolitan areas. Numerous small to medium-sized metropolitan areas exhibit diverse fertility distribution patterns, with some central cities recording higher fertility rates than their suburbs. The results suggest that regional fertility differences in Japan are shaped by a complex interplay of urban development, housing policies, and demographic behaviors rather than a simple dichotomy between central cities and suburbs. This study highlights the need for further interdisciplinary research integrating population geography and urban studies to refine the understanding of fertility variations within Japanese metropolitan areas.

1. Introduction

According to “Population Census of Japan” and “Specified Report of Vital Statistics,” Japan’s total fertility rate has remained below the population replacement level since 1974, and the working-age population has been declining continuously since 1995. The OECD estimates that, if Japan’s total fertility rate remains around 1.3 and net immigration stays constant, its population will decline from 125 million to approximately 96 million by 2060 and fall below 70 million by 2100 [1] (p. 7). If this projection becomes reality, Japan’s working-age population will decline by more than half by the end of this century. Jones [1] points out that one disadvantage of population decline and falling fertility is the difficulty in maintaining social security systems that provide medical care, nursing care services, and pensions. Indeed, Japanese experts have also pointed out similar problems [2,3]. Furthermore, population decline is expected to cause problems such as a decrease in domestic demand [2,4], an increase in land with unknown owners and vacant houses [5], and a widening of regional disparities [6]. Jones [1] described Japan, facing such circumstances, as “Japan is at the frontier of the global demographic transition” (p. 6). In other words, elucidating the mechanisms underlying Japan’s demographic phenomena is of crucial importance both socially and academically.

To understand Japan’s demographic trends, some aspects of Western population theories may apply well, while others do not. Specifically, the “first demographic transition” [7]—where societies shift from high birth and high death rates, through high birth and low death rates, to low birth and low death rates as they modernize—effectively explains Japan’s demographic dynamics from the 1870s to the 1960s [8,9]. The persistence of low fertility due to changing values and structural factors is termed the “second demographic transition” [10], a concept that also aptly describes Japan, where low fertility has persisted since the 1960s [9]. However, the “third demographic transition” does not fit Japan well. In European countries, the number of immigrants has remained high since the 1950s, and the demographic phase has shifted from the second to the third demographic transition [11]. In Japan, the government remains reluctant to accept immigrants even as fertility declines and the population ages, and the foreign population has barely increased (Figure 1). Based on the above, Japan could be positioned as a special case where the “second demographic transition” persisted for an extended period without experiencing a “third demographic transition.” Alternatively, it may be undergoing a unique transition distinct from the “third demographic transition” experienced by Western nations.

Since existing population theories do not perfectly apply to Japan, how should we accurately assess the current state of Japan’s population under these unique circumstances? In order to solve this problem, an approach that observes regional differences in fertility rates within the country is highly effective. Boyle argues that “Spatial variations (or the lack of them) can be a useful test of the comprehensiveness of grand theories of population change and it would be a shame if population geographers did not take up the challenge to investigate these theories a little more enthusiastically” [12] (p. 622). Changes in demographic trends, such as population transition, do not occur without spatial variation within a country; for example, declines in fertility often originate in urban areas. Specifically, it has been reported that the decline in fertility associated with the first demographic transition [13,14,15], shrinking household composition related to the second demographic transition [16,17,18], and the decline in fertility in developing countries [19] are occurring first in urban areas. Therefore, if a trend entirely different from previous patterns is confirmed regarding regional variations in Japan’s fertility, it could be interpreted as Japan potentially entering a new phase in terms of its population.

This study aimed to observe regional differences in fertility within Japanese metropolitan areas and test the validity of the hypothesis that fertility is low in central cities and high in suburban areas. In addition, it examined the relationship between the life cycle of Japanese cities and regional differences in fertility. As discussed in detail in the next chapter, the phenomenon of low fertility in suburban areas has been reported as an example where the traditional grounded theories and population geographical perspectives on fertility do not fit well (Section 2.1). Furthermore, despite the possibility that Japanese metropolitan areas may also be experiencing a reversal in fertility distribution between central cities and suburbs, research reports on regional fertility differences within metropolitan areas remain insufficient (Section 2.2). Therefore, achieving this research objective could potentially capture a new phase in population in Japan.

The years analyzed in this study were 2000, 2010, and 2015. Japan’s metropolitan areas experienced a phase of suburbanization from the 1960s to the 1990s. However, in the late 1990s, the Tokyo and Osaka metropolitan areas began to show signs of population recovery in their central cities (urban centers) [20,21]. Subsequently, in the first half of the 2000s, population recovery in central cities (urban centers) progressed remarkably, and in the second half of the 2000s, the trend of population recovery in central cities (urban centers) became more pronounced in local metropolitan areas with prefectural capitals as the central cities [21,22,23,24,25,26,27,28,29,30,31,32,33]. Japanese urban geographers have reported the phenomenon of population recovery in central cities (urban centers), such as Tokyo [21,22,23,24], Osaka [25], Sapporo [26], Sendai [27,28], Kyoto [29], and Fukuoka [30]. Similar phenomena have been observed in local cities, such as Mito [31], Niigata [32], and Toyohashi [33]. Therefore, by setting the period 2000–2015 as the year for analysis, this study tested whether the hypothesis of Rodrigo-Comino et al. [34] that population recovery in central cities (urban centers) may affect regional differences in fertility within metropolitan areas applies to Japan. However, the year 2005 was excluded from the analysis because “the Heisei municipal mergers” made it difficult to prepare fertility and metropolitan area data on a municipal scale, and the year 2020 was expected to be affected by COVID-19.

2. Previous Research

2.1. Population Studies in Western Countries

Given this context, it is important to review what regional variations in fertility have been identified in previous studies. In order to define research objective, this Section first reviews research on regional fertility variations in Western countries, followed by an introduction to reports from Japan. The patterns of fertility spatial distribution pointed out by research in Western countries can be summarized as follows. First, one of the most widely known patterns of fertility distribution within a country is that fertility is low in urban (metropolitan) areas and high in rural (non-metropolitan) areas. This pattern has been widely confirmed in European countries [35], such as England and Wales [36], the Netherlands [37,38], Italy [39], West Germany [40], Austria and Poland [41], Estonia [42], Nordic countries [43,44], and Australia [45]. Similar trends were observed in the United States [46,47].

Second, within metropolitan areas low fertility in central cities (urban centers) and high fertility in the suburbs is observed. As described by [48,49] a similar pattern has been reported in the United States since the 1950s [50,51,52]. Freedman et al. [53] reported that differences in family size between the central city and the suburbs were subtle but pointed to the concentration of Catholics in the central city, who had more children than Protestants. In addition, they clarified that Protestants and Catholics living in the suburbs had significantly larger family sizes than Protestants and Catholics living in the central city. Freedman et al. [53] employed a method that controlled for individual attributes to properly assess regional differences in fertility. Their research is noteworthy because it is the origin for a common method that decomposes “the compositional effect,” which emphasizes the attributes of residents that comprise the local population, and “the contextual effect,” which emphasizes the environment that influences the fertility behavior of individual residents. Freedman et al. [53] also examined whether couples residing in the suburbs wanted more children (selective mobility effect) or whether residing in the suburbs led them to want more children (suburban contextual effect). In the late 1980s, several studies in the fields of family sociology and rural sociology focused on differences in family formation and residential patterns between central cities and suburbs [46,54,55,56]. In the 2000s, patterns of fertility distribution within metropolitan areas began to be examined again in European countries, and high fertility trends in suburban areas were reported in Finland [48], Sweden [49], Scotland [57]. In Australia, higher fertility was found in the outer suburbs of capital cities than in the inner and middle suburbs of capital cities, although this was limited to first births [45].

However, in recent years, research that re-examines the hypothesis that “fertility in central cities is lower than fertility in suburbs” has emerged. Rodrigo-Comino et al. examined the regional differences between central cities and suburbs in European countries by calculating the mean crude birth rate from 2013 to 2018 for 671 Functional Urban Areas [34]. They found that approximately 70% of Functional Urban Areas in Eastern Europe and 55% of Mediterranean countries had higher suburban birth rates than central-city birth rates. A contrasting pattern was found in Western (approximately 20%), Northern (approximately 25%), and Central European countries (approximately 30%), confirming the diversity in the distribution patterns of fertility within metropolitan areas. Rodrigo-Comino et al. also noted that regional differences in fertility may reflect the phase of the urban life cycle [34]. Specifically, they argued that metropolitan areas with relatively high fertility in the suburbs, which are typical of Eastern European countries, fell into a phase of suburbanization, whereas metropolitan areas with relatively high fertility in central cities, which are typical of Western and Northern European countries, fell into a phase of re-urbanization.

2.2. Population Studies in Japan

In the following paragraphs, we introduce perspectives on regional differences in fertility within Japan reported in previous studies, clarifying the achievements and limitations of Japanese demography and population geography. Historical demography studies have revealed that regional differences in fertility occurred in Japan before the demographic transition. Akira Hayami, who introduced the family reconstitution method developed at the Institut National d’etudes Démographiques in France to Japan, analyzed the population reproduction behavior of married couples in various regions of Japan, including the Suwa region of Nagano Prefecture [58] and Saijo Village, Anpachi-gun, Gifu Prefecture [59]. Subsequently, Hayami organized The EurAsia Project on Population and Family History from 1995 to 2000, in which historical demographers and family historians worked to clarify the actual fertility patterns and family formation in premodern Japan [60]. These results are summarized in [60], in which the regional fertility differences in pre-modern Japan are as follows: northeastern Japan, where the average age at first marriage, age at first childbirth, and the number of births is low; central Japan, where the average age at first marriage, age at first childbirth, and the number of births is high; and Southwest Japan, which has a high average age at first marriage, medium age at first childbirth, and a high birth rate.

In Japan, since the enactment of the “Family Registration Law” in 1898 and the development of the modern demographic statistics system in 1899, the following demographic transitions are reported: a gradual decline in fertility from the 1910s to the end of the Pacific War, baby boom from 1947 to 1949 after the war, sharp decline in fertility after the 1950s, and low fertility level since 1960. Kawabe [61] and Hama [62] reported the regional differences in Japan after 1920s, stating that prefectures in Japan’s three largest metropolitan areas (Tokyo, Ibaraki, Saitama, Chiba, Kanagawa, Aichi, Kyoto, Osaka, Hyogo, Nara, and Wakayama prefectures) had relatively low fertility rates, except during the 1960s and the 1970s. Consequently, regional differences in fertility in Japan today show a pattern of low fertility in metropolitan areas and high fertility in non-metropolitan areas, as reported in Western countries [35,36,37,38,39,40,41,42,43,44,45,46,47].

Though the historical changes and spatial distribution of fertility in Japan are similar to those in Western countries, they exhibit some contrary characteristics. First, the illegitimacy rate in Japan has remained low even after the emergence of a low fertility maintenance trend, which could be interpreted as a second demographic transition (Figure 2). Therefore, population studies analyzing regional differences in fertility in Japan have often used the method that decomposes fertility into nuptiality and marital fertility (e.g., Ref. [63]). Second, Government of Japan remains reluctant to accept immigrants even as fertility declines and the population ages, and the foreign population has barely increased (Figure 1).

In the 2000s, fertility studies in Japan using micro and panel data became popular under the influence of research trends in Western countries, as demonstrated by researchers such as Boyle [48,49,57] and Kulu [35,42,48,49]. These studies utilized data from “The National Fertility Survey” and “The National Survey on Migration” conducted by the National Institute of Population and Social Security Research of the Ministry of Health, Labour and Welfare to reveal the mechanisms underlying regional differences in fertility in Japan [64,65,66,67,68,69,70,71,72]. For example, Sasai [64], who used the 7th (1977), 8th (1982), 9th (1987), 10th (1992), 11th (1997), 12th (2002), and 13th (2005) “The National Fertility Survey” data to analyze the factors for regional differences in marital fertility, found that, while compositional effects such as the wife’s age at marriage and parents’ cohabitation determined fertility to some extent, region-specific contextual effects might have occurred in Southern Kanto (Saitama, Chiba, Tokyo, and Kanagawa Prefectures), Kyushu, and Okinawa (Fukuoka, Saga, Nagasaki, Kumamoto, Oita, Miyazaki, Kagoshima, and Okinawa Prefectures). Sasai [64] also determined variables that had different relationships with the fertility behavior of married couples in different regions. Specifically, while in the multiple regression model for Japan as a whole and for South Kanto (Saitama, Chiba, Tokyo, and Kanagawa prefectures), the number of children among couples where the wife was employed was suppressed, in the multiple regression model for Tohoku (Aomori, Iwate, Miyagi, Akita, Yamagata, and Fukushima prefectures), Hokuriku, and Koshinetsu (Niigata, Toyama, Ishikawa, Fukui, Yamanashi, and Nagano), there was no significant relationship between the wife’s employment status and fertility. The possibility that the relationship between variables may differ by region is also pointed out by Kamata and Iwasawa [66], who used the geography-weighted regression model, total fertility rate by municipality in 2005 as the dependent variable and ratio of workers in the primary industry, male unemployment rate, migration rate, ratio of nuclear families, ratio of university graduates (females aged 15–49), employment rate (females aged 15–49), ratio of unmarried people (females aged 30–39), number of marriages divided by unmarried population aged 15–49, and number of nursery schools per 100,000 people aged 0–5 years as explanatory variables. They found significant differences in the coefficients of eight variables, other than the number of nursery schools, among the regions. Kamata and Iwasawa [66] interpreted that many determinants of fertility identified in previous studies may be more suitable for explaining fertility variations in metropolitan areas.

Nuptiality [67] and marital fertility [65,68,69] are lower in the Tokyo metropolitan area (Saitama, Chiba, Tokyo, and Kanagawa prefectures) than in other regions. Yamauchi [65] compiled micro data from the 4th (2008) and 5th (2013) “The National Survey on Family” and found that the average number of children in the Tokyo metropolitan area was 1.96, whereas the average number of children in the non-Tokyo metropolitan area was 2.16. The tendency for the average number of children to be smaller in the Tokyo metropolitan area than in the non-Tokyo metropolitan area was maintained when the data were controlled by birth cohort, educational background, and marital age. This way, Yamauchi [65] demonstrated the existence of contextual effects; however, the specific content of the contextual effect could not be clarified. Yamauchi [68] also used micro data from 15 social surveys conducted between 2000 and 2010 to calculate the average number of children and parity expansion rate. The results showed that the average number of children and the parity expansion rate from the first to the second child in the southern Kanto region (Saitama, Chiba, Tokyo, and Kanagawa prefectures) were significantly lower than those in regions except Hokkaido and Kinki regions (Shiga, Kyoto, Osaka, Hyogo, Nara, and Wakayama prefectures), and the parity expansion rate from the second to the third child in the southern Kanto region was significantly lower than those in other regions, except Hokkaido, Northern Kanto region (Ibaraki, Tochigi, and Gumma prefectures), Tokai region (Gifu, Shizuoka, Aichi, and Mie prefectures), and Kinki region. Yamauchi et al. [69] used micro data from the 8th (2016) “National Survey on Migration” as their data source and found that the average number of children of married women in the Tokyo metropolitan area (Saitama, Chiba, Tokyo, and Kanagawa prefectures) was lower than that in non-Tokyo metropolitan areas, even when age at marriage and residential migration experience were treated as control variables, indicating the existence of contextual effects.

Koike discussed the possibility of internal migration leading to low fertility in Japan’s metropolitan areas [70,71,72]. Koike divided Japan into metropolitan (Tokyo, Saitama, Chiba, Kanagawa, Gifu, Mie, Osaka, Kyoto, and Hyogo prefectures) and non-metropolitan areas, and established four migration categories: stayers within non-metropolitan areas, stayers within metropolitan areas, migrants from non-metropolitan areas to metropolitan areas, and migrants from metropolitan areas to non-metropolitan areas [70,71,72]. Four hypotheses were proposed regarding the mechanism by which internal migration reduces fertility: Selection, Disruption, Adaptation, and Socialization [41]. Koike [70] analyzed the 5th (2001) “The National Survey on Migration” and found that the Adaptation hypothesis was more likely, whereas Koike [71,72] analyzed the 6th (2006) and 7th (2011) “The National Survey on Migration” and found that the Selection and Adaptation hypotheses were more likely.

Finally, high fertility trends have been observed in Okinawa Prefecture, a remote island region in Japan [73,74,75]. Yamauchi et al. [75] compared fertility behavior and family values between women living in Okinawa Prefecture and those living in other regions of Japan using data from the National Institute of Population and Social Security Research’s 4th “The National Survey on Family” and an original sample survey conducted in Okinawa Prefecture. The results showed that married women in Okinawa Prefecture had more children than those in mainland Japan–a trend that could not be explained by compositional effects alone. Nishioka [76] examined the parity expansion rate and sex ratio of the youngest child for each sex combination of children among women over 40 years old living in Okinawa Prefecture. The results showed that individuals in Okinawa Prefecture had strong patrilineal successionism and tended to continue giving birth until the birth of a male child, contributing to the high level of fertility in the prefecture. Thus, it is widely believed that, in Okinawa Prefecture, family values such as the desirability of having many children, tolerance for pregnancy before marriage, and patrilineal succession tend to have contextual effects and contribute to the high fertility rate [75,76]. However, Sawada [73] found that the availability of fertility control measures in Okinawa Prefecture, where abortion and contraception were illegal under U.S. military rule until 1972, differed significantly from those in mainland Japan, and this might have contributed to regional differences in fertility.

As shown above, Japanese population geographers have introduced methodologies from Western countries and reported regional differences in fertility within Japan. These reports provide useful references for interpreting fertility statistics in Japan. However, in Japan, regional differences in fertility are often interpreted as a dichotomy between metropolitan and non-metropolitan areas, or urban and rural areas, whereas spatial variation in fertility within metropolitan areas has not received considerable attention. Observing the distribution of fertility in metropolitan areas and clarifying its mechanisms are important for Japan’s population and national land policies. Therefore, the present study addresses the following research question: How is the development of Japan’s metropolitan area related to fertility distribution? To answer this question, we compared fertility rates between central cities and suburban areas within Japan’s metropolitan areas and interpreted the results using discussions by Japanese urban geographers about changes in metropolitan structure.

3. Materials and Methods

As longitudinal data using administrative record information have not been developed in Japan, cross-sectional data must be used to examine regional differences in fertility. “The National Fertility Survey” conducted by the National Institute of Population and Social Security Research has a small sample size, making it difficult to use it for analysis at the prefectural level [64]. For example, in the 16th survey, conducted in June 2021, 9401 questionnaires were distributed to married couples, of which only 6834 were valid. Therefore, this study used the total fertility rate by municipality from the “Specified Report of Vital Statistics” (https://www.mhlw.go.jp/toukei/list/list58-60.html (accessed on 20 November 2025)) as a fertility indicator. These are statistics on births, deaths, stillbirths, marriages, and divorces calculated by the Ministry of Health, Labour and Welfare and published by the Japanese government (https://www.e-stat.go.jp/ (accessed on 20 November 2025)). In municipalities with lower populations, the number of births is low, and the total fertility rate may be unstable owing to contingencies. Therefore, in the “Specified Report of Vital Statistics,” the total fertility rate by municipality is calculated from five-year statistics (female population in the 5-year age group and number of births by mother’s 5-year age group), and a Bayesian estimation is applied to stabilize the statistical values. Specifically, statistics for the years 1983–1987, 1988–1992, 1993–1997, 1998–2002, 2003–2007, 2008–2012, and 2013–2017 are available.

Table 1. Number of municipalities for which the total fertility rate (TFR) is calculated.

	2000	2010	2015
Municipalities for which the TFR is calculated	3356	1897	1896
Municipalities in Metropolitan Employment Areas (MEAs)	1573	1091	1075
Central cities in MEAs	263	288	276
Suburbs in MEAs	1310	803	799
Municipalities in Micropolitan Employment Areas (McEAs)	669	319	332
Central cities in McEAs	137	93	96
Suburbs in McEAs	532	226	236

Source: Specified Report of Vital Statistics, CSIS (https://www.csis.u-tokyo.ac.jp/UEA/index.html; accessed on 24 August 2024).

lists the municipalities for which the total fertility rate is calculated. The large decrease in the number of municipalities from 2000 to 2010 is due to the “the Heisei municipal mergers,” (1999–2010) and should be considered when comparing the results over time.

In this study, the “Urban Employment Area” (hereinafter referred to as UEA) was used as a criterion for classifying municipalities into central cities, suburbs, and municipalities that are not included in metropolitan areas (Figure 3). UEA was established by Kanemoto and Tokuoka [77], and data on the UEA are available on the Web (https://www.csis.u-tokyo.ac.jp/UEA/index.html (accessed on 24 August 2024)). In the definition of UEA, a municipality is considered a central city when (1) it has a Densely Inhabited Districts (DID) population of 10,000 or more and is not a suburb of another city or (2) it meets the conditions of a suburban municipality but has a permanent employee population ratio of at least 1 and a DID population of at least one-third of that of the central municipality or a DID population of 100,000 or more. A municipality is considered a primary suburban municipality when its commuting rate to the central city is at least 10%, and a municipality is considered a secondary or lower suburban municipality when its commuting rate to the suburban municipality exceeds 10% and no other municipality with a higher rate exists. Thus, though the definition of UEA is complex, it reflects the reality of Japanese urban clusters with complex exchange patterns in a small country [77].

UEAs whose central cities have DID populations of at least 50,000 are called “Metropolitan Employment Areas” (hereinafter referred to as MEA) and UEAs whose central cities have a DID population of at least 10,000 and less than 50,000 are called “Micropolitan Employment Areas” (hereinafter referred to as McEA). When the total fertility rates within UEAs are presented in this research, the results of ten major MEAs (Tokyo, Osaka, Nagoya, and ordinance-designated cities as of 2000), other MEAs, and McEAs are presented in order.

Prior to the development of the UEA data, individual researchers defined metropolitan areas based on their own criteria [77] (p. 2). Despite official definitions of “Major Metropolitan Area” and “Metropolitan Area” by the Statistics Bureau of the Ministry of Internal Affairs and Communications of the Japanese government, only the special wards of Tokyo and ordinance-designated cities are considered the central cities constituting a “Major Metropolitan Area” and only cities with a population of 500,000 or more that are not included in a “Major Metropolitan Area” are considered the central cities constituting a “Metropolitan Areas” (https://www.stat.go.jp/data/kokusei/2000/guide/2-01.htm (accessed on 20 November 2025)). Therefore, the analysis was limited to large metropolitan areas when using these definitions. For example, based on the results of the 2020 Census, the following 11 “Major Metropolitan areas” were set up: “Sapporo Major Metropolitan Area,” “Sendai Major Metropolitan Area,” “Kanto Major Metropolitan Area,” “Niigata Major Metropolitan Area,” “Shizuoka and Hamamatsu Major Metropolitan Area,” “Chukyo Major Metropolitan Area,” “Kinki Major Metropolitan Area,” “Okayama Major Metropolitan Area,” “Hiroshima Major Metropolitan Area,” “Kitakyushu and Fukuoka Major Metropolitan Area,” and “Kumamoto Major Metropolitan Area.” There are only 3 “Metropolitan Areas”: “Utsunomiya Metropolitan Area,” “Matsuyama Metropolitan Area,” and “Kagoshima Metropolitan Area.”

UEA is similar to the criteria for establishing metropolitan areas used in previous studies in European countries and has the advantage of facilitating international comparisons. For example, Kulu and Boyle [48] set the criterion for central cities as cities and towns with a population of 30,000 or more, and the standard commuting rate for suburban municipalities as 10% or more. Rodrigo-Comino et al. [34] set the standard for a central city as a municipality with a population of 50,000 or more, and the standard commuting rate for suburban municipalities as 15%, which is in accordance with Eurostat’s Functional Urban Area. In the definition of “Major Metropolitan Areas” and “Metropolitan Areas” by the Statistics Bureau of the Ministry of Internal Affairs and Communications, the standard commuting rate for suburban municipalities is set at 1.5% or higher (https://www.stat.go.jp/data/kokusei/2000/guide/2-01.htm (accessed on 20 November 2025)). Table 1 shows the results of the classification and tabulation of the central cities of the MEAs or McEAs and the suburbs of the MEAs or McEAs.

4. Results

4.1. Major Metropolitan Employment Areas

Figure 4 shows the total fertility rate by municipality in the Tokyo MEA from 2000 to 2015. The total fertility rate for the central city is shown in orange, the total fertility rate for the suburbs is in blue, and the dots in the figure are organized by taking prefectures and government-designated cities on the vertical axis. The central city municipalities of the Tokyo MEA, which have a lower total fertility rate than any suburban municipality, were 13 municipalities in 2000 (Shibuya, Meguro, Nakano, Suginami, Shinjuku, Setagaya, Toshima, Bunkyo, Chiyoda, Shinagawa, Minato, Taito wards, and Musashino city), eight municipalities in 2010 (Toshima, Shinjuku, Nakano Meguro, Shibuya, Setagaya, Bunkyo wards, and Musashino City), and only five municipalities in 2015 (Toshima, Nakano, Shinjuku, Meguro, and Suginami wards). The total fertility rate by ward for Yokohama, Kawasaki, and Chiba was about 1.10–1.50 in all years, which can be interpreted as being at the same level as that of suburban municipalities.

As shown in Figure 4, the trend of extremely low total fertility rates in some administrative wards in the central city of the Tokyo MEA did not change from 2000 to 2015. Certainly, in the central city, the total fertility rate has increased in administrative wards such as Minato (0.94 in 2000, 1.27 in 2010, and 1.39 in 2015), Koto (1.13 in 2000, 1.24 in 2010, and 1.35 in 2015), and Chuo (1.01 in 2000, 1.1 in 2010, and 1.39 in 2015) wards. However, the impact of the population recovery in the central city on the regional differences in fertility between the central city and suburbs was limited.

In the Osaka MEA (Figure 5), suburban municipalities with extremely low total fertility rates exist, such as Toyono Town in Osaka Prefecture (total fertility rate of 1.0 in 2000, 0.82 in 2010, 0.84 in 2015) and Heguri Town in Nara Prefecture (total fertility rate of 1.05 in 2000, 1.07 in 2010, 1.15 in 2015). Even if these two suburban municipalities are excluded as special cases, there were eight central city municipalities with total fertility rates lower than any of the suburban municipalities in 2000 (Kita, Chuo, Naniwa, Tennoji, Nishinari, Nishi, Abeno, and Miyakojima wards), four municipalities in 2010 (Naniwa, Kita, Chuo, and Nishi wards), and three municipalities in 2015 (Naniwa, Kita, Chuo, and Nishi wards).

The Osaka MEA also shows a trend of low total fertility rates in some of the administrative wards of the central city and a tendency for fertility recovery to be observed only in a limited spatial area. In the case of the central city of Osaka MEA, the fertility rates in Konohana (1.34 in 2000, 1.45 in 2010, and 1.54 in 2015), Tennoji (1.05 in 2000, 1.16 in 2010, and 1.29 in 2015), and Abeno (1.07 in 2000, 1.14 in 2010, and 1.31 in 2015) wards have recovered by more than 0.2.

Figure 6 shows a similar trend for the Nagoya MEA, with seven central city municipalities having lower total fertility rates than any of the suburban municipalities in 2000 (Naka, Showa, Higashi, Nakamura, Chikusa, Mizuho, and Atsuta wards), six municipalities in 2010 (Naka, Showa, Nakamura, Higashi, Chikusa, and Atsuta wards), and only two in 2015 (Naka and Higashi wards). In the case of the central city of Nagoya MEA, Chikusa (1.07 in 2000, 1.17 in 2010, 1.28 in 2015), Nakamura (1.06 in 2000, 1.13 in 2010, 1.30 in 2015), Showa (1.04 in 2000, 1.12 in 2010, 1.28 in 2015), Mizuho (1.12 in 2000, 1.26 in 2010, 1.39 in 2015), and Moriyama ward (1.45 in 2000, 1.65 in 2010, 1.66 in 2015). Notably, with the exception of the Naka and Higashi wards, the fertility levels of the central cities and suburbs were similar in Nagoya MEA in 2015. This trend differed significantly from those of Tokyo MEA and Osaka MEA.

Figure 7 shows the total fertility rates for the central cities and suburbs of the government-designated MEAs in 2000 (Sapporo, Sendai, Kyoto, Kobe, Hiroshima, Kitakyushu, and Fukuoka). In these MEAs, the total fertility rate is particularly low in the location of city halls, as exemplified by Chuo Ward, Sapporo (0.9 in 2000), Aoba Ward, Sendai (1.08 in 2000), Nakagyo Ward, Kyoto (0.94 in 2000), Chuo Ward, Kobe (0.98 in 2000), Naka Ward, Hiroshima City (0.93 in 2000), and Chuo Ward, Fukuoka City (0.81 in 2000). In contrast, Kitakyushu MEA continued the trend of similar levels of fertility in the central city and the suburbs from 2000 to 2015.

Among the central cities shown in Figure 7, the total fertility rate of 15 administrative wards recovered to more than 0.2 between 2000 and 2015. These included Shiroishi Ward, Sapporo City (1.09 in 2000, 1.2 in 2010, 1.29 in 2015), Higashinada Ward, Kobe City (1.25 in 2000, 1.28 in 2010, 1.46 in 2015), Nada Ward, Kobe City (1.14 in 2000, 1.23 in 2010, 1.35 in 2015), Tarumi Ward, Kobe City (1.31 in 2000, 1.45 in 2010, 1.56 in 2015), Naka Ward, Hiroshima City (0.94 in 2000, 1.09 in 2010, 1.15 in 2015), Minami Ward, Hiroshima City (1.21 in 2000, 1.35 in 2010, 1.48 in 2015), Asaminami Ward, Hiroshima City (1.58 in 2000, 1.71 in 2010, 1.80 in 2015), Aki Ward, Hiroshima City (1.43 in 2000, 1.78 in 2010, 1.72 in 2015), Wakamatsu Ward, Kitakyushu City (1.33 in 2000, 1.41 in 2010, 1.55 in 2015), Tobata Ward, Kitakyushu City (1.39 in 2000, 1.53 in 2010, 1.61 in 2015), Yawatanishi ward, Kitakyushu City (1.44 in 2000, 1.55 in 2010, 1.65 in 2015), Higashi Ward, Fukuoka City (1.28 in 2000, 1.42 in 2010, 1.5 in 2015), Minami Ward, Fukuoka City (1.19 in 2000, 1.27 in 2010, 1.39 in 2015), Nishi Ward, Fukuoka City (1.37 in 2000, 1.46 in 2010, 1.57 in 2015), and Sawara Ward, Fukuoka City (1.25 in 2000, 1.38 in 2010, 1.45 in 2015).

In the ten major MEAs listed above, the average total fertility rate in the suburbs was higher than the average total fertility rate in the central city for the entire period from 2000 to 2015. However, as shown in Figure 4, Figure 5, Figure 6 and Figure 7, variations in fertility exist among the municipalities and wards that comprise the central cities, and only a few municipalities and wards have lower total fertility rates than suburban municipalities.

4.2. Other Metropolitan Employment Areas

Figure 8, Figure 9 and Figure 10 show the distribution of the total fertility rate between the central city and suburbs in the MEAs, excluding the ten MEAs mentioned in the previous Section. The distribution ranges from 1.11 to 1.54 in Hokkaido Prefecture (from Hakodate to Chitose), from 1.22 to 1.96 in the Tohoku region (from Aomori to Iwaki), from 1.19 to 1.73 in the Kanto region (from Mito to Odawara), and from 1.2 to 1.8 in the Chubu region (from Niigata to Matsusaka). The total fertility rate is distributed within the range of 1.22 to 2.03 in the Kinki region (from Hikone to Wakayama), 1.14 to 1.75 in the Chugoku region (from Tottori to Iwakuni), 1.19 to 1.83 in the Shikoku region (from Tokushima to Kochi), and 1.13 to 1.89 in the Kyushu region (from Omuta to Kagoshima). The total fertility rate in Okinawa Prefecture (Naha and Okinawa), where unique contextual effects have been pointed out in previous studies [75,76], is distributed within the range of 1.55 to 2.02.

The above results comparing fertility distributions among the MEAs suggest the following three points: First, the differences in fertility between Northeast, Central, and Southwest Japan, which have been pointed out in previous studies [60], have been obscured since at least the 2000s. Second, diversity exists even within the same region, such as in MEAs with high and low fertility. For example, Aizu-Wakamatsu MEA in 2000 (with a minimum value of 1.67 and a maximum value of 1.96) is prominent in the Tohoku region, and Joetsu MEA in 2000 (with a minimum value of 1.54 and a maximum value of 1.8) is prominent in the Chubu region. Third, except for the ten cities mentioned in the previous Section, no municipalities in MEAs had a total fertility rate below 1.0 in any year.

Figure 8, Figure 9 and Figure 10 show that the hypothesis that “central cities have lower total fertility rates than suburbs” is not necessarily true for any given year. MEAs with the lowest total fertility rates in central cities were only 24 of 99 (24.2%) in 2000, 23 of 90 (25.6%) in 2010, and 19 of 84 (22.6%) in 2015. Contrarily, MEAs with the highest total fertility rate in the central city are also small in number: 13 out of 99 (13.1%) in 2000, 19 out of 90 (21.1%) in 2010, and only 15 out of 84 (17.9%) in 2015. The number of MEAs with neither maximum nor minimum total fertility rates in the central city is large in all years: 62 out of 99 (62.6%) in 2000, 48 out of 90 (53.3%) in 2010, and 50 out of 84 (59.5%) in 2015.

The MEAs where the average total fertility rate in the central city was lower than the average rate in the suburbs accounted for 53 of 99 (53.5%) of all MEAs in 2000, 41 of 90 (45.6%) in 2010, and 40 of 84 (47.6%) in 2015. In contrast, 44 of 99 (44.4%), 46 of 90 (51.1%), and 41 of 84 (48.8%) MEAs had higher average total fertility rates in their central cities than in their suburban areas in 2000, 2010, and 2015, respectively. As of 2015, Joetsu, Hikone, and Niihama fell into patterns in which the average values in the central cities and those in the suburbs are almost equal (coincident with two decimal places).

The spatial distribution of MEAs with relatively high fertility in central cities and suburban areas is geographically dispersed to some extent. In 2000, Aomori (central city total fertility rate 1.35, suburban average 1.38), Morioka (central city total fertility rate 1.31, suburban average 1.5), Niigata (central city total fertility rate 1.27, suburban average 1.49), Kanazawa (central city total fertility rate 1.35, suburban average 1.46), Yamaguchi (central city total fertility rate 1.4, suburban average 1.53), Tokushima (central city total fertility rate 1.32, suburban average 1.47), Saga (central city total fertility rate 1.52, suburban average 1.61), Nagasaki (central city total fertility rate 1.29, suburban average 1.52), and Kumamoto (central city total fertility rate 1.43, suburban average 1.55) can be cited as representative examples of relative high fertility in suburbs. These MEAs have the prefectural capital as their central city, which, together with the ten cities introduced in the previous Section, may suggest that fertility tends to be dispersed to the suburbs in MEAs with strong political and economic centripetal forces.

Contrarily, MEAs with a pattern of relatively high fertility in the central city are distributed in the periphery of major MEAs (e.g., Tokyo, Osaka, and Nagoya), such as Mito (central city total fertility rate 1.45, suburban average 1.42), Utsunomiya (central city total fertility rate 1.48, suburban average 1.46), and Takasaki (central city total fertility rate 1.49, suburban average 1.41). Total fertility rate in central cities is relatively high in Nagano (central city total fertility rate 1.5, suburban average 1.497) and Matsumoto (central city total fertility average 1.55, suburban average 1.48), which have vast mountainous areas on their outer edges, and Okayama (central city total fertility rate 1.51, suburban average 1.41), Takamatsu (central city total fertility rate 1.54, suburban average 1.44), and Matsuyama (central city total fertility rate 1.29, suburban average 1.27), which are located along the coast of the Seto Inland Sea.

4.3. Micropolitan Employment Areas

Figure 11, Figure 12 and Figure 13 show the distribution of the total fertility rate between the central city and suburbs within McEAs. For example, in 2000, the distribution ranged from 1.23 to 1.79 in Hokkaido prefecture (from Rumoi to Shizunai), from 1.32 to 2.05 in the Tohoku region (from Goshogawara to Nihommatsu), from 1.27 to 1.89 in the Kanto region (from Shimodate to Tateyama), from 1.19 to 1.98 in Chubu region (from Kashiwazaki to Owase), from 1.29 to 1.94 in the Kinki region (from Nagahama to Shingu), from 1.38 to 2.06 in the Chugoku region (from Kurayoshi to Hagi), from 1.44 to 1.81 in the Shikoku region (from Marugame to Nakamura), from 1.34 to 2.29 in Kyushu region (from Iizuka to Kokubu) and from 1.77 to 2.45 in Okinawa Prefecture (Hirara and Nago). Total fertility rates of McEAs are generally higher than those of MEAs shown in the previous Section. In addition, we found a low fertility trend in the Hokkaido region, high fertility trend in the southern Tohoku region (from Yonezawa to Nihonmatsu), high fertility trend in Saga, Nagasaki, and Kumamoto prefectures (from Karatsu to Yamaga), and high fertility trend in Kagoshima and remote island areas in Okinawa (Naze, Taira, and Nago), suggesting that the differences in McEA’s total fertility rate among the regions are more conspicuous than that of McEAs’ total fertility rate.

The McEAs with the smallest central city total fertility rates were only 30 of 132 (22.7%) in 2000, 18 of 89 (20.2%) in 2010, and 29 of 92 (31.5%) in 2015. In contrast, McEAs, with the largest central city total fertility rate were 44 of 132 (33.3%) in 2000, 44 of 89 (49.4%) in 2010, and 42 of 92 (45.7%) in 2015. The number of McEAs with neither maximum nor minimum total fertility rates in the central city was 58 of 132 (43.9%) in 2000, 25 of 89 (28.1%) in 2010, and 20 of 92 (21.7%) in 2015.

The number of McEAs for which average total fertility rate in the central city was lower than that in the suburbs was 60 out of 132 (45.5%) in 2000, 27 out of 89 (30.3%) in 2010, and 40 out of 92 (43.5%) in 2015. The average total fertility rate in the central city was higher than that in the suburbs in 69 of 132 (52.3%) McEAs in 2000, 61 of 89 (68.5%) in 2010, and 52 of 92 (56.5%) in 2015. Considering McEAs, there were more cases of relatively low total fertility in the suburbs than in the central cities in all years, and a strong tendency for fertility to return to the central cities was observed, especially in 2010. McEAs whose average value of the central city and average value of the suburbs were very close (when they are equal to the second decimal places) included Tochigi, Tanabe, and Nichinan in 2000 and Takayama in 2010.

Trends similar to those observed for MEAs are confirmed in the spatial distribution of McEAs, with relatively low fertility in the suburbs. First, many cases are located on the outer edges of major MEAs (e.g., Tokyo, Osaka, and Nagoya). Specific examples include Tatebayashi (total fertility rate of central city is 1.55 and suburban average is 1.33), Tomioka (total fertility rate of central city is 1.46 and suburban average is 1.35), Chichibu (total fertility rate of central city is 1.58 and suburban average is 1.51), Nagahama (total fertility rate of central city is 1.71 and suburban average is 1.51), and Yokaichi (total fertility rate of central city is 1.7 and suburban average is 1.61). Second, this pattern is more common in McEAs in Nagano Prefecture, which has a vast mountainous area on its outer edge, such as Ueda (central city total fertility rate 1.58, suburban average 1.55), Okaya (central city total fertility rate 1.56, suburban average 1.51), Suwa (central city total fertility rate 1.66, suburban average 1.59), Saku (central city total fertility rate 1.73, suburban average 1.61), and Ina (central city total fertility rate 1.73, suburban average 1.65). Third, there are many McEAs with relatively high fertility in the central cities located along the Seto Inland Sea coast, including Marugame (central city total fertility rate 1.7, suburban average 1.46), Kanonji (central city average 1.63, suburban average 1.53), and Uwajima (central city total fertility rate 1.74, suburban average 1.68).

5. Discussion

5.1. Do Suburbs Have Higher Fertility than Central Cities?

One of the purposes of this study was to observe regional differences in fertility within Japanese metropolitan areas and to test the validity of the hypothesis that fertility is low in central cities and high in suburban areas. We compared the average fertility values between municipalities comprising central cities and those comprising suburbs and focused on the ranking of fertility in the central city within each MEA or McEA. The results of our study indicate that the tendency that fertility in central cities is lower than that in suburbs applies specifically to metropolitan areas with huge populations but not to all metropolitan areas in Japan.

The following three points are clear regarding the spatial distribution patterns of fertility in metropolitan areas. First, in ten major MEAs (Tokyo, Osaka, Nagoya, and ordinance-designated cities as of 2000 (Sapporo, Sendai, Kyoto, Kobe, Hiroshima, Kitakyushu, and Fukuoka), some wards in the central cities have significantly lower total fertility rates than those in the suburbs; however, the spatial range is extremely limited. In some central cities, such as Yokohama, Saitama, Chiba, and Kitakyushu, the total fertility rate is almost the same as that of the suburbs.

Second, in many MEAs or McEAs in Japan, the total fertility rate in the central city is higher than that in the suburbs, and the tendency of “low fertility in the central city and high fertility in the suburbs” is not absolute. The pattern of the lowest total fertility rate in the central city is only a quarter of MEAs and about 20–30% of McEAs. The pattern in which the average total fertility rate in the central city is lower than the average total fertility rate in the suburbs is about half of that and is in the 30–40% range among McEAs.

Third, when the distribution pattern of fertility within MEAs or McEAs was plotted on a map (Figure 14 and Figure 15), spatial maldistribution became clear. MEAs or McEAs, where the fertility of the central city is lower than that of the suburbs, are distributed not only in the ten major MEAs (Tokyo, Osaka, Nagoya, and government-designated cities as of 2000) but also in many MEAs with prefectural capitals (e.g., Aomori, Morioka, Niigata, Kanazawa, Yamaguchi, Tokushima, Saga, Nagasaki, and Kumamoto). On the other hand, MEAs or McEAs with higher fertility in the central city than in the suburbs are concentrated at the outer edges of major MEAs (e.g., Mito, Maebashi, Takasaki, and eastern Shiga prefectures), Nagano Prefecture, which has many mountainous areas (e.g., Nagano and Matsumoto), and the Seto Inland Sea coastal areas (e.g., Okayama, Takamatsu, and Matsuyama).

The results of this study indicate that, at least in the case of Japan, it is insufficient to capture regional differences in fertility through a simple contrast between low fertility in central cities (urban centers) and high fertility in suburban areas, as pointed out by [48,49]. A more accurate way to describe regional differences in fertility in Japan would be to argue that “a part of areas in central cities of Japan’s metropolitan areas have significantly lower fertility than the suburbs” or that “some metropolitan areas in Japan have higher fertility in the suburbs while others have higher fertility in the central cities.”

5.2. Relationship Between Fertility Distribution and the Urban Life Cycle Model

Another important question this study addressed was whether the regional differences in fertility observed within or across metropolitan areas in Japan are related to the urban life cycle model, as in [34]. Based on our findings and the findings of Japanese urban geography studies, we conclude that the impact of the urban life cycle on regional differences in fertility within the metropolitan areas of Japan is limited. As shown in Section 4.1, in Japan’s major metropolitan areas, there are some wards in the central city where the total fertility rate increased between 2000 and 2015. However, the trend of extremely low fertility rates in some administrative wards in the central city did not significantly change between 2000 and 2015.

Why are there many areas where the results obtained in this study do not agree with those of [34]? To answer this question, we need to focus on the two differences in population recovery in urban centers observed between Japan, Europe, and the United States. First, the spatial scale at which population recovery is observed in urban centers differs significantly among Japan, Europe, and the United States. Ref. [20] (p. 206) states that, in large cities in Europe and the U.S., population recovery is observed in all cities or metropolitan areas, whereas in large cities in Japan, population recovery is observed only in some administrative districts. This study shows that, in Japan’s major metropolitan areas, the trend of low fertility has been maintained in some administrative wards of the central city, whereas fertility has recovered in some administrative wards of the central city. Although these results appear contradictory at first glance, they may be related to the small spatial scale of population recovery in Japan’s major cities.

Second, in Japan, condominium development is the most important factor in population recovery in central cities (urban centers) [29]. Japanese urban geographers have explored the socioeconomic attributes and residential history of condominium dwellers to elucidate the factors contributing to population recovery in central cities (urban centers) (e.g., [20,22,23,24,25,26,27,28,32,33]). These reports are useful for discussing whether population recovery in urban centers, as well as discussion in Rodrigo-Comino et al. [34], causes a recovery in fertility because they show whether condominium development and population recovery in urban centers have increased the demographic attributes that improve fertility.

In the early 2000s, many reports noted that most condominium residents in the urban centers of major Japanese cities were single-person households or married-couple households without children. In other words, Japanese urban geographers have shown results that imply that condominium development in urban centers depresses rather than increases fertility. For example, Yabe [22], who surveyed Minato Ward, Tokyo, in October 2001, reported that 37.0% of private condominium residents were single-person households under the age of 65, 27.6% were married couples under the age of 65 with no children, and 18.9% were married couples under the age of 65 with children. Approximately 80% of the single-person households under the age of 65 years were female. Yabe [22] conducted the same survey in public housing in the same area and reported that 4.1% of the residents were single-person households under the age of 65 years, 13.7% were married couples under the age of 65 years with no children, and 43.8% were married couples aged <65 years with children. Tomita [25], who conducted a questionnaire survey in four condominiums in central Osaka from February to April 2004, reported that 31.5% of those surveyed were single-person households, 33.8% were married couples without children, and 21.9% were nuclear families. In Tomita’s study [25], women headed 62.3% of single-person households. Several similar surveys have been conducted in Sendai City. Hirose [28] surveyed the Aoba Ward in Sendai City in 1997 and reported that more than half of the households that moved into a new condominium were single-person households or married couples without children, whereas only 26.2% were households with children. Sakakibara et al. [27] conducted a questionnaire survey of 37 condominiums in central Sendai in November 2001 and reported that single-person households and married-couple households without children constituted the majority and that approximately 60% of single-person households were women. However, approximately 46% of households in this study [27] had children.

However, some reports have shown that condominium development in urban centers contributes to the recovery of fertility. Yabe’s [22] questionnaire survey in the Toyosu area of Koto Ward in October 2009 found that nearly 40% of skyscraper condominium residents had young children. Specifically, 7% of those surveyed were single households, 25% were households without children, 37% had children under the age of 20 living with them, and 30% had children over the age of 20 living with them or with children living away from home. According to Yabe [22], approximately 30% of households with young children living in skyscraper condominiums stated that their children’s education was their reason for moving.

The results of the above studies indicate that some condominium developments contribute to the fertility recovery in urban centers, whereas some do not. These differences may be explained by differences in the year of the survey and not just district differences. Kubo and Yui [78], who examined the supply of condominiums by major real estate agents in central Tokyo, noted that while “compact condominiums” for small households have been actively supplied since the late 1990s, condominiums to meet the demands of nuclear families have also been supplied since the late 2000s. Furthermore, we found that some urban central wards experienced a recovery in fertility from the 2000s to the 2010s (e.g., Minato, Koto, and Chuo wards in Tokyo), possibly as a result of the supply of housing available to families with children. On the other hand, it is possible that there was not a large supply of housing available for families with children in the urban centers of metropolitan areas, such as Sendai MEA, where a recovery in fertility has not been confirmed.

What can be said with certainty is that, at least in the case of Japan, there is no simple causal relationship whereby condominium development and population recovery in urban centers directly contribute to the fertility recovery. In addition, it is clear that the population recovery in the urban centers of Japan’s major cities has not had the influence to disrupt the contrast between low fertility in the urban centers and high fertility in the suburbs. Despite many differences between the results of this study and those of Rodrigo-Comino et al. [34], both share the assertion that we need to focus on cycles of real estate supply and population growth when examining fertility fluctuations in a metropolitan area. However, to accurately discuss the relationship between the life cycle and fertility recovery in the future, we must focus not only on the number of properties supplied and population growth but also on changes in population attributes. Unfortunately, Tokyo is the only city where the attributes of population groups that contributed to population recovery in the city center from the late 1990s to the late 2010s have been continuously examined.

5.3. The Diversification Process of Regional Differences in Fertility Within Japanese Metropolitan Areas

Section 5.1 argued that the contrast of lower fertility in central cities compared to suburbs is by no means absolute. This implies that the traditional theories of population and metropolitan development cannot fully explain the findings of this study. Furthermore, Section 5.2 demonstrated that the relationship between regional differences in fertility within metropolitan areas and the urban life cycle is not direct, and that the mechanism proposed by Rodorigo-Comino et al. [34] does not apply well to Japan. Therefore, based on the analysis results, this study presents an original model illustrating the process by which regional differences in fertility within Japanese metropolitan areas diversified (Figure 16). This model argues that regional differences in fertility within Japanese metropolitan areas are determined by two factors: the scale of metropolitan spatial expansion prior to the 2000s, and the occurrence of population recovery in central cities since the 2000s.

First, the scale of metropolitan spatial expansion prior to the 2000s determines the fundamental structure of population reproduction within metropolitan areas, whether high fertility areas are distributed in the central city or the suburbs. On the one hand, as a specific example of metropolitan areas where the fertility rate in suburbs is higher than in central cities, there are ten major MEAs (Tokyo, Osaka, Nagoya, and government-designated cities as of 2000 (Sapporo, Sendai, Kyoto, Kobe, Hiroshima, Kitakyushu, and Fukuoka)) or many MEAs with prefectural capitals (e.g., Aomori, Morioka, Niigata, Kanazawa, Yamaguchi, Tokushima, Saga, Nagasaki, and Kumamoto). On the other hand, even before the 2000s, when population recovery began occurring in central cities in Japan, a certain number of metropolitan areas existed where central cities had higher fertility rates than their suburbs (Figure 4, Figure 5, Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12 and Figure 13). The tendency for central cities to have higher fertility rates is frequently observed in metropolitan areas that are subordinate to other large metropolitan areas or located in mountainous regions.

Second, the occurrence of population recovery in central cities since the 2000s may reduce regional differences in metropolitan areas where low fertility regions are concentrated in central cities and high fertility regions in the suburbs. Chapter 4 confirmed a recovery trend in fertility rates in certain areas of central cities within major metropolitan areas (Figure 4, Figure 5, Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12 and Figure 13). In Japan, the phenomenon of population recovery in central cities is not a common trend, and the demographic composition of redevelopment areas is diverse, with some cases lowering fertility rates and others raising them. Therefore, when attempting to theorize the trends in fertility distribution within Japanese metropolitan areas, the framework does not become a simple one where all cities fit into an urban life cycle as seen in Western countries [34]. Instead, it diverges into various types based on the characteristics of metropolitan development and structural transformation.

However, the mechanisms determining regional differences in fertility are very complex, and it is important to note that contextual effects pointed out in previous studies are also at work, in addition to the model presented in this research. First, the high fertility trend in Okinawa Prefecture and the low fertility trend in the Tokyo MEA, identified in previous studies as contextual effects, are reflected in the total fertility rate by municipality. For example, the total fertility rates of the municipalities comprising the Naha MEA are distributed in the range of 1.5 to 2.3, those of the municipalities within the Okinawa McEA are distributed in the range of 1.7 to 2.5 and those within the Nago McEA in the range of 1.7 to 2.3. On the other hand, the total fertility rates of the municipalities comprising the Tokyo MEA are distributed within the range of 0.7 to 1.7. Second, we were unable to identify clear differences in fertility between Northeastern Japan, Central Japan, and Southwestern Japan. There are more detailed regional differences in fertility within the same region, such as the Aizu-Wakamatsu MEA in 2000 (minimum value of 1.67, maximum value of 1.96), which has higher fertility in the Tohoku region, and the Joetsu MEA in 2000 (minimum value of 1.54, maximum value of 1.8), which has higher fertility in Niigata Prefecture. This trend has not received considerable attention in previous studies conducted in Japan. Third, municipalities with extremely low total fertility rates below 1.0 are mostly concentrated in ten major MEAs (Tokyo, Osaka, Nagoya, and government-designated cities as of 2000 (Sapporo, Sendai, Kyoto, Kobe, Hiroshima, Kitakyushu, and Fukuoka). The only exception was the Moroyama McEA (total fertility rate of 0.94 in 2010 and 0.97 in 2015 in Moroyama town) adjacent to the Tokyo MEA. These results indicate that the distribution pattern of fertility in Japan is multilayered and that there is diversity in fertility among MEAs and McEAs.

6. Conclusions

This study showed the diversity in fertility among metropolitan areas in Japan and the distribution patterns of fertility within metropolitan areas. Metropolitan areas with a central city population of less than one million do not necessarily fall into the traditional contrast of low fertility in the central city (urban center) and high fertility in the suburbs. In Japan’s ten major metropolitan areas, fertility tends to be extremely low in some central city administrative wards, but many central city administrative wards exhibit fertility comparable to that of suburban areas.

In addition, this study examined the relationship between the life cycles of Japanese cities and regional differences in fertility. Despite some central city administrative wards in the major metropolitan areas where the total fertility rate increased between 2000 and 2015, the trend of extremely low total fertility rates in some central city administrative wards did not change significantly. In other words, it is difficult to assert that condominium development and population recovery in urban centers are directly related to fertility recovery.

Many of the results of this study are inconsistent with those of Rodrigo-Comino et al. [34], and we reject their hypothesis that the urban life cycle is associated with changes in fertility. However, this study and Rodrigo-Comino et al.’s study [34] agree that, when examining fertility variations within metropolitan areas, it is necessary to focus on the cycles of real estate supply and population growth. Population recovery has only been observed in a limited number of districts in the urban centers of Japan’s major metropolitan areas, and the demographics of condominium residents, a factor in population recovery, vary by district and year. Condominiums with more single female-headed households and households with no children would have the effect of pushing down fertility, whereas condominiums with more households with children would have the effect of improving fertility. Thus, in Japan, population recovery in urban centers does not directly affect fertility recovery. Therefore, a more detailed analysis is required not only of which locations within metropolitan areas are experiencing population growth but also of which types of households are targeted for housing supply.

The limitations of this study were as follows. First, the total fertility rate by municipality used in this study is a composite indicator, making it impossible to test for significant differences. Furthermore, it cannot observe regional differences within the same municipality between inner-city and suburban areas. The total fertility rate by municipality is not a perfect analytical indicator; therefore, it is necessary to supplement this study with analyses based on microdata collected through original questionnaire surveys and analyses of 500 m or 1 km grid survey data. These analyses are currently underway in a separate project, and we aim to publish them at a later date. Second, regarding the diverse spatial distribution patterns of fertility rates in Japanese small and medium-sized metropolitan areas, we were unable to objectively verify causes other than population recovery in central cities. However, it may be possible to overcome this problem by focusing on the real estate supply cycle and population growth, similar to the major Japanese metropolitan areas considered in this study. For example, in the MEAs or McEAs of Nagano Prefecture, which has vast mountainous areas on the periphery, total fertility rates are relatively high in the central city; however, this trend may be the result of a concentrated supply of housing for family households in the flat land of the central city. Third, the approach adopted in this study and that of Rodrigo-Comino et al. [34], who attempted to explain the mechanism of the spatial distribution of fertility within metropolitan areas from cycles of real estate supply and population growth, may underestimate the contextual effects of the surrounding environment on child-rearing households. Components of the contextual effects include not only the supply of housing for families, but also economic conditions, cultural factors, and the supply of amenities related to child-rearing and education. Therefore, by extracting areas with similar housing supply and comparing how family households’ reproductive behavior and residential mobility decisions change depending on differences in the conditions related to ‘the contextual effects,’ it is expected that we will be able to gain a deeper understanding of the specific content of the contextual effects.

Elucidating the mechanisms by which the diversity in fertility distribution arises is a future challenge. However, as mentioned in Section 3, longitudinal data using administrative record information is not available in Japan, and “The National Fertility Survey” and “The National Survey on Migration” conducted by the Ministry of Health, Labor, and Welfare have small sample sizes and cannot withstand analysis at the prefectural or metropolitan scale [64]. In recent years, population geography studies in Japan have made progress in analyzing longitudinal data, targeting low fertility trends in the Tokyo metropolitan area, high fertility trends in Okinawa Prefecture, and fertility contrasts between metropolitan and non-metropolitan areas. However, few analyses using cross-sectional data have been conducted in recent years, and the basic task of observing actual fertility distribution patterns has been neglected. This study introduced the diversity of spatial distribution patterns of fertility in Japan using cross-sectional data and presented new research questions, such as the mechanism by which fertility returns to central cities in some MEAs or McEAs and the reasons why fertility in certain MEAs or McEAs in a region is significantly higher than in others. In the future, Japanese population geographers must strive to ensure the three processes of improving statistical data, accurately observing population phenomena using cross-sectional data, and systematically clarifying mechanisms using a longitudinal data function.

Finally, we argue that not only population geographers but also urban geographers can contribute to the task of interpreting regional differences in fertility within Japan. Although Japanese urban geographers are interested in population recovery in urban centers, fertility recovery has never been examined. Therefore, the amount of research is insufficient for generalizing the relationship between the cycles of real estate supply, population growth, and fertility. If there were further reports on real estate supply and population changes in central Tokyo, as in Yamagami [20] and Yabe [22], and in the urban center of Sendai City, as studied by Sakakibara et al. [27] and Hirose [28], the interpretation of the variation in total fertility rates presented in this study would seem more convincing. In addition, in not only newly developed condominium districts but also in districts that have an important impact on population dynamics in the future, it is necessary to examine factors that directly affect fertility, such as the ratio of households with and without children and the residential patterns of women aged 15 to 50, which is the denominator of the total fertility rate. We hope that this study will provide population and urban geographers with the opportunity to examine their interpretations of regional differences in fertility within Japan in more detail.