Introduction

Ageing is a phenomenon that human societies are experiencing as a consequence of demographic transition (Kurek, 2011). Because older people usually have a limited working capacity and sometimes insufficient cognitive function to live by themselves, appropriate care should be provided by both the public and private sectors. Generally, rural populations are more aged compared to urban populations because rural-to-urban migration occurs more frequently in young people than in older people (Kinsella, 2001; Organisation for Economic Co-operation and Development [OECD], 2015). However, urban areas are also ageing, and the growth of their older populations is currently faster than the growth in rural areas (OECD, 2015; United Nations, Department of Economic and Social Affairs, 2015).

One of the challenges presented by urban ageing is that the spatial and temporal distribution of older people is highly heterogeneous, causing pronounced geographical disparities in ageing. For example, the mass migration of reproductive-age households to new towns (large-scale public residential estates) in the past could cause the sudden emergence of ageing hot spots (Miyaki et al., 2018). Very low fertility rates and/or out-migration of younger people can increase the proportion of older people in specific areas (Kurek, 2011; OECD, 2015). Residential age segregation can also occur because variations in neighbourhood characteristics attract people from different age groups (Feng, 2015; Sabater et al., 2017). Urban redevelopment in central districts, such as the construction of high-rise residential buildings, can also lead to population recovery and the growth of certain age groups (Kurek, 2011; Tsubomoto & Uozumi, 2012; Xie et al., 2016). Thus, there is a need to investigate the spatial and temporal distribution of ageing in urban areas to explain the population ecology of Homo sapiens who inhabit the urban environment.

Previous studies of urban ageing have investigated the spatial characteristics of ageing in urban areas. In a case study of 10 European cities, it was observed that, in older cities, the older populations were more widely scattered (Arup et al., 2015). In several studies, urban ageing is marked by a city-centre–suburb division. For example, in the same 10 European cities, population ageing was higher in the suburbs than in the city centre. In contrast, in Beijing, China, population ageing is higher in the city centre than in the suburbs, but heterogeneity occurs even within individual suburbs. Population ageing was found to decrease in areas surrounding the city centre (inner suburbs), but increase in the outer suburbs (Xie et al., 2016). Except for Japan, Korea, and the United Kingdom, most OECD metropolitan areas also display a similar spatial pattern of a greater proportion of older adults among the population in city centres, rather than in the suburbs (OECD, 2015).

Another group of studies that investigated the temporal characteristics of urban ageing revealed a tendency for the ageing population to move from city centre to the suburbs over time. In metropolitan areas in Germany, Belgium, Sweden, Denmark, Norway, and the United States, areas with a greater proportion of older adults among the population shifted from the city centre to suburban areas between 2001 and 2011 (OECD, 2015). The same pattern was observed in Brisbane, Australia, during 1996–2006 (Han & Corcoran, 2014). Based on projected population data, this reversal in the spatial pattern was predicted to occur in large cities in Germany between 1990 and 2025 (Swiaczny, 2010). This was because young families moved to suburban housing areas in the 1960s and 1970s, resulting in an older population in the city centre. Later, the young population began to migrate to the city centre, which decelerated the city-centre’s ageing process. At the same time, the suburban population aged rapidly due to low in-migration of the young, whereas the large cohorts who moved there in earlier decades are ageing in place (Naganuma et al., 2006; Kurek, 2011; OECD, 2015; Swiaczny 2010).

Most studies that have investigated the spatial aspects of urban ageing have used either structural (i.e., proportion of older people in the total population) or numerical (i.e., number or density of older people) measures. Structural ageing is a basic measure of the proportion of older people in an area (Cheng et al., 2012; Han & Corcoran, 2014; Kurek, 2011; OECD, 2015; Xie et al., 2016). This measure is commonly used to guide the funding and prioritisation of goods and services (Atkins & Tonts, 2016; Rowland, 2012). In contrast, numerical ageing is used to estimate the potential demand for services and infrastructure (Han & Corcoran, 2014; Atkins & Tonts, 2016); it must be determined even in areas with low structural ageing (Rowland, 2012). Because the two measures can reflect different aspects of urban ageing, it is theoretically useful to use both of them when studying the spatial and temporal heterogeneity of urban ageing (Han & Corcoran, 2014; Kaneyasu, 1987; Shiode et al., 2014). In their novel investigation of urban ageing in Perth, Australia, Atkins and Tonts (2016) proposed the use of the interplay between the change in numerical and structural ageing measures. The interplay of the change in numerical and structural ageing produces four types of ageing patterns: accumulation, decline, dilution, and concentration. Accumulation occurs when numerical and structural ageing are increasing, whereas decline occurs when both are decreasing. Dilution occurs when numerical ageing is increasing but structural ageing is decreasing. In contrast, concentration occurs when numerical ageing is decreasing but structural ageing is increasing. In their investigation in Perth, Australia, Atkins and Tonts (2016) showed that ageing accumulation was prevalent across the city, particularly in the suburbs, whereas the city centre experienced an ageing decline.

In the present study, we adopted the approach of Atkins and Tonts (2016) to investigate the spatial and temporal distribution of urban ageing in the Tokyo Metropolis (Tokyo prefecture) in Japan. Tokyo prefecture is the capital of Japan with 23 special wards (known as the ‘centre’ of Tokyo) and 39 municipalities. Tokyo had 1.2 million people aged 75 and older (9.4% of the total population) in 2015 (Statistics Bureau Ministry of Internal Affairs and Communication, 2019); this number is projected to increase to 2.3 million (16.7%) by 2045 (National Institute of Population and Social Security Research, 2018). The population of the city centre rapidly increased between 1945 and 1970, then slightly decreased until 1995 because of suburbanisation during the bubble economy. After the bubble economy ended, a population recovery was observed in the city centre that reflected the ‘Back to the City Centre’ phenomenon (PwC Japan Group, 2016; Wakabayashi & Koizumi, 2018).

Several studies have investigated the spatial distribution of the older adult population in Tokyo after the bubble economy using standard grid square data (i.e., data for small area grids approximately 250 m2, 500 m2, or 1 km2 with constant boundaries over time). Using 1-km grids, Naganuma et al. (2006) found that the proportion of older adults (aged 65 and older) among the population in 2000 was highest within 10 km of the former Tokyo Metropolitan Government Building and lowest in areas 30–50 km away. Wakabayashi and Koizumi (2018) used 500-m grids to examine the 23 special wards; they noted an increase in the proportion of older adults (aged 65 and older) among the population in 2005 compared to 1985. In 2005, the greatest proportion of older adults among the population was found within 3 km of the Tokyo Central Station, and the proportion was smaller in areas 3–15 km away.

We examined the spatial patterns and temporal trends of ageing occurring in 500-m grids by focusing on the population aged 75 years and older. The population aged 75 and older was analysed because these individuals are regarded as ‘later-stage older people’ or koki koreisha in Japan; they are likely to have a lower functional capacity and more disease burden, thus requiring greater medical care (Santoni et al., 2015; Suzuki, 2018). Specifically, this study aimed to: (1) classify geographic areas in Tokyo into the four types of population ageing; (2) examine changes in the spatial patterns and temporal trends of the types of population ageing from 2000 to 2015; and (3) determine the geographic structure of ageing using Tokyo’s city centre and train stations as reference points. Understanding how these types of ageing unfold within the city over time is essential for managing limited urban resources, and planning service delivery and infrastructure, with the aim of making Tokyo Metropolis an inclusive and sustainable city where older people can thrive.

Methods and Materials

Study Area

Tokyo Metropolis, referred as ‘Tokyo’ in this study, is the capital of Japan and one of its 47 prefectures. Tokyo includes 23 special wards (each governed as an individual city) and the Tama area with 26 cities, three towns, and one village (Fig. 1). Unlike the administrative area of Greater Paris or Greater London which is composed of the city centre and surrounded by suburbs, Tokyo has an elongated shape that covers about 2,194 km2, and extends for 90 km from east to west and 25 km from north to south. The western-most part of Tokyo is a mountainous area and is bordered by Mount Kumotori (Tokyo Metropolitan Government, n.d.), while the eastern-most part faces Tokyo Bay. The 23 special wards in the eastern part of Tokyo, especially wards around JR Yamanote Line stations, are centres of commercial and administrative activity. Although there is no official definition, Greater Tokyo Area commonly refers to Tokyo and three neighbouring prefectures that usually expands along major train routes to central Tokyo (Tokyo Metropolitan Government, n.d.) (Fig. 1).

Fig. 1
figure 1

Map of the population per 500-m grid (population density) in Tokyo Metropolis’ 23 special wards and the Tama area and its neighbouring prefectures in 2015

Full size image

Data and Data Sources

This study used population data at a spatial resolution of 500-m grids from the national population censuses conducted in 2000, 2005, 2010, and 2015. The 500-m grids were used because they were not influenced by changes in administrative boundaries over time, and they provided fine details regarding the heterogeneity of urban ageing (Shiode et al., 2014). The 2000 and 2005 statistical map databases in shapefile format were created by the PASCO Corporation and were provided by the Center for Spatial Information Science at the University of Tokyo under a collaborative project scheme (project number: 537). The 500-m grid statistics for 2010 and 2015 were downloaded from the official statistics portal (https://www.e-stat.go.jp/) of the Statistics Bureau of Japan. The locations of train stations in 2004, 2012, and 2015 were obtained from the Japan Ministry of Land, Infrastructure, Transport and Tourism, and are publicly available at https://nlftp.mlit.go.jp/ksj/.

Study Variables

In this study, we defined the ‘older population’ as the population aged 75 and older. Two measures of population ageing and changes were used. The first was numerical ageing, with population ageing and changes represented as the number of older adults in the population and the relative percentage change in number of older adults in the population between census periods, respectively. The second was structural ageing, with population ageing and changes represented as the older adult proportion of the overall population and the difference in the older adult proportion of the overall population between census periods. Based on the interplay between changes in the number and proportion of the older adult population, grids were classified into four types of ageing (accumulation, decline, dilution, and concentration; Atkins & Tonts, 2016) as explained above. Accumulation was caused by an older population moving into the area or staying in place, or a younger population moving out. Decline was caused by high mortality rates among the older population or by an older population moving out of the area, combined with a younger population moving into the area (Atkins & Tonts, 2016; Helminen et al., 2017). Dilution was caused by an older population moving into the area or ageing in place, combined with large numbers of younger individuals moving in or staying; therefore, the increase in the older population is diluted by the increase in the younger population. Finally, concentration is caused by some older adults and many younger individuals moving out of the area, which results in an overall older population; the decrease in the number of older adults may also be related to high mortality (Helminen et al., 2017).

Two variables were also used to characterise the study grids geographically. Distance from the nearest train station was calculated as the distance from the nearest train station to the centroid of the grid in kilometres (km) using the Euclidean distance. Distance from the JR Yamanote Line was calculated as the distance from the nearest station on the JR Yamanote Line to the centroid of the grid, and was also expressed in km using the Euclidean distance. The JR Yamanote Line was used as the reference point for defining the city centre in this study (Fig. 1). This train line forms a circular route that runs through many of Tokyo’s major business and entertainment districts, such as Shinjuku, Harajuku, Shibuya, Shinagawa, Tokyo, Akihabara, Ueno, and Ikebukuro. Some of these stations are also transfer hubs between the city centre and the suburbs (PwC Japan Group, 2016).

Using the distance from the JR Yamanote Line, we classified the urban landscape of Tokyo into three types: city centre, suburbs, and mountainous areas. The city centre was defined as the grids within 3 km of the JR Yamanote Line (Fig. 1), including the grids inside and outside the circular route. The suburbs are the grids between 3 and 40 km from the JR Yamanote Line (Fig. 1). They include 13 of the 23 special wards of Tokyo and municipalities in west Tokyo, where most of the residential districts are located. About 40% of the residents of the suburbs commute to the city centre (Tada, 2020), which can be reached by train in less than 1 h. The mountainous areas were in grids located more than 40 km from the JR Yamanote Line. The mountainous areas are serviced by local trains, but they run at infrequent intervals. Commuting to the city centre takes at least 1 h by train.

Data Analysis

We conducted our analysis in two steps. First, we used the population ageing matrix proposed by Atkins and Tonts (2016). For each 500-m grid, we evaluated the change in the number of older adults in the population and the changes in the older adult proportion of the population during each 5-year period from 2000 to 2015. Then, each grid was classified into four types: accumulation, decline, dilution, and concentration. To visualise the distribution of these classifications over time, a population ageing matrix scatterplot was generated for each 5-year period from 2000 to 2015. The spatial distribution of the types of ageing was mapped using ArcGIS Pro version 2.7. Second, to determine the geographic structure of the different types of population ageing in relation to distance from the JR Yamanote Line and distance from the nearest train station, a restricted cubic spline logistic regression was employed at seven knots (3, 10, 20, 30, 50, and 60) and five knots (0.5, 1, 2, 4, and 8), respectively. The analyses were carried out using Stata version 14.0.

Results

Spatial Patterns and Temporal Trends of Numerical and Structural Ageing

Table 1 shows that the median number of individuals in the older population in the 500-m grids increased steadily from 115 to 2000 to 253 in 2015. The interquartile range (IQR) also widened over time, implying the presence of a very low or very high number of individuals in the older population in some grids in recent years. The median older adult proportion of the population increased from 5.8 to 10.5% during the period between 2000 and 2015.

Table 1 Median values of the number of proportion of the older adult population (aged 75 and over) in the study grids in Tokyo during 2000–2015
Full size table

Figure 2 shows the median, IQR, and minimum and maximum non-outlier thresholds (1.5 × IQR) of the number (Fig. 2a) and proportion (Fig. 2b) of the older adult population in the 500-m grids by distance from the JR Yamanote Line. The median number of individuals in the older population in the 500-m grids was lower in grids located far from the JR Yamanote Line. The median older adult proportion of the population was slightly lower at greater distances within 40 km from the JR Yamanote Line, while it was higher at greater distances (i.e., between 40 and 70 km) from the JR Yamanote Line in 2000, 2005, and 2010. In 2015, the median older adult proportion of the population was almost the same in grids within 40 km from the JR Yamanote Line, but was higher at greater distances.

Fig. 2
figure 2

Number (a) and proportion (b) of the older adult population (aged 75 and over) in the study grids by distance from the JR Yamanote Line in Tokyo

Full size image

Figure 3 shows the median, IQR, and minimum and maximum non-outlier thresholds (1.5 × IQR) of the number (Fig. 3a) and proportion (Fig. 3b) of the older adult population in the 500-m grids by distance from the nearest train station. The median number of individuals in the older population was lower in grids far from the nearest train station. Between 2000 and 2010, the median older adult proportion of the population decreased in grids within 4 km of the nearest train station and increased in grids located between 4 and 16 km from the nearest train station. In 2015, the median older adult proportion of the population was almost the same in grids within 4 km and higher in grids located further away.

Fig. 3
figure 3

Number (a) and proportion (b) of the older adult population (aged 75 and over) in the study grids by distance from the nearest train station in Tokyo

Full size image

Types of Population Ageing in Tokyo based on a Population Ageing Matrix

Figure 4 shows the distribution matrix of different types of ageing from 2000 to 2015. Accumulation grids were the most prevalent in Tokyo, although the proportion of grids classified as accumulation consistently decreased between 2000 and 2015. Interestingly, decline grids, which experienced a decrease in both the number and proportion of the older adult population, were the second most common type. Decline grids increased steadily over time. Dilution occurred in about 6% of the grids in 2000 and slightly increased in prevalence until 2015. Concentration grids were the least common. The number of concentration grids decreased between 2000 and 2010, but then increased sharply in 2015.

Fig. 4
figure 4

Population ageing matrices for Tokyo for (a) 2000–2005 (n = 4,932), (b) 2005–2010 (n = 5,171), and (c) 2010–2015 (n = 5,140)

Full size image

Over the three time periods from 2000 to 2015, most grids (63%) had the same type of ageing, while the remaining grids changed in at least one period. Table 2 shows the transition of the types of ageing between the first (2000 − 2005) and last (2010 − 2015) periods. Nearly 70% of the grids exhibited persistent accumulation, mostly located between 3 − 30 km from the JR Yamanote Line. Some accumulation grids within 10 km of the JR Yamanote Line changed to either dilution grids (5.4%) or decline grids (4.7%). Some dilution grids within 10 km and 20 − 30 km changed to accumulation grids (4.2%).

Table 2 Transition matrix of types of population ageing among grids in Tokyo between the first (2000 − 2005) and last (2010 − 2015) time periods studied
Full size table

Spatial Patterns and Temporal Trends of Types of Population Ageing

Figure 5 shows the spatial patterns of the different types of ageing from 2000 to 2015. Accumulation grids were widespread over Tokyo. The bulk of these grids were in the east of Tokyo, including the city centre and its nearby areas. Clusters of dilution grids were found in the east of Tokyo, and their prevalence increased over time within the city centre. In certain areas of the city centre, dilution grids formed clusters together with decline grids. In the west of Tokyo, the number of decline grids increased over time, often co-occurring with concentration grids. Concentration grids were largely located in west Tokyo. Additional data concerning the distributions of these ageing types within each municipality/special ward are provided in Online Resource 1. The special wards in the city centre and municipalities in the far west of Tokyo generally exhibited heterogenous types of ageing.

Fig. 5
figure 5

Map of the spatial distribution of the type of population ageing in Tokyo during (a) 2000–2005, (b) 2005–2010, (c) 2010–2015

Full size image

Spatial Patterns and Temporal Trends of the Types of Population Ageing in Relation to Distance from the JR Yamanote Line

Accumulation grids were prevalent within 40 km of the JR Yamanote Line. They peaked between 10 and 40 km from the JR Yamanote Line (Fig. 6a). The number of accumulation grids decreased over time within 10 km and at more than 40 km from the JR Yamanote Line, but remained almost the same between 10 and 40 km. Decline grids were mostly located more than 40 km from the JR Yamanote Line, but some clusters of grids were found within 10 km of the JR Yamanote Line. The number of decline grids increased over time at both distances from the JR Yamanote Line (Fig. 6b). Dilution grids were less prevalent with distance within 10 km of the JR Yamanote Line, but the number increased over time (Fig. 6c). Concentration grids were prevalent between 40 and 60 km from the JR Yamanote Line but they decreased over time (Fig. 6d).

Fig. 6
figure 6

Restricted cubic spline logistic regression plot with seven knots for different types of population ageing: (a) accumulation, (b) concentration, (c) decline, and (d) dilution, by distance from the JR Yamanote Line in Tokyo during 2000–2015

Full size image

Spatial Patterns and Temporal Trends of the Types of Population Ageing in Relation to Distance from the Nearest Train Station

Figure 7 shows the spatial distribution of the types of ageing according to distance from the nearest train station. In general, the spatial pattern of each type of ageing in relation to distance from the nearest train station was consistent over time. Accumulation grids were more prevalent within 1 km of the nearest train station, but the number decreased over time (Fig. 7a). Decline grids and concentration grids were more prevalent at greater distances from the nearest train station, with the bulk of them located more than 8 km away. Although the number of decline grids increased over time, the number of concentration grids decreased (Fig. 7b and d). Dilution grids occurred within 1 km of the nearest train station and their prevalence increased over time (Fig. 7c).

Fig. 7
figure 7

Restricted cubic spline logistic regression plot with five knots for different types of population ageing: (a) accumulation, (b) concentration, (c) decline, and (d) dilution, by distance from the nearest train station in Tokyo during 2000–2015

Full size image

Discussion

This study revealed the spatial patterns and temporal trends of numerical ageing, structural ageing, and four types of population ageing in Tokyo according to the distance from the city centre and the nearest train station. Three main patterns and trends emerged. First, the number of individuals in the older population was higher in grids closer to the JR Yamanote Line and to the nearest station, while the older adult proportion of the population was higher in the grids closest to and farthest from the JR Yamanote Line and the nearest train station. Second, in relation to distance from the JR Yamanote Line, accumulation grids and dilution grids were more prevalent within 40 km of the JR Yamanote Line. Decline and concentration grids were more prevalent at distances more than 40 km from the JR Yamanote Line, and the number of decline grids increased over time, while the number of concentration grids decreased. Third, regarding the types of ageing in relation to the nearest train station, accumulation and dilution grids were more prevalent within 1 km of the nearest train station, and the number of accumulation grids decreased over time while the number of dilution grids increased. Decline and concentration grids were more prevalent at greater distances from the nearest train station and their prevalence increased over time.

Spatial Patterns and Temporal Trends of Numerical Ageing, Structural Ageing, and Ageing Population Types in Relation to Distance from the City Centre

Both the number and proportion of the older adult population in Tokyo exhibited a concentric pattern. The number and proportion of the older adult population was highest in the city centre, lower toward the suburbs, and small in number but large in proportion in the mountainous areas. An east-west divide was also evident in both the number and proportion of the older adult population, which was related to Tokyo’s distinct geography. A larger number of individuals in the older population, but a smaller overall proportion, lived in the eastern part of Tokyo, where there is a concentration of businesses and residential areas. The western most part of the city is composed of several mountains with wild boars roaming freely around the area; hence, it is sparsely inhabited by the older population, although they were highly concentrated in terms of the proportion of the overall population.

Both the number and proportion of the older adult population exhibited a city centre-suburb divide. The number of individuals in the older population in Tokyo was higher in the city centre than in the suburbs. This spatial pattern is similar to that observed in Perth, Australia (Atkins & Tonts, 2016), and in regional cities in Japan, such as Sapporo, Sendai, Hiroshima, and Fukuoka (Feng, 2015). The difference in the numbers of individuals in the older population between the city centre and the adjacent suburbs decreased over time, which may indicate the movement of older members of the population toward the suburbs close to the city centre (Kikuchi & Sugai, 2018). The older adult proportion of the population was higher in the city centre and decreased toward the suburbs. The decreasing older adult proportion of the population from the city centre to the suburbs was consistent with the results of Naganuma et al. (2006) in 2000 and Wakabayashi and Koizumi (2018) in 2005. The general pattern of a higher older adult proportion of the population living in the city centre than in the suburbs was also similar to the pattern observed in Beijing, China, but it contrasted with cities in Europe, such as Milan, Madrid, Berlin, Brussels, Dublin, Amsterdam, London, Lisbon, Copenhagen, and Paris (Arup et al., 2015). Considering the metropolitan areas in OECD member countries, the older adult proportion of the population living in metropolitan areas in Japan, Korea, Mexico, and the United Kingdom was smaller in the urban core (densely inhabited areas) than in the hinterlands (defined as the worker catchment area outside the urban core) in 2001. This contrasted with the results of this study, which may imply that Tokyo’s spatial pattern may also be different from other metropolitan areas in Japan. Notably, the older adult proportion of the population in the city centre and the suburbs was almost the same in 2015. This may indicate a movement of the young population toward the city centre and suburbs, or a movement of the older population to the suburbs.

The distinct spatial patterns in the different types of ageing were related to the geography of Tokyo. Accumulation was prevalent, but it decreased over time in the city centre along with an increase in dilution and decline. Accumulation remained high in the suburbs and this trend did not change substantially over the 15-year study period. Decline and concentration mainly occurred in the mountainous areas. These patterns and trends imply that ageing in Tokyo is more diverse and dynamic in the city centre than in the suburbs and mountainous areas. In comparison with Atkins and Tonts (2016), the spatial patterns of the types of ageing in Perth, Australia, were quite different from the spatial patterns in Tokyo. In general, different types of ageing occurred in the suburbs of Perth compared to the city centre. Accumulation, concentration, and dilution occurred in the suburbs, while decline occurred in the city centre. The older population in Perth increased over time in the suburbs where there have been recent subdivision developments and the construction of a commuter railway (Atkins & Tonts, 2016).

The ‘Back to the City Centre’ phenomenon had a major impact on the structure of ageing in Tokyo after the bubble economy ended in the 1990s. During its economic boom, the Tokyo Metropolis experienced a substantial increase in land prices in the city centre (PwC Japan Group, 2016). New towns were then developed in the western suburbs of the Tokyo Metropolis to provide reasonable housing (OECD, 2015). This development triggered an outflow of people to the suburbs, including large cohorts of young families; the outflow led to an increase in the proportion of older people in the city centre. When the bubble economy ended in the 1990s, land prices began to decline in Tokyo’s city centre, which resulted in changes in land use (Doteuchi, 2003; Tsubomoto & Uozumi, 2012). To revive the city centre, a massive redevelopment of the business, shopping, and entertainment districts occurred throughout the 2000s, along with the construction of condominiums and public housing complexes, as part of the ‘Urban Renewal’ program (Doteuchi, 2003; PwC Japan Group, 2016; Yabe, 2018). This program encouraged both older adult and young populations to reside in the city centre because of the convenient lifestyle (Kikuchi & Sugai, 2018; Yui et al., 2017). The resulting population shift is reflected in the increasing occurrence of dilution in the city centre along the JR Yamanote Line. In some areas of the city centre, changes in land use subsequently displaced the residents, including the older population, thus contributing to decline (Tsubomoto & Uozumi, 2012). The ‘Back to the City Centre’ phenomenon also affected the attitude of the young population; notably, it attracted young families toward city-centre living. More young workers, especially professional workers, chose to live in the 23 wards of Tokyo rather than move to the outer suburbs (Yabe, 2018). Young families with fewer children also remained and raised their children in the city centre rather than buying a house and settling in the suburbs, as their parents had tended to do (Yabe, 2018; Yui et al., 2017). These changes in attitude led to population shrinkage in some areas outside the city centre (Yui et al., 2017). The older population chose to age in place while the younger population moved out of the suburbs, leading to accumulation in the suburbs (Naganuma et al., 2006; Yui et al., 2017). Because the internal migration mobility of the older population in Japan is generally low, the size of the ageing population in the suburbs will likely increase in the future.

In the mountainous areas in the westernmost part of Tokyo, depopulation has been occurring since the 1980s because of the decline in the local timber market (Nishino, 2007). The older adults who remain have chosen to age in place. The resulting trend is reflected in the concentration and decline observed in this area. This indicates that the older population, although it was small, lived in disadvantaged areas, making them more vulnerable. There is a need to ensure that the older population in these areas are not ‘left behind’ in terms of living conditions, health, and other services.

Spatial Patterns and Temporal Trends of Numerical Ageing, Structural Ageing, and Types of Population Ageing in Relation to Distance from the Nearest Train Station

The results indicated that areas with a high number of older adults in the population, and areas with relatively small proportion of the older adult population, were both located within a walkable distance of the train station. These findings are consistent with the World Health Organization (2015) recommendation that the older population should have access to public transportation within a walkable distance of 500 m from home.

Together, the trends in the number and proportion of the older adult population showed that although the number of individuals in the older population increased over time in areas within 4 km of the train station, the older adult proportion of the population increased in areas far from the station (> 2 km). This may indicate that a younger population is moving into areas near to train stations, while the older population is ageing in place within areas far from the stations. Ito et al. (2011) showed that, compared with France and Germany, the older population in regional cities in Japan tended to live in areas where the frequency of train services was low and in areas far from train stations, where transportation was difficult and inconvenient to access. According to Ariga and Matsuhashi (2016), the majority of the older population in Tokyo Metropolitan Area in 2010 had no adequate train access, meaning they were living more than 1 km from a train station.

As discussed earlier, there was a redevelopment program in Tokyo city centre in the 2000s, mostly near to the train stations, which has had an impact on the structure of ageing. The movement of the young and older population into condominiums and housing complexes constructed near train stations reflects the occurrence of accumulation and dilution within 1 km of train stations. On the other hand, the ageing in place of the older population in the suburbs and mountainous areas, where train density is lower, reflects the high occurrence of concentration and decline more than 8 km from the nearest train station.

Policy Implications

Our findings have several policy implications. Because changes in the ageing population will likely alter local needs across the city, the locations and supplies of urban infrastructure and the levels of public service provision must be adjusted (OECD, 2015). For example, different ageing types exist in the city centre; therefore, these areas will likely experience increased demands for services and infrastructure related to both young and old populations. Local governments must determine how to maintain good levels of various services and infrastructure to meet the needs of both populations (OECD, 2015). In suburban municipalities, ageing accumulation occurred within most areas. Suburban spatial structures are car-dependent because of low train station density, the presence of detached housing and medium-rise residential housing on steep terrain, and the placement of services far from communities (Doteuchi, 2003; Yui et al., 2017). As the numbers of older people increase in these areas, there is a need for local governments and the private sector to improve transportation, provide age-friendly housing, and develop compact city strategies (OECD, 2015). In mountainous areas, as the numbers of older adults decrease and their geographic distribution becomes sparse, local governments will experience increasing costs in terms of service provision. Cooperation among neighbouring municipalities, the private sector, and local communities is needed to jointly provide services and reduce costs, thereby ensuring that no one is ‘left behind’ (OECD, 2022).

In Japan, a ‘Community-based Integrated System’ was implemented to provide integrated and comprehensive care services at the community level; this system facilitates ageing in place with cooperation from local residents (Tsutsui, 2014). The identification of different ageing types at the neighbourhood level may guide local governments in the efficient implementation of this system based on local needs. For example, in municipalities with high ageing accumulation and concentration, there may be small numbers of local young volunteers to care for the older population. Local governments may need to provide strong support to older people who care for other older people; they may also need to ensure strong cooperation from the private sector and non-profit organisations (Japan International Cooperation Agency (Japan International Cooperation Agency [JICA], 2022; Tsutsui, 2014). In areas with broader ageing distributions, an integrated care system that caters to all generations may be necessary to provide sufficient support to the community (JICA, 2022).

Limitations

This study had some limitations. First, some grids had inconsistent population data, such as a higher older population than the total population, and were therefore excluded from the study. However, this only represented 0.5% of the total number of grids. Second, because we did not have information for the exact location of the older adult population in each grid, we used the centroid of each grid to denote their location. We used a fine resolution 500-m grid as the unit of analysis, and for the purpose of our study this level of accuracy was considered sufficient. Third, we did not conduct a scale sensitivity analysis although our spatial analysis was sensitive to the scale of aggregation. However, Shiode et al. (2014) also observed that a 500-m grid was adequate to provide information on urban ageing. In future studies, a sensitivity analysis should be conducted to determine whether the results are consistent at different spatial scales.

Conclusion

Investigation of the spatial and temporal distributions of the older population in urban areas is essential for understanding the dynamics of urban ageing, which has implications for resource allocation, service provision, and infrastructure needs. We found that the spatial structure of ageing in Tokyo is related to geography and land use in the city. The older adult population in Tokyo is becoming diluted in the city centre, accumulating in the suburbs, and shrinking in mountainous areas. If these trends continue in the future, the main challenge will involve identifying methods that can meet the increasing demand for age-related facilities and services such as health care, transportation, and housing in the low-density suburbs and in remote mountainous areas, while ensuring that these services and infrastructure are preserved in the city centre. There is a need to optimise the allocation of urban resources to meet the changing needs of the older population.