Can the Relationship Population Contribute to Sustainable Rural Development? A Comparative Study of Out-Migrated Family Support in Depopulated Areas of Japan

Abstract

This comparative study examines how geographic isolation and out-migrant motivations shape support systems in two aging, depopulated rural areas of Japan: Shimogo Town and Nanmoku Village. Challenging the prevailing policies’ focus on internal migration as the primary solution for regional revitalization, we highlight the “relationship population”—a specific group of out-migrated family members who maintain ties to their hometowns—and their diverse contributions to rural sustainability. We employed a mixed-methods approach, including quantitative analyses of aging-oriented household survey questionnaires (using multiple correspondence analysis, multinomial logistic regression, k-means, and two-step cluster analysis) and qualitative analyses of interviews with local government officials and residents (case studies in Nanmoku). Our analysis reveals contrasting support patterns: Shimogo exhibits a localized system driven by resident needs, while Nanmoku employs a strategic approach balancing practical support and community engagement. These findings underscore the limitations of one-size-fits-all migration policies and indicate the need for strategies tailored to the local characteristics of each community. By recognizing the diverse motivations behind hometown visits and the contributions to both residents and communities from the “relationship population”, this research advocates shifting the focus from promoting migration to the interplay of family ties, local support systems, and the agency of residents and out-migrated families. This perspective offers actionable insights for policymakers, local leaders, and researchers working on rural revitalization.

1. Introduction

Rural decline is a global phenomenon linked to industrialization and urbanization and varies significantly across regional and local contexts [1,2]. The specific characteristics and severity of this decline vary greatly, especially in geographically remote peripheral rural areas [3] that face compounded challenges concerning their resource-dependent economies and limited infrastructure [4]. This issue, shaped by forces such as globalization, urbanization, and economic restructuring [5,6], extends beyond individual nations and reflects a global trend impacting various regions, including Europe, North America, and Asia [5,6,7,8]. In Southern Europe, rural depopulation is intensified by economic recession, austerity policies, and youth emigration [7], while in North America, rural areas face challenges such as agricultural restructuring and a decline in local industries. Similarly, in parts of Africa, rural areas face persistent challenges such as poverty, environmental degradation, and the out-migration of younger generations [5]. In China, rural hollowing, excessive land conversion, and inadequate infrastructure contribute to imbalanced urban–rural growth, exacerbating socioeconomic disparities and their related consequences [6]. Although rural–urban migration is often seen as problematic, it also facilitates resource flows and adaptation in affected communities, offering opportunities for redistribution [8]. The phenomenon of rural abandonment, often exemplified by the emergence of “ghost towns”, not only underscores the severe consequences of depopulation but also highlights the untapped potential for cultural-, heritage-, and tourism-based regeneration [9]. However, in regions undergoing significant economic and demographic transitions, such as mountainous rural areas, redistribution processes often fail to address challenges stemming from a limited local capacity, necessitating external support [10]. This issue highlights the critical need for innovative revitalization strategies tailored to the unique constraints of these areas. Thus, while the diverse manifestations of rural decline underline the need for tailored, actionable policies, the shared vulnerabilities of these regions call for strategies specifically targeted to their unique conditions.

The mountainous, depopulated rural areas of Japan provide a striking example of the unique challenges arising from both national and local contexts. As the world’s most rapidly aging society [11], Japan faces acute depopulation in its mountainous rural communities where long-standing family-based mutual aid traditions have been eroded by low fertility rates, rapid aging, and substantial youth out-migration [12,13]. Up until the economic bubble burst in the 1990s, Japan’s rural policies primarily ranged from large-scale infrastructure-focused efforts (“Chiikikaseika” [14,15], regional revitalization) to initiatives promoting in-migration (“Chihousousei” [16,17], regional creation). However, in mountainous areas, these policies often proved inadequate in terms of fully addressing the specific needs of aging, depopulated communities. This inadequacy stemmed from a limited scope, which focused on permanent migration, failing to acknowledge the larger depopulation trends since the 2000s and the consequently increasing need to support existing residents for a sustainable future [14]. This limitation of existing approaches necessitates a shift in perspective, changing from focusing on permanent in-migration to embracing a more flexible, mobile, and dynamic understanding of support systems to be used in supporting depopulated mountainous rural areas.

At the heart of this shift is the concept of the “relationship population”, a term broadly used by the Japanese Ministry of Internal Affairs and Communications (MIC) [18] to describe people with diverse relationships with rural communities that transcend a desire for permanent relocation. In this research, we focus on a specific subset of this “relationship population”: out-migrated family members (or “Tashutsushi”) who maintain active ties to their rural hometowns, primarily through regular visits [19]. Unlike permanent migration strategies, which rely on attracting new residents, the “relationship population” embodies a form of mobility that connects rural and urban areas through periodic hometown visits. These visits often provide essential informal care for elderly residents, addressing gaps left by weakened family-based support due to separate habitation and inadequate formal care services [13,18], but their contributions remain underexplored in both policy-related and academic discussions. This study directly addresses this gap by emphasizing the importance of out-migrants’ attachment and mobility [20] and revealing hometown visits as an often-overlooked but vital component for support. Addressing rural decline requires moving beyond simplistic urban–rural dichotomies [21]. The socioeconomic disparities within urban areas significantly contribute to rural shrinkage and the separation of family structures [22,23], leading to the exclusion of non-co-resident members from being considered part of the family unit. To move beyond this limited, co-residence-based viewpoint and to truly understand rural revitalization, it is necessary to acknowledge the diverse motivations for maintaining ties with rural areas [24,25,26], including non-economic factors such as lifestyle preferences, family obligations, and a desire to maintain cultural traditions [27,28,29,30] that influence support networks beyond co-residency.

Thus, this study positions the out-migrated family as the “relationship population” at the center of the rural sustainability discussion, shifting the focus from demographic decline to the interplay of family ties, local support systems, and mobility. This approach is intended to help us analyze how the relationship population’s specific motivations and patterns of hometown visits foster dynamic support systems anchored in community and family ties. We address these complex dynamics through three key research questions: (1) How do out-migrated family members provide informal care and contribute to rural resilience, and how do these contributions vary across different geographic contexts? (2) How do out-migration patterns (i.e., migration to metropolitan areas vs. local towns) shape familial and community support? (3) What policy implications emerge from these findings for rural revitalization strategies, particularly concerning future migration back to one’s hometown? These questions will guide this research, providing crucial insights for policy development and giving rise to more targeted approaches to integrating the “relationship population” into revitalization efforts.

2. Theoretical Frameworks

2.1. Overview of the Conceptual Frameworks

To effectively understand how to support sustainable aging amidst the challenges faced in Japan’s depopulated mountainous rural areas—where life is shaped by geographic isolation, dispersed settlements, infrastructure limitations [12], and reinforced cycles of economic and regional decline [31,32]—this study revisits and builds upon existing theoretical frameworks to provide new insights. This section integrates global and Japan-specific frameworks, as also summarized in

Table 1. Comparative theoretical frameworks for supporting depopulated rural areas of Japan.

Feature	Counter-Urbanization and “Denenkaiki”	UIJ-Turn Migration	“Relationship Population”
Definition	The movement of populations from urban to rural areas, driven by lifestyle preferences, environmental concerns, and a desire for community-based living [24,26].	A Japanese model categorizing internal migration: U-turn (return to hometown), I-turn (regional in-migration), and J-turn (migration to a nearby hometown) [33,34].	Defined as individuals maintaining active ties with rural areas despite residing elsewhere, particularly in regard to out-migrated families in this study, as well as those who previously resided in a given area or have prior residence or professional connections [18].
Characteristic	Voluntary direction of rural in-migration.	Policy-driven initiatives intended to address the issue of excessive population concentration in Tokyo.	A complement to the migration approach, highlighting non-residents.
Advantages	It highlights the appeal of rural lifestyles, e.g., self-sufficiency, a slower pace of life, and natural surroundings [30,33].	It provides actionable strategies for internal migration; U-turn migration emphasizes the role of “Chien” ties (place attachment) in fostering long-term settlement and integration [20,35].	It focuses on dynamic mobility and contributions beyond residency, such as informal care, cultural participation, and community engagement [18].
Understanding in Japan	It has been incorporated into early rural revitalization efforts (e.g., “Denenkaiki”), promoting the appeal of rural living to attract urban populations [36].	It has been widely adopted in Japan’s “Chihousousei” policies to mitigate urban over-concentration and revitalize regional areas [15,16].	It is emerging as a critical concept in addressing aging and depopulation in mountainous areas, emphasizing the importance of non-resident engagement [18].
Implications in sustaining depopulated rural areas	It focuses on idyllic rural living but lacks alignment with the realities of aging, isolation, and limited infrastructure in Japan’s mountainous rural area [20,37].	It is dependent on attracting migrants, which is increasingly challenging given the decline in the national population and zero-sum competition among municipalities [38,39].	It requires further exploration to quantify long-term impacts and integrate them into policies targeting depopulated mountainous areas.

Table 1 provides a comparative analysis of three conceptual frameworks used for understanding rural population dynamics in Japan. The definitions, characteristics, advantages, and limitations were synthesized by the authors based on existing research, as cited in the text. While counter-urbanization and UIJ-turn migration offer valuable perspectives on migration patterns, their original designs reflect assumptions not fully aligned with the conditions of Japan’s aging, depopulated mountainous areas. In contrast, the “relationship population” framework offers an alternative lens for understanding rural sustainability, emphasizing mobility, dynamic changes, and actionable patterns of support rather than permanent migration alone.

, ultimately positioning the “relationship population” concept as a critical lens for understanding and sustaining rural support systems (Figure 1) while also exploring the potential of encouraging post-retirement migration through mobility-based hometown visits.

2.2. Counter-Urbanization and “Denenkaiki”

Counter-urbanization refers to the movement of populations from urban to rural areas, as we outlined in Table 1, driven by lifestyle preferences, environmental concerns, and a desire for community-based living [24,26,27]. Existing studies, covering topics such as counter-urbanization [23], “back-to-the-land” movements [28], and lifestyle migration [29], largely focus on permanent relocation driven by lifestyle preferences. Even the Japanese concept of “Denenkaiki” (a return to idyllic rural life) [30] overlooks the dynamic, temporary mobility of the relationship population and its specific support contributions. This phenomenon has been extensively studied in Western contexts, particularly in Europe and North America, although the issue of accessibility—wherein rural areas often present an option for attaining improved living conditions and more favorable housing compared to overcrowded urban centers—is also relevant in this context [40,41]. Japan, influenced by global trends, incorporated counter-urbanization concepts into its early rural revitalization efforts, such as “Denenkaiki” (a return to idyllic rural life), which highlights the appeal of self-sufficiency, natural surroundings, and slower-pace lifestyles [30,33,36,37]. The dominant counter-urbanization narrative simplifies this process as a straightforward return to rural living, overlooking the complex interplay between rural mobility and local attachment [42,43,44,45]. However, the inherent Japanese concept of “Chien” [46], rooted in early cultural and social practices, reflects an understanding that maintaining ties with one’s hometown remains important even after outward migration. While the recognition of this concept allowed for emphasis to be placed on the value of relationships, it did not initially evolve into a broader framework for understanding how such ties could actively support and sustain rural communities.

2.3. The UIJ-Turn Migration

While counter-urbanization offers a valuable perspective on urban-to-rural migration, the UIJ-turn migration framework (summarized in Table 1), driven by a strong desire to achieve visible outcomes in boosting rural populations, pinned a great deal of hope on addressing regional demographic imbalances. In Japan, while population decline at the national level began in the mid-2000s, many regional areas, particularly rural ones, had already been experiencing depopulation since the late 1990s [47]. By the 2010s, the structural changes brought by population decline—such as widespread rural depopulation and an excessive concentration of people in the Tokyo metropolitan area—started to affect Japanese society as a whole. In response to these challenges, the government introduced the “Chihousousei” (Regional Creation) policy in 2014, which focused on promoting internal migration as a potential solution [15,16]. Japan developed the UIJ-turn migration framework to address urban–rural population flows. This model categorizes migration into three distinct types: (1) U-turn, that is, returning to one’s rural hometown after living in an urban area; (2) I-turn, i.e., moving from an urban area to a rural or urban one in which one has no prior ties; and (3) J-turn, that is, relocating to a local town nearby one’s rural hometown [33,34] (Table 1). This framework, refined through policy implementation and analysis, has demonstrated that U-turn migration—with its foundation in pre-existing local connections—is generally more stable and sustainable than I-turn migration, which lacks such relational ties [20,35], underscoring the significance of place attachment in fostering long-term settlement in rural communities.

Although UIJ-turn migration has shaped many local government initiatives, its focus on permanent relocation reveals a hesitance to fully address Japan’s population decline and changing societal dynamics. However, despite a decade of various “Chihousousei” initiatives implemented by municipalities across Japan, the persistent concentration of Japan’s population in the “Kanto Region” (the Tokyo metropolitan area)—where 35.3% of the national population resides, with Tokyo itself accounting for 32.5%—underscores the limitations of viewing the UIJ-turn framework as a panacea for nationwide depopulation [38,39]. Given Japan’s persistently low fertility rate and the continued population decline [39], municipal efforts to attract migrants have devolved into an ineffective zero-sum game.

2.4. The “Relationship Population” Concept

This is where the concept of the “relationship population” becomes crucial. Though existing approaches acknowledge that rural areas need support mechanisms, this need is often understood through the lens of “formal support only, or a permanent migration of new residents”, neglecting a group that is not physically present in communities. As defined by Japan’s MIC, the concept of the “relationship population” expands our understanding of rural population dynamics by recognizing individuals who maintain meaningful connections with rural areas beyond those connections established via permanent residency [18] (Table 1). However, despite being widely used in the existing policy discourse, the term “relationship population” is too broad and may include various types of relationships that do not necessarily translate to long-term support or engagement. Therefore, in this study, we focus on a specific subset of the relationship population: out-migrated family members who maintain strong ties to their rural hometowns through regular visits [19].

As shown in Table 1 and Figure 1, unlike UIJ-turn approaches, which focus solely on residents physically present in an area or rely on a relocation-driven migration framework, the “relationship population” framework emphasizes dynamic social ties as a complementary perspective. By understanding “U-turn like” hometown visit patterns (as highlighted in Figure 1 with respect to the UIJ-turn migration part), we seek to further explore the full spectrum of family and community ties and their potential for promoting post-retirement return migration. To further highlight the importance of mobility, we selected two contrasting rural areas for analysis: Shimogo, located far from metropolitan centers, and Nanmoku, situated within a metropolitan sphere. Therefore, in this paper, we aim to present those overlooked social forces provoking people to frequently—or less frequently—visit their hometowns, providing various forms of support while primarily residing in metropolitan or local urban areas. The following section will provide a detailed description of these two selected areas, highlighting their geographical contexts, transportation and accessibility, and key population dynamics.

3. Study Area

This comparative study focuses on two contrasting depopulated municipalities: Shimogo Town (Aizu Area, Fukushima Prefecture) and Nanmoku Village (Seimo Area, Gunma Prefecture). These municipalities were chosen for their contrasting demographic, geographic, and migration characteristics, making them ideal for a comparative analysis of hometown visit patterns and support dynamics (

Table 2. A comparison of the basic information for the two study areas: Shimogo and Nanmoku.

Feature	Shimogo Fukushima Prefecture	Nanmoku Gunma Prefecture
Geography Context 1	Depopulated municipality; located in the easternmost part of Aizu Area; bordering Tochigi; dispersed settlements; reliance on local areas.	Depopulated municipality; located within the Tokyo metropolitan sphere; geographically isolated; mountainous terrain; limited transportation. Surrounded by other depopulated municipalities.
Transportation and Accessibility 2	Private railway connections to Tokyo and Aizu-Wakamatsu; 37 km from Shirakawa IC; 35 km from Aizu-Wakamatsu by car.	Approximately 150 km from Tokyo; accessible via Kan-etsu and Joshin-etsu Expressways (over 2.5 h, 135 km toll roads); up to 5 h away when traveling via national highways.
Population Dynamics 3	Latest data based on local office and 2020 National Census
	Population: 4818 Number of households: 2110 Aging rate: 45.5% (8th within Fukushima)	Population: 1427 Number of households: 844 Aging rate: 67.83% (1st within Japan)
	Population (2015): 5800; aging rate: 40.1% Population (2010): 6461; aging rate: 37.1% Population (2005): 7053; aging rate: 34.3%	Population (2015): 1979; aging rate: 60.5% Population (2010): 2423; aging rate: 61.9% Population (2005): 2929; aging rate: 53.4%
Economic Characteristics 4	Traditional agriculture (e.g., rice, buckwheat, flowers, and vegetables), forestry, tourism (e.g., “Ouchi-juku”, “Yunokami Onsen”, etc.), and cultural events	The economy involves small-scale farming, local produce (e.g., flowers and konjac), and links to urban markets; specific local products for urban sales.
Cultural and Social Ties 5	Strong cultural heritage and festivals; external engagement; reliance on familial networks for community cohesion; out-migrant support for local events.	Cultural ties are influenced by urban proximity; mountainous terrain leads to social isolation; out-migration is influenced by regional and metropolitan connections.

¹ and ²: For geographical information, see Figure 2 and Figure 3, and for depopulated area information, see Figure 4. ³ For Shimogo, the population and household numbers are from the Shimogo Town Office, and the aging rate is from the 2020 National Census [49,50]; for Nanmoku, the data are from the Nanmoku village Office (2024) [51], and the past data are from the 2005–2020 National Census [50]. ⁴ and ⁵: Information is from the National Overview of Municipalities, Reiwa 5 Edition, published by the MIC, Local Administration Bureau, Municipal Affairs Division [48].

). Shimogo, located outside the Tokyo metropolitan area, exhibits dual migration patterns: out-migration to both nearby towns and the Tokyo metropolitan area. This region has placed great emphasis on farming and cultural activities, which are significantly supported by family members who return home [48]. Nanmoku, within the Tokyo metropolitan area but geographically isolated, has one of the highest aging rates in Japan and predominantly experiences local out-migration. Figure 2 and Figure 3 illustrate the contrasting geographic contexts and transportation networks of these areas, highlighting these factors’ potential influence on migration and return-visit patterns. To further highlight the shared challenge of depopulation and the local settings’ influences, Figure 4 shows the distribution of depopulated municipalities around both Shimogo and Nanmoku.

As shown in Table 2, Shimogo Town, a government-designated depopulated municipality, is located in the easternmost part of the Aizu region in Fukushima Prefecture, bordering Tochigi Prefecture to the south (Figure 2 and Figure 3). Characterized by dispersed settlements and a reliance on nearby areas for resources, Shimogo is a mountainous area with limited access to vital services, such as healthcare, social services, and transportation (Figure 3). It is accessible by private railway, with connections from Tokyo via the Tobu Railway and from Aizu-Wakamatsu via the Aizu Railway, and it is about 37 km from the Shirakawa IC and 35 km from Aizu-Wakamatsu by car. The town’s economy is centered on traditional agriculture, forestry, and tourism (e.g., Ouchi-juku), and the region is known for local produce and traditional cultural events. There is an increasing trend in which new farmers focus on cultivating flowers to be sold nationwide. In 2020, the population of individuals aged 65 years and over in Shimogo constituted 45.5% of the total population, which is significantly higher than the averages of Fukushima Prefecture (33.8%) [49] and Japan (29.3%) [11]. The town has also experienced a decline of over 45% between 2005 and 2020. But the strong cultural heritage and festivals attract external engagement, and the town is known for its traditional cultural events and festivals.

Nanmoku Village, also a depopulated municipality located in Gunma Prefecture, is situated within the Tokyo metropolitan area but remains isolated in terms of its geographical characteristics (Figure 3 and Figure 4). Surrounded by other depopulated municipalities (Kanna, Ueno, and Shimonita, for which Figure 4 contains more information) within Gunma Prefecture, Nanmoku is characterized by mountainous terrain and limited transportation infrastructure, necessitating a more strategic reliance on both local and regional connections to urban hubs. Located approximately 150 km from Tokyo, Nanmoku is accessible via the Kan-etsu and Joshin-etsu Expressways, requiring over 2.5 h of travel time (including 135 km on toll roads), and travel via national highways can take up to 5 h (Table 2 and Figure 3). Nanmoku’s economy involves small-scale farming and local produce (e.g., flowers and konjac), which are mostly sold in urban centers. Nanmoku has one of the highest aging rates in Japan and experiences predominantly local out-migration [48,51] (Table 2). While it is geographically closer to Tokyo, its mountainous terrain leads to more social isolation and limited accessibility. The out-migrants are more influenced by both regional and metropolitan connections.

Shimogo and Nanmoku were chosen as contrasting cases to explore how different geographic contexts and out-migration patterns influence familial support. While both municipalities are mountainous and depopulated, they differ in terms of their proximity to local urban centers, their transportation infrastructure, and their reliance on different types of support. This contrast allows us to explore the complexities of rural sustainability and the role of the “relationship population” in different settings. Furthermore, it highlights the need for a multi-method approach to capture those differences.

4. Materials and Methods

In this study, we adopted a comparative, mixed-methods approach to investigate the role of out-migrated family members in supporting rural community resilience, building upon findings from our Shimogo study (Phase 1) [52] and extending the investigation to Nanmoku (Phase 2). Figure 5 provides a visual overview of our methodological path, which began with an exploratory phase in Shimogo (Phase 1), followed by a comparative analysis in Nanmoku (Phase 2). This comparison was made to reveal how geographic proximity, migration patterns, and aging populations shape family support dynamics.

4.1. Overview of Research Design

Phase 1—the Shimogo study: In this exploratory phase, we investigated the daily activities of elderly residents (aged 65+) and the support provided by out-migrated family members during hometown visits. Data were collected through surveys distributed to all eligible households and interviews with local government officials. The analysis of these data revealed four distinct types of return visits: obligation-driven, high-support, community-maintenance-focused, and leisure-oriented visits. These findings highlighted the significant influence of geographic proximity on the frequency and types of support provided.

Phase 2—the Nanmoku study: Building upon the Shimogo findings, in this phase, we focused on a geographically isolated area within the Tokyo metropolitan region with a high proportion of elderly residents. Data were collected through surveys (distributed to all households with the support of local officials) and semi-structured interviews with 12 resident families. Family members were categorized as either frequent (Family A) or less frequent (Family B) visitors based on survey responses. Here, we analyzed how this division of labor shaped the provision of caregiving, household support, and community engagement.

4.2. Questionnaire Design

As shown in Figure 5, the questionnaire was designed to capture objective behavioral indicators, rather than subjective satisfaction measures [53], and reduce the burden of aging respondents, focusing on daily activities, hometown visit patterns, and support behaviors. The questionnaire, which is available in Appendix A 

Table A1. The main contents of the survey questionnaire distributed in Nanmoku.

Questionnaire Summary	Resident Households 1	Most (Family A) and Second-Most (Family B) Frequent Hometown Visit Relatives 2
Basic Information 3	1. Residential area (name of settlement)	5. Location of Visit Relatives
	2. Age and Gender of all family members	6. Age
		7. Frequency of Hometown Visit
		8. Relationship (kinship) to Resident Households
Healthcare And Shopping 4	3. Health conditions:
	Select the health conditions that apply to your family (multiple choice).
	3.1. All family members are in good health
	3.2. Need to go to the hospital regularly
	3.3. Uses a home visit helper
	3.4. Uses a day service (bathing, rehabilitation, etc.).
	3.5. Uses home visit medical services
	3.6. Need caregiving.
	4. Shopping (medical care) methods:
	Select the shopping (medical care) conditions that apply to your family (multiple choice)
	4.1. Completed by family without others’ help
	4.2. Requires relatives to go shopping (go to hospital) with family members during their hometown visit
	4.3. Requires nearby neighbors to go shopping (go to hospital) with family members
	4.4. Relatives bring the items to our family
	4.5. Neighbors bring the items to our family
	4.6. Use mobile vending vehicles in communities
	4.7. Use online services from co-ops, Amazon, and others
Purpose of Hometown-Visiting		9. Purpose of Hometown-Visiting
		Select the main purpose of hometown-visiting by relatives that apply to your family (multiple choice).
		9.1. Family/relative gatherings or reunions
		9.2. Supporting household chores
		9.3. Supporting farm work, yard work (gardening),
		9.4. Enjoying outdoor leisure activities (e.g., walking, fishing)
		9.5. Engaging in individual social activities/events
		9.6. Visiting gravesites
Frequency of Daily Shopping		10. Providing Daily Essentials Shopping
Frequency of Agricultural Work		11. Providing Yard Work 12. Providing Farmland Cultivation, including Planting in Pots, Gardening, Home Vegetable Gardening
Frequency of Outdoor Leisure Activities		13. Outdoor Leisure Activities Frequency of engaging in Outdoor Leisure Activities during Hometown Visits, including: Barbecue (BBQ) Foraging for Wild Plants (Mountain Vegetables), Fishing, Hiking, Walking/Strolling, Carpentry Work, Agricultural Work
Frequency of Personal Social Activities		14. Social Interactions and Leisure Activities in the Community during hometown visits, Including: Spending Time with Friends (e.g., Teatime) Engaging in Hobbies Going Out
Frequency of Community Activities		15. Participating on Behalf of Residents with Community Activities, including: Community Lawn Mowing Community Clean-Up (Trash Collection) Local Festival

To ensure the data were consistent and minimize the burden on the respondents, several survey design choices were implemented. ¹ Responses were collected from a household perspective, ² focusing on the most (Family A) and second-most (Family B) frequent visitors when multiple relatives visited. ³ Open-ended questions allowed for the accurate recording of the settlement names and locations of visiting relatives not included in the provided options. ⁴ Household health status was based on the most severe health condition reported among members. All shopping (medical care) methods employed by household members were recorded.

, was designed based on prior research and the relevant literature [52,54], including four types of daily life behaviors that potentially involve participation by both residents and out-migrated family members and are often considered key aspects of the unique characteristics of rural communities. These behaviors included essential activities (daily living, e.g., medical care and shopping), productive activities (e.g., farming and yard work), social and leisure pursuits (e.g., teatime with friends), and community engagement (e.g., community meetings, clean-up, and festivals) (

Table 3. Summary of questionnaire content in Nanmoku.

Category	Questionary in Nanmoku
Demographic information of resident	1. Location in Nanmoku 2. Household composition 3. Health conditions 4. Shopping and medical care methods
Demographic information of out-migrated family	5. Age 6. Relationship with residents 7. Location 8. Hometown visit frequency 9. The main purpose of a hometown visit
Support during visit	10. Shopping 11. Yard work 12. Farm work
Leisure during visit	13. Personal hobbies 14. Outdoor leisure
Community activities	15. Community meetings (e.g., clean-up, festivals, etc.)

The questionnaire captured detailed information on demographics, daily activities, hometown visit patterns, the division of labor among family members, contributions to support (caregiving, household chores, farming assistance, and community participation), and the impact of aging on support needs. Question 2 asks for the age and gender of all household members. Questions 5–15 differentiate between Family A (most frequent visitors) and Family B (second most frequent visitors). See Appendix A Table A1 for further details.

); at the same time, we also considered the specific context of rural communities in Japan. The Shimogo questionnaire focused on resident households, while the Nanmoku questionnaire included questions specifically targeting out-migrated family members (Family A and Family B) to better understand their motivations and support behaviors.

4.3. Data Collection

This study adhered to the ethical guidelines established by the Chiba University Research Ethics Review Committee, focusing on non-interventional research methods. All data collection and handling procedures complied with relevant university regulations, ensuring the privacy of all participants (see Institutional Review Board Statement).

In Shimogo, the survey was administered through the town office, ensuring participant anonymity and eliminating direct researcher contact (Figure 5). In Nanmoku, with village office approval, surveys were distributed directly to households, accompanied by a comprehensive written explanation of the study, the voluntary nature of participation, and measures for privacy protection (Figure 5). In total, we collected 378 completed surveys in Shimogo and 155 in Nanmoku. Basic descriptive statistics regarding the respondents in Nanmoku are presented in the Appendix A in

Table A2. Basic information on resident households.

Survey Questions	Survey Items		Details and Proportion
1. Location of Resident Households by Settlements N = 155	Settlement Name		Quantity	Proportion
	Ōhinata		39	25.2%
	Iwado		24	15.5%
	Ozawa		23	14.8%
	Mukuruma		18	11.6%
	Ōshiozawa		17	11.0%
	Hisawa		9	5.8%
	Ōnita		7	4.5%
	Hazawa		6	3.9%
	Tozawa		6	3.9%
	Hoshio		4	2.6%
	Chihara		1	0.6%
	Not Provided		1	0.6%
2. Household Composition N = 150	Categorizations of Household Composition		Details of Demographic Characteristics	Quantity	Proportion
	Single-Senior Household N = 49 (33.8%)		Gender
			Male	12	75.5%
			Female	37	24.5%
			Age
			Early Senior (65–79)	13	26.5%
			Later Senior (≥80)	36	73.5%
	Multi-Senior Households N = 66 (42.1%)		Same-Generation Senior Household	50	82.0%
			Intergeneration Senior Household	11	18.0%
	Intergenerational Households N = 35 (24.1%)		Two-Generation Household (Early Senior)	20	57.1%
			Two-Generation Household (Later Senior)	12	34.3%
			Two-Generation Household (Junior Only)	3	8.6%
3. Health Conditions 2 N = 146	Categorizations of Health Conditions 2		Options (Multiple-Choice Questions) 1	Quantity	Selection Rate 1
	Healthy Households N = 49 (33.6%)		3.1. All Family Members are In Good Health	49	33.6%
	Self-Reliant Medical Care Households N = 78 (53.4%)		3.2. Require Regular Medical Attention	93	64.0%
	Number of Support-Reliant Households N = 19 (13.0%)	Support-Needed Households N = 10 (6.2%)	3.3. Utilize In-Home Care Services (for basic assistance)	5	2.5%
			3.4. Attend Adult Day Care	11	7.5%
		Care-Needed Households N = 9 (6.8%)	3.5. Receive Home Visit Medical Services	1	0.8%
			3.6. Require Long-Term Care Assistance	8	5.4%
4. Shopping Medical-care Methods 3 N = 147	Categorizations of Shopping Medical-Care Methods 3		Options (Multiple-Choice Questions) 1	Quantity	Selection Rate 1
	Self-help N = 77 (53.0%)		4.1. Shop (Seek Medical Care) Independently with Cohabiting Family	121	82.3%
	Mutual aid N = 12 (8.2%)		4.2. Shop (Seek Medical Care) with Hometown Visiting Relatives	27	18.3%
			4.3. Shop (Seek Medical Care) with Community Residents	15	10.2%
			4.4. Have Shopping Performed on Their Behalf by Hometown Visiting Relatives	18	11.6%
			4.5. Have Shopping Performed on Their Behalf by Community Residents	3	2.0%
	Combined support N = 58 (38.8%)		4.6. Utilize Community Mobile Vending for Shopping	15	9.5%
			4.7. Utilize Home Delivery Services (e.g., Co-op) for Shopping	36	24.5%

Due to voluntary participation, N varies across items. Percentages are based on valid responses. ¹ Selection rates for options in multiple-choice questions 3 and 4 are shown in Table 3. ² Household health conditions are categorized as follows: healthy households (all family members are in good health, and there is no formal long-term care; only option 3.1 should be selected); self-reliant-medical-care households (one or more family members require regular medical attention but not formal long-term care; only option 3.2 should be selected); and support-reliant households (utilizing formal long-term care services; any combination of options 3.3–3.6 can be selected). Support-reliant households are further categorized as support-needed (options 3.3 or 3.4) and care-needed (options 3.5 or 3.6) households. ³ This categorization focuses on the primary method of obtaining daily necessities and healthcare: self-help (option 4.1 only), mutual aid (options 4.2–4.5 alone or in combination, highlighting the reliance on relatives and neighbors), and combined support (all other combinations of options 4.1–4.7), reflecting the complex support networks observed in Nanmoku.

,

Table A3. Basic information on resident households’ out-migrated family (A and B).

Survey Questions	Survey Items		Quantity (Proportion)
			Family A 3		Family B 3
5. Location of Visit Relatives	Wide range division	Prefecture/City/Town	N = 147		N = 118
	Seimou area 1	Total	N = 112	76.2%	N = 75	63.6%
		Tomioka City	32	28.6%	29	38.7%
		Takasaki City	32	28.6%	22	29.3%
		Nanmoku Village	25	22.3%	14	18.7%
		Shimonita Town	12	10.7%	4	5.3%
		Others	11	9.8%	6	8.0%
	Other areas in Gunma Prefecture	Total	N = 17	11.6%	N = 13	11.0%
		Maebashi City	7	41.2%	6	46.2%
		Others	10	58.8%	7	53.8%
	Kanto Region 2 (Kanto Region exclude Gunma Prefecture)	Total	N = 15	10.2%	N = 27	22.9%
		Tokyo	4	26.7%	12	44.5%
		Saitama	4	26.7%	9	33.3%
		Chiba	7	46.6%	1	3.7%
		Kanagawa	-	-	5	18.5%
	Other Regions	Others	N = 3	2.0%	N = 3	2.5%
6. Age	Total		N = 152		N = 121
	Under 50		47	31.0%	46	37.5%
	50~64		61	40.1%	35	29.1%
	65~74		18	11.8%	20	16.7%
	Over 75		26	17.1%	20	16.7%
7. Frequency of Hometown Visit 5	Total		N = 151		N = 119
	More than once a week		50	33.1%	17	14.3%
	More than once a month		64	42.4%	46	38.7%
	More than once a year		37	24.5%	56	47.0%
	Less than once a year (excluded from total)		4	-	3	-
8. Relationship of Hometown-visiting Relatives to Residents	Total		N = 153		N = 123
	Parent		14	9.1%	2	1.6%
	Child		110	71.9%	84	68.3%
	Sibling		24	15.7%	30	24.4%
	Others		5	3.3%	7	5.7%
9. Main Purpose of Hometown-Visiting	Multiple-choice Question			Selection Rate 4		Selection Rate 4
	9.1. Family/relative gatherings or reunions		127	83.0%	104	83.2%
	9.2. Supporting household chores		27	17.6%	20	16.0%
	9.3. Supporting farm work, yard work (gardening, planting in pots)		14	9.2%	11	8.8%
	9.4. Enjoying outdoor leisure activities (e.g., walking, fishing)		25	16.3%	22	17.6%
	9.5. Engaging in personal social activities/events (e.g., teatime with friends, engaging in hobbies)		20	13.1%	10	8.0%
	9.6. Visiting gravesites		94	61.4%	64	51.2%

Due to voluntary participation, N varies across items. Percentages are based on valid responses. ¹ The Seimou area includes Tano District (Kanna Village and Ueno Village), Kanra District (Kanra Town, Shimonita Town, and Nanmoku Village), and the urban area (Takasaki City, Tomioka City, Fujioka City, and Yasunaka City). ² The Kanto Region does not have a legally defined boundary. In this study, the Kanto Region refers to the prefectures of Tokyo, Ibaraki, Tochigi, Gunma, Saitama, Chiba, and Kanagawa (Figure 2). ³ Family A corresponds to the most frequent visitor, and Family B corresponds to the second most frequent visitor, also shown in Table A1. ⁴ The selection rates for options in the ninth set of multiple-choice questions are also shown in Table A1. ⁵ Q7 “Less than once a year” sample 4 for Family A and 3 for Family B were excluded from the total.

and

Table A4. Comparison of support from Family A and Family B.

Survey Questions On Questionnaire	Details	Family A Most Frequent		Family B Second-Most Frequent
		Quantity	Proportion	Quantity	Proportion
10. Daily Essentials Shopping	Total	N = 147		N = 118
	More than once a week	21	14.3%	8	6.8%
	More than once a month	16	10.9%	11	9.3%
	Several times a year	21	14.3%	22	18.6%
	Almost never	89	60.5%	77	65.3%
11. Yard Work around house	Total	147		N = 116
	More than once a month	22	15.0%	4	3.4%
	Several times a year	11	7.5%	11	9.5%
	Almost never	114	77.5%	101	87.1%
12. Farm Work	Total	N = 149		N = 117
	More than once a month	16	10.7%	5	4.2%
	Several times a year	19	12.8%	12	10.3%
	Almost never	114	76.5%	100	85.5%
13. Outdoor Leisure Activities	Total	N = 145		N = 115
	More than once a month	19	13.1%	7	6.1%
	Several times a year	67	46.2%	59	51.3%
	Almost never	59	40.7%	49	42.6%
14. Personal Social Activities	Total	N = 148		N = 121
	More than once a week	14	9.4%	7	5.8%
	More than once a month	7	4.7%	8	6.6%
	Several times a year	39	26.4%	21	17.4%
	Almost never	88	59.5%	85	70.2%
15. Community Activities	Total	N = 148		N = 119
	Participate All	19	12.8%	8	6.7%
	Participate Part	30	20.3%	21	17.6%
	Almost never	99	66.9%	90	75.6%

Due to voluntary participation, N varies across items. Percentages are based on valid responses. The analysis of visiting relatives’ activities (questions 10–15) is based on respondents who reported that relatives visit at least once a year (Table A3, Q7); “Less than once a year” samples were excluded.

(for the Shimogo data, see [52]). In Nanmoku, 12 resident families also participated in semi-structured interviews. All data collection and handling procedures complied with relevant university regulations, ensuring the privacy of all participants was maintained. All respondents were informed that returning completed questionnaires constituted consent. Further, all personally identifiable information was removed from the data before analysis. A total of 12 Nanmoku residents participated in the semi-structured interviews, and the cases presented in Section 5.4 were anonymized to protect participant confidentiality. Participation in both surveys and interviews was voluntary. These procedures ensured the ethical treatment of all participants and the integrity of the data collected for this comparative analysis of rural support systems.

4.4. Data Analysis

In this study, we employed a comparative, mixed-methods approach to investigate the role of out-migrated family members in supporting rural community resilience. The data analyses were conducted using IBM SPSS Statistics version 27.0.0.0. Figure 5 provides a visual overview of our methodological path. As shown in

Table 4. Data analysis methodology framework: a sequential and progressive approach.

Stage	Method and Analysis	Rationale for Using the Method	Implications for This Study
1. Data Overview	Questionary design, primarily including categorical and factual frequency questions Table A1	To objectively capture actions: It allowed the use of categorized variables based on behaviors while also minimizing subjective bias and cognitive burden [53].	Provided a baseline understanding of demographics, support needs, and visiting patterns in Shimogo and Nanmoku.
	Descriptive Statistics Table A2, Table A3 and Table A4
2. Dimensionality Reduction	Multiple correspondence analysis (MCA)	To Simplify complex data: MCA identifies underlying relationships between multiple categorical variables [55,56,57,58].	Allowed the generation of numerical scores representing underlying dimensions of resident needs (Shimogo) and out-migrant motivations/behaviors (Nanmoku).
	Table A5 and Table A10 Figure A1 and Figure A3	It allowed us to create interpretable scores for further analysis and provided visual joint category plots of variables.	Enabled the use of these scores in subsequent regression and cluster analyses.
3. Dimension Interpretation	Qualitative labeling of dimensions Table A6 and Table A11	To connect abstract dimensions to social context: Dimensions were interpreted based on variable loadings, the existing literature, and the specific context of the study [59].	Provided a clear understanding of what each dimension represents in terms of resident needs, hometown visit motivations, and behaviors. Facilitated the interpretation of subsequent analyses.
4. Causal Relationship	Multinomial logistic regression [55,58] Table A7 and Table A12	To link latent structures with visit patterns: Multinomial logistic regression was used to examine the relationship between MCA dimension scores and visit frequency.	Identified significant associations between dimensions and the likelihood of more frequent visits in different subgroups based on how each support dimension or support pattern influences real behavioral patterns.
5. Cluster Identification	Cluster analysis [60] K-means: Table A8 and Table A9, Figure 3 Two Step [61]: Table A13, Table A14 and Table A15, Figure 4	To provide simplified profiles: Clustering transformed data from numeric-score-based latent dimensions into clear profiles that represent a specific type of social action.	Identified distinct groups of relatives with shared patterns of engagement and support provision.
6. Qualitative Exploration	Thematic coding of interviews, case studies (Section 5.4)	Objective measurements in daily life: Thematic coding of interviews and case studies provided in-depth understanding of lived experiences and contextual factors influencing support patterns.	Provided in-depth understanding of the lived experiences and contextual factors influencing support patterns. Validated and illustrated the findings from the quantitative analyses.

Shimogo data: For Shimogo, the questionnaire design, descriptive statistics, MCA, regression, and cluster analysis were based on a previously published study [52]. Nanmoku data: For Nanmoku, details on the design of the questionnaire, descriptive statistics, MCA, regression, and cluster analysis are given in tables and figures in the Appendix A.

, to address the limitations imposed by using only categorical data and the lack of numerical variability, we employed a carefully designed, sequential multi-method approach, where each method was specifically chosen to build upon and address the limitations of the previous one.

Detailed information on the data and analyses for Shimogo can be found in our previously published study [52], while all detailed analyses, tables, and figures for Nanmoku are provided in the Appendix A. The key steps of our data analysis, as Table 3 shows, included collecting descriptive statistics, which were used to summarize the characteristics of resident households, out-migrated relatives, and the support behaviors in both areas. Detailed descriptive statistics are provided in Appendix A Table A2, Table A3 and Table A4.

MCA was applied to identify key support dimensions, reflecting residents’ needs in Shimogo and the motivations and engagement behaviors of Families A and B in Nanmoku [55,56,57,58]. The full results of the Nanmoku MCA analysis are provided in the Appendix A in

Table A5. Model summary for MCA of Family A’s activities.

Dimension	Cronbach’s Alpha	Variance Accounted For
		Total (Eigenvalue)	Inertia	Variance Explained (%)
1	0.778	2.845	0.474	61.27%
2	0.532	1.797	0.300	38.73%
Total		4.643	0.774	100.0%
Mean	0.683	2.321	0.387

Eigenvalues represent the amount of variance explained by each dimension. Inertia is proportional to the eigenvalue and represents the total variance explained by each dimension. Cronbach’s alpha is a measure of internal consistency reliability; values above 0.6 indicate acceptable reliability. The mean Cronbach’s alpha is based on the mean eigenvalue.

and

Table A6. Discrimination measures (contributions) for MCA of Family A’s activities.

Variables	Dimension 1 Daily Support	Dimension 2 Community Engagement	Mean
10A. Daily Shopping Support	0.523	0.225	0.473
11A. Yard Work Support	0.577	0.042	0.309
12A. Farm Work Support	0.559	0.279	0.419
13A. Outdoor Leisure Activities	0.474	0.274	0.374
14A. Social Activities	0.324	0.413	0.369
15A. Participation in Community Activities as a Substitute	0.387	0.564	0.476
Age (Supplementary)	0.099	0.023	0.061
Relationship (Supplementary)	0.023	0.008	0.015
Geographic Location (Supplementary)	0.031	0.028	0.030
Active Total	2.845	1.797	2.321

Higher values indicate a stronger association between a variable and a given dimension. Age, relationship with residents, and geographic location were included as supplementary variables.

for Family A and

Table A10. Model summary for MCA of Family B’s activities.

Dimension	Cronbach’s Alpha	Variance Accounted For
		Total (Eigenvalue)	Inertia	Variance Explained (%)
1	0.770	2.793	0.466	56.9%
2	0.633	2.115	0.353	43.1%
Total		4.909	0.818	100.0%
Mean	0.711	2.454	0.409

Eigenvalues represent the amount of variance explained by each dimension. Inertia is proportional to the eigenvalue and represents the total variance explained by each dimension. Cronbach’s alpha is a measure of internal consistency reliability; values above 0.6 indicate acceptable reliability. The mean Cronbach’s alpha is based on the mean eigenvalue.

and

Table A11. Discrimination measures (contributions) for MCA of Family B’s activities.

Variables	Dimension 1 On-Demand Support	Dimension 2 Social Contact	Mean
10B. Daily Shopping Support	0.614	0.405	0.510
11B. Yard Work Support	0.605	0.513	0.559
12B. Farm Work Support	0.499	0.334	0.417
13B. Outdoor Leisure Activities	0.401	0.200	0.301
14B. Social Activities	0.279	0.527	0.403
15B. Participation in Community Activities as a Substitute	0.395	0.135	0.265
Age (Supplementary)	0.058	0.067	0.062
Relationship (Supplementary)	0.022	0.175	0.099
Geographic Location (Supplementary)	0.030	0.136	0.083
Active Total	2.793	2.115	2.454

Higher values indicate a stronger association between a variable and a dimension. Age, relationship with residents, and geographic location were included as supplementary variables.

for Family B, as well as in Figure A1 and Figure A3.

Multinomial logistic regression was used to examine the factors influencing hometown visit frequencies, focusing on the relationships between MCA dimensions and visit patterns for Shimogo families and Nanmoku Families A and B [56,57,58,59]. The detailed results of the regression analysis for Nanmoku are provided in Appendix A

Table A7. Multinomial logistic regression predicting the hometown visit frequency of Family A based on MCA dimensions.

Hometown Visit Frequency	B	Std. Error	Wald	df	p	Exp(B)	95% CI for Exp(B)
More than once a month
Intercept	0.366	0.247	2.189	1	0.139	-	-	-
Dimension 1	−1.001	0.328	23.843	1	0.000	0.202	0.106	0.384
Dimension 2	−0.909	0.271	11.241	1	0.001	0.403	0.237	0.686
More than once a year
Intercept	−1.546	0.753	4.222	1	0.040	-	-	-
Dimension 1	−4.133	1.167	12.548	1	0.000	0.016	0.002	0.158
Dimension 2	−0.838	0.455	3.395	1	0.065	0.433	0.177	1.055

This study highlights the direction and significance of relationships. Model fit significantly outperformed the intercept-only mode (χ² = 96.329, df = 4, p < 0.001). Pseudo R²: Cox and Snell = 0.472, Nagelkerke = 0.534, and McFadden = 0.297. Dimension 1: negatively associated with return frequency; Dimension 2: weaker negative association. Hometown visit frequency is based on respondents who reported at least one annual visit (Question 7, Table A3).

and

Table A12. Multinomial logistic regression predicting the hometown visit frequency of Family B based on MCA dimensions.

Hometown Visit Frequency	B	Std. Error	Wald	df	p	Exp(B)	95% CI for Exp(B)
More than once a month
Intercept	1.619	0.391	17.13	1	0.000	-	-	-
Dimension 1	1.066	0.303	12.424	1	0.000	2.905	1.605	5.255
Dimension 2	−0.287	0.21	1.866	1	0.172	0.75	0.497	1.133
More than once a year
Intercept	−1.546	0.753	4.222	1	0.040	-	-	-
Dimension 1	−4.133	1.167	12.548	1	0.000	0.016	0.002	0.158
Dimension 2	−0.838	0.455	3.395	1	0.065	0.433	0.177	1.055

This study highlights the direction and significance of relationships. Model fit significantly outperformed the intercept-only mode (χ² = 40.092, df = 4, p < 0.001). Pseudo R²:Cox and Snell = 0.286, Nagelkerke = 0.331, and McFadden = 0.168. Dimension 1: higher scores are linked to lower return frequencies, with the direction of association depending on the comparison category. Dimension 2: No significant association. Hometown visit frequency is based on respondents who reported at least one annual visit (Question 7, Table A3).

.

Cluster analysis was used to group relatives based on MCA scores [59], revealing distinct engagement patterns among Shimogo relatives and Nanmoku families (Section 5.3). K-means clustering was applied to Shimogo and Nanmoku Family A, while a two-step clustering approach was used for Nanmoku Family B. For the uneven K-means clustering distributions, we used two-step cluster analysis, which allowed us to analyze both categorical and continuous variables while identifying clusters automatically [60,61,62]. The detailed results of the cluster analysis are provided in Appendix A 

Table A8. Explained variance and case distribution in MCA dimension scores by number of K-means clusters (N = 155).

Number of Clusters	Explained Variance of Dimension		Distribution of Cases
	1	2	Cluster A1		Cluster A2		Cluster A3		Cluster A4
2	0.776	0.024	116	74.8%	39	25.2%
3	0.622	0.498	32	20.6%	34	21.9%	89	57.4%
4	0.780	0.776	27	17.4%	12	7.7%	30	19.4%	86	55.5%

This table shows the explained variance and case distribution for 2, 3, and 4 clusters in Nanmoku Family A based on MCA dimension scores (Table A5). Explained variance reflects the proportion of total variance captured by the clusters, calculated as the ratio of between-cluster variance to total variance. A higher explained variance indicates a greater ability of the cluster solution to capture the variance in the data. While the 4-cluster solution achieves the highest explained variance in Dimension 1, its uneven distribution (with cluster 2 containing only 7.7% of the cases) makes the 3-cluster solution more balanced and interpretable for analyzing behavioral patterns.

and

Table A9. Descriptive statistics of Family A clusters based on MCA dimension scores.

Variables		Percentages in Clusters 1						Chi-Square Tests (p-Value) 2	Characteristic
		Cluster A1		Cluster A2		Cluster A3
Age	<50 years	8	25.8%	3	9.1%	36	40.9%	<0.05	Cluster A3: predominantly younger; Cluster A2: predominantly older; Cluster A1: mix
	≥50 years	23	74.2%	30	90.9%	52	59.1%
Relationship	Parent	5	16.1%	3	9.1%	6	6.7%	-	Predominantly children in all clusters, with higher sibling representation in Cluster A3
	Child	22	71.0%	26	78.8%	62	69.7%
	Sibling	3	9.7%	3	9.1%	18	20.2%
Relatives from Seimou area		24	77.4%	25	86.2%	63	72.4%	-	Seimou-based across all clusters
Hometown visits (weekly)		8	25.0%	29	87.9%	13	15.1%	<0.001	Cluster A2: high frequent visits
Hometown visits (>monthly)		30	93.7%	33	100%	38	44.2%		Cluster A1: frequent visits
Hometown visits (yearly only)		2	6.3%	-	-	25	40.7%		Cluster A3: high rate of visits (both annual and less frequent)
Chore support 3		8	25.0%	12	35.3%	7	7.9%	<0.001	Clusters A1 and A2: low levels of chore support; Cluster A3: lowest level of chore support
Grave visit 3		20	62.5%	21	61.8%	53	59.6%	-	Common across all clusters
Outdoor leisure activities (weekly)		2	6.7%	15	50.0%	2	2.4%	<0.001	Cluster A2: high level of weekly outdoor leisure activities
Outdoor leisure activities 3		26	86.7%	25	83.3%	35	41.2%		Cluster A3: least amount of engagement
Shopping support (>monthly)		11	37.9%	20	58.8%	6	7.2%	<0.001	Cluster A2: frequent shopping support; Cluster 1: lower level
Shopping support (yearly only)		9	31.0%	12	35.3%	68	81.0%		Cluster A3: infrequent shopping support
Yard work support (weekly)		3	11.1%	18	54.5%	-	-	<0.001	Cluster A2: high-level support; Cluster A1: low-level support
Yard work support 3		8	29.6%	24	72.7%	1	1.2%		Cluster A3: infrequent yard work support
Farm work support (weekly)		2	6.7%	14	42.4%	-	-	<0.001	Cluster A2: high-level support; Cluster A1: low-level support
Farm work support 3		16	53.3%	17	51.5%	2	2.3%		Cluster A3: infrequent farm work support
High-frequency social activities (>monthly)		3	10.0%	14	43.8%	4	4.7%	<0.001	Cluster A2: high level of social activity
Social activities 3		23	76.7%	23	71.9%	14	16.3%		Clusters A1 and A2 high social participation; Cluster A3: lower participation
Community activities 3		22	73.3%	18	56.2%	9	10.5%	<0.001	Cluster A1: high engagement; Cluster A2: moderate engagement; Cluster A3: lower engagement
Shopping and medical care support needed by residents 4		17	54.8%	22	78.6%	27	31.0%	<0.001	Cluster A2: higher support needs; Cluster A3: high unmet support needs

¹ Percentages are within-cluster proportions based on valid responses; sample sizes vary across variables. ² p-values are based on chi-square tests and indicate whether a characteristic significantly differs across clusters. ³ Chore support, grave visits, outdoor leisure activities, yard work support, farm work support, social activities, and community activities indicate participation, not frequency. ⁴ Support needed by residents refers to the proportion of households within each cluster requiring external support for shopping or healthcare (Table A2, Q4, referring to the quantities of mutual-aid and combined-support households).

for Family A,

Table A13. The auto-clustering results for Family B obtained via two-step cluster analysis.

Number of Clusters	Schwarz’s Bayesian Criterion (BIC)	BIC Change	Ratio of BIC Changes	Ratio of Distance Measures
1	430.669
2	267.024	−163.645	1.000	2.094
3	203.842	−63.183	0.386	3.526
4	206.468	2.627	−0.016	1.092
5	211.296	4.827	−0.029	1.750
6	226.340	15.044	−0.092	1.623
7	246.615	20.275	−0.124	1.205
8	268.318	21.703	−0.133	1.214
9	291.251	22.933	−0.140	1.286
10	315.461	24.210	−0.148	1.253
11	340.574	25.113	−0.153	1.163
12	366.185	25.612	−0.157	1.298
13	392.500	26.315	−0.161	1.363
14	419.443	26.943	−0.165	1.053
15	446.473	27.030	−0.165	1.025

Silhouette measure of 0.8 for cohesion and separation in a “good” cluster quality. BIC Change: the change in Schwarz’s Bayesian information criterion (BIC) from the previous number of clusters. Ratio of BIC Changes: the ratio of the BIC change for the current number of clusters to the BIC change for the two-cluster solution. Ratio of Distance Measures: The ratio of the distance measures between the current number of clusters and the previous number of clusters. This table shows the results of the auto-clustering phase of the two-step clustering method, where we used categorical visit frequency (Table A3, Q7, N = 119) and MCA dimension 1 and 2 scores (Table A8) as input variables. The three-cluster solution was chosen because the BIC change showed a substantial decrease from 2 to 3 clusters, while further increases in the number of clusters did not lead to a similarly large change, indicating that three clusters captured most of the meaningful patterns in the data.

,

Table A14. Cluster distribution and centroids of Family B’s two-step cluster analysis.

Cluster	N	Percentage of Combined	Object Score Dimension 1 (On-Demand Support)		Object Score Dimension 2 (Social Contact)
			Mean	Std.	Mean	Std.
Cluster B1	17	11.6%	−1.475	1.720	0.328	2.021
Cluster B2	46	38.7%	0.062	0.811	0.029	1.294
Cluster B3	56	47.1%	0.368	0.353	−0.114	0.469
Combined	119	100%	−0.013	1.044	0.004	1.148

Dimension 2 scores are provided for reference but were not found to be significant predictors of visit frequency in the regression analysis (Table A12).

and

Table A15. Descriptive statistics of Family B clusters based on MCA dimension scores.

Variables		Percentages in Clusters 1						Chi-Square Tests (p-Value) 2	Characteristic
		Cluster B1		Cluster B2		Cluster B3
Age	<50 years	10	58.8%	27	65.9%	40	71.4%	-	B3: predominantly younger; Cluster B1 and B2: mix
	≥50 years	7	41.2%	14	34.1%	16	28.6%
Relationship	Parent	1	5.9%	1	2.2%	-	-	-	Predominantly children and a small proportion of siblings in all clusters
	Child	12	70.6%	31	68.9%	39	70.9%
	Sibling	3	17.6%	10	22.0%	14	25.5%
Relatives from Seimou area		15	88.2%	35	81.4%	25	45.5%	<0.001	Cluster B1 and B2: Seimou-based
Relatives from Kanto region		1	5.9%	4	9.3%	19	34.5%		Cluster B3: mix
Hometown visits (weekly)		17	100%	-	-	-	-	<0.001	Cluster B1: highly frequent visits Cluster B2: frequent visits; Cluster B3: infrequent visits
Hometown visits (monthly)		-	-	46	100%	-	-
Hometown visits (yearly)		-	-	-	-	56	100%
Chore support 3		5	29.4%	7	15.2%	7	12.5%	-	Low-level chore support across all clusters
Grave visit 3		6	35.3%	24	52.2%	30	53.6%	-	Low levels in Cluster B1
Outdoor leisure activities (weekly)		4	25.0%	2	4.8%	1	1.9%	<0.05	Cluster B1: high levels of engagement in outdoor leisure activities Cluster B2 and B3: moderate levels of engagement
Outdoor leisure activities 3		12	75.0%	21	50.0%	31	59.6%
Shopping support (>monthly)		9	52.9%	9	22.0%	-	-	<0.001	Cluster B1: frequent shopping support; Cluster B2: lower-level shopping support; Cluster B3: infrequent shopping support
Shopping support (yearly only)		2	11.8%	8	19.5%	11	20.4%
Yard work support (weekly)		3	17.6%	1	2.5%	-	-	<0.001	Cluster B1: moderate-level support; Cluster B2 and B3: infrequent yard work support
Yard work support 3		8	47.1%	4	10.0%	2	3.6%
Farm work support (weekly)		3	17.6%	1	2.4%	1	1.9%	<0.05	Cluster B1: moderate level support; Cluster B2 and B3: infrequent farm work support
Farm work support 3		7	41.2%	3	7.3%	2	11.1%
High-frequency social activities (>monthly)		4	23.5%	3	6.7%	-	-	<0.001	Cluster B1: high level of social activity engagement;
Social activities 3		12	71.6%	12	26.7%	12	21.8%		Clusters B2 and B3: lower engagement
Community activities 3		8	53.7%	10	23.8%	10	17.9%	<0.001	Cluster B1: moderate levels of engagement; Cluster B2 and B3: lower engagement
Shopping and medical care support needed by residents 4		9	46.2%	17	69.0%	19	66.7%	-	Cluster B1: lower support needed; Cluster B2 and B3: higher support needed

¹ Percentages are within-cluster proportions based on valid responses; sample sizes vary across variables. ² p-values are based on chi-square tests and indicate whether a characteristic significantly differs across clusters. ³ Chore support, grave visits, outdoor leisure activities, yard work support, farm work support, social activities, and community activities indicate participation, not frequency. ⁴ Support needed by residents refers to the proportion of households within each cluster requiring external support for shopping or healthcare (Table A2, Q4; the numbers of mutual-aid and combined-support households).

for Family B, and Figure A2 and Figure A4.

The qualitative analysis involved the thematic coding of the interview data from Nanmoku, offering insights into motivations, support dynamics, and intergenerational contributions. Case studies were conducted to contextualize the quantitative findings (Section 5.4). Visualizations support these analyses, including a Sankey diagram [63] (Section 5.2) illustrating the flow of support behaviors to MCA dimensions and maps (Section 5.1 and Section 5.3) highlighting geographic contexts and pathways of engagement.

In this study, we compare Shimogo’s localized support networks with Nanmoku’s more strategic and regionally distributed systems to reveal how differing community needs and out-migrated relatives’ motivations shape rural resilience strategies.

5. Results

5.1. Geographic Context, Out-Migration, and Frequent Visitor Patterns

This section presents a comparative analysis of household structures and support networks in Shimogo and Nanmoku, illustrating how contrasting forms of geographic isolation shape both household demographics and the nature of support provided in these depopulated rural areas. Shimogo, located in the easternmost part of the Aizu region in Fukushima Prefecture, exhibits a form of localized isolation defined by its remoteness and dispersed settlements, leading to a reliance on nearby areas for resources. Nanmoku, situated in the westernmost part of the Seimo region in Gunma Prefecture, exhibits a different form of geographic isolation. Despite being within the Tokyo Metropolitan region, it is characterized by a mountainous terrain and limited transportation infrastructure, necessitating a more strategic reliance on both local and regional connections to urban hubs. These distinct forms of isolation shape the patterns of out-migration and the support networks that emerge in these two areas.

These distinct forms of isolation significantly influence the dynamics of household composition, the age profiles of residents, and the strategies for accessing external support.

5.1.1. Resident Characteristics and Geographic Context

The contrasting forms of geographic isolation influence household structures and support mechanisms, as detailed in

Table 5. Characteristics of resident households in Shimogo and Nanmoku.

Survey Items		Details and Proportion
		Shimogo		Nanmoku
Household composition	Total	N = 376		N = 150
	Single senior	146	38.8%	49	33.8%
	Multi-senior	201	53.5%	66	42.1%
	Intergenerational	29	7.7%	35	24.1%
Age of single-senior Household member	Total	N = 146		N = 49
	65–79	89	60.9%	13	26.5%
	Over 80	55	37.7%	36	73.5%
	Not provided	2	1.4%	-	-
Health conditions 1	Total	N = 357		N = 146
	Healthy	102	28.6%	49	33.6%
	Only outpatient care	207	58.0%	78	53.4%
	Support-reliant	48	13.4%	19	13.0%
Support methods 2 (shopping and medical care)	Total	N = 364		N = 147
	Self-help	233	64.0%	77	52.4%
	Mutual aid	38	10.4%	12	8.2%
	Combined support	93	25.6%	58	39.4%

Due to voluntary participation, N varies across items. Percentages are based on valid responses. ¹ Health conditions: healthy (no formal healthcare needed); outpatient care only (regular medical attention required, with no long-term care); and support–reliant (reliant on the Long-Term Care Insurance system). ² Support methods: self-help (independent); mutual aid (neighbor/family assistance required, excluding community services); and combined support (a mix of independent and external assistance is required, including community services). For the design of the questionnaire, see Appendix A Table A1. Detailed data tables and additional analyses related to these findings are available in our Shimogo study [52] and Appendix A Table A2, Table A3 and Table A4. Due to voluntary participation, N varies across items. Percentages are based on valid responses.

. Shimogo’s localized isolation is reflected in its higher proportion of multi-senior households (53.5%), which are reliant on local support, while Nanmoku’s mountainous isolation and higher aging rate are reflected in a significantly greater reliance on peer support networks and a greater proportion of single-senior household members aged 80 and over (73.5%). Despite these differences, both communities show comparable levels of need for support, indicated by a similar reliance on formal support services (Shimogo: 13.4%; Nanmoku: 13.0%). However, they diverge in terms of their access to support and the ways in which they engage with these systems. Shimogo residents exhibit greater self-reliance (64% self-help), while Nanmoku residents utilize combined support systems more frequently (39.4%), underscoring the influence of geographic accessibility on their respective support strategies.

5.1.2. Out-Migration Patterns and the Geography of Frequent Visitors

The contrasting forms of geographic isolation shape the kinship networks and patterns of hometown visits. Table 5 and

Table 6. Characteristics of relatives in Shimogo and Nanmoku.

Survey Items		Details and Proportion
		Shimogo 1			Nanmoku 1
						Family A		Family B
Location	Total		N = 342			N = 147		N = 118
	Local Area	Aizu Area	171	50.0%	Seimou Area	112	76.2%	75	63.6%
	Main local city	Aizuwakamatsu City	89/171	52.1%	Tomioka City	32/112	28.6%	20/75	26.7%
					Takasaki City	32/112	28.6%	14/75	18.7%
		Others	82/171	47.9%	Others	28/112	42.8%	41/75	54.6%
	Nearby Region	Within Fukushima	52	15.2%	Within Gunma	17	11.6%	66	11.0%
	Tokyo Metropolitan	Kanto Region 2	100	29.2%	Kanto Region 2 Exclude Gunma	15	10.2%	35	22.9%
	Other Region		19	5.6%		3	2.0%	3	2.5%
Age	Total		N = 355			N = 152		N = 121
	Under 50		163	48.7%		47	31.0%	46	37.5%
	50–64		109	32.5%		61	40.1%	35	29.1%
	65–74		48	14.3%		18	11.8%	20	16.7%
	Over 75		15	4.5%		26	17.1%	20	16.7%
Hometown visit frequency	Total		N = 313			N = 151		N = 119
	Weekly		66	21.1%		50	33.1%	17	14.3%
	Monthly		101	32.3%		64	42.4%	46	38.7%
	Yearly		146	46.6%		37	24.5%	56	47.0%
Relationship 3		This question was not present in the Shimogo survey			Total	N = 153		N = 123
					Parent	14	9.1%	2	1.6%
					Child	110	71.9%	84	68.3%
					Sibling	24	15.7%	30	24.4%
					Others	5	3.3%	7	5.7%
Main purpose of hometown visit 3 (multiple choice)		This question was not present in the Shimogo survey			Total	N = 153		N = 125
					Family reunion	127	83.0%	104	83.2%
					Chore support	27	17.6%	20	16.0%
					Farm work	14	9.2%	11	8.8%
					Outdoor leisure	25	16.3%	22	17.6%
					Personal social activities	20	13.1%	10	8.0%
					Grave visits	94	61.4%	64	51.2%

Due to voluntary participation, N varies across items. Percentages are based on valid responses. ¹ For Shimogo, the data represent the family with the most hometown visits; for Nanmoku, Family A and Family B represent the most and second most frequently visited families, respectively. ² The Kantō region includes Chiba, Saitama, Tochigi, Ibaraki, Gunma, Kanagawa, and Tokyo; for Nanmoku, this count excludes Gunma Prefecture. ³ Data on relative relationships and main purposes for hometown visits were only collected from the Nanmoku respondents; percentages reflect selection rates for each multiple-response option.

and Figure 6 demonstrate that in Shimogo, 91.3% of the most frequent visitors (monthly or more) were localized within the Aizu area, particularly Aizuwakamatsu City, illustrating a strong reliance on a localized support network. In contrast, Nanmoku had a more strategic pattern, with frequent visitors (Family A) roughly split between nearby Tomioka (28.6%), a locally accessible city, and Takasaki (28.6%), a major transportation hub with regional connections. Conversely, a greater proportion of less frequent visitors (Family B) from Nanmoku are found in the wider Gunma region (11.0%) or the more distant Tokyo Metropolitan Region (22.9%). Despite these differences in migration choices, both Shimogo and Nanmoku reveal evidence of sustained connections with hometowns through regular visits. Within Nanmoku, while children are the predominant family tie for both Families A (71.9%) and B (68.3%), suggesting their shared primary responsibility for familial support, siblings represent a significantly higher proportion of Family B (24.4%) compared to Family A (15.7%). While Nanmoku is located within the Tokyo Metropolitan area (“Kanto” Region), Figure 6 clearly demonstrates that it is primarily the regional connections within Seimou, not proximity to Tokyo itself, that are most strongly associated with frequent visits. Furthermore, while family reunions were the most reported purpose for visits (approximately 83%), Nanmoku Family A shows significantly higher involvement in practical support activities (chore support, farm work, and grave visits) than Family B, highlighting the multifaceted nature of their sustained engagement in family and community life.

This analysis reveals that geographic isolation—which is localized in Shimogo versus strategically dispersed in Nanmoku—shapes support systems and mobility patterns, a fact that has direct implications for future rural engagements. Shimogo’s localized networks, centered in Aizuwakamatsu, suggest a foundation for future return migration, while Nanmoku’s strategically balanced use of local and regional hubs indicates a more complex pathway for maintaining ties. These findings necessitate context-specific rural revitalization strategies.

5.2. Motivation Shapes Contribution: Analyzing Rural Support in Shimogo and Nanmoku

This section explores how the motivations of out-migrated family members shape their support contributions to Shimogo and Nanmoku. Building on the analysis of geographic isolation (Section 5.1), we used multiple correspondence analysis (MCA) [55,56,57,58,59], multinomial logistic regression (

Table 7. MCA dimensions, top contributions, and associated visit frequencies.

Out-Migrated Family	MCA Dimension 1	Top Contributing Behaviors (Loadings)		Dimension Traits	Influence on Visit Frequency 2
Shimogo Family	Resident Self-Reliance	Farm Work	0.592	This dimension reflects the residents’ independence and capacity for self-sufficiency.	Higher scores are associated with less frequent visits (p < 0.001)
		Yard Work	0.589
		Shopping Support	0.511
	Resident Community Support Needs	Chore Support	0.656	This dimension reflects the residents’ need for community resources in the absence of family support.	Higher scores are associated with less frequent visits (p < 0.001)
		Outdoor Leisure	0.664
Nanmoku Family A	Daily Support	Shopping Support	0.523	This dimension captures Family A’s motivation to provide routine, practical assistance for daily living and household maintenance.	Higher scores are associated with more frequent visits (p < 0.001)
		Yard Work	0.577
		Farm Work	0.559
	Community Engagement (Community Support)	Community Activities	0.564	This dimension captures Family A’s motivation for active community integration and social participation with their family support activities.	Higher scores are associated with more frequent visits (p < 0.001)
		Social Activities	0.413
		Outdoor Leisure	0.274
Nanmoku Family B	On-Demand Support	Shopping Support	0.614	This dimension captures Family B’s motivation to provide routine, practical assistance for daily living and household maintenance.	Higher scores are associated with more frequent visits (p < 0.001)
		Yard Work	0.605
		Farm Work	0.499
	Community Engagement (Social Contact)	Social Activities	0.527	This dimension captures Family B’s motivation for active community integration and social participation with their family support activities.	-
		Outdoor Leisure	0.200

Cronbach’s alpha values above 0.6 are generally considered to indicate acceptable internal consistency reliability for a scale. The overall Cronbach’s alpha values for all models were as follows: Shimogo (0.778) and Nanmoku Family A (0.683), Family B (0.711). ¹ For Shimogo, the MCA dimensions, top contributing behaviors, dimension traits, and the influence on visit frequency are based on a previously published study [52]. The MCA dimensions and associated traits are derived from behaviors identified in Shimogo and Nanmoku. For details on the MCA methodology, analysis results, and model specifications, refer to Appendix A Table A5, Table A6, Table A10 and Table A11 and Figure A1 and Figure A3. ² Influence on visit frequency was determined using multinomial logistic regression. Detailed regression results and model metrics are provided in Appendix A Table A7 and Table A12.

) [58], and a Sankey diagram [63] (Figure 7) to reveal that while Shimogo’s support networks are primarily a reflection of localized self-sufficiency and the expressed needs of its residents, the networks in Nanmoku are driven by the more diverse motivations and individual support preferences of its out-migrated relatives.

Table 7 presents a comparative overview of the MCA dimensions and their relation to visit frequencies, highlighting the distinct forms of support in Shimogo and Nanmoku. In Shimogo, the dimensions reflect resident needs: “resident self-reliance”, characterized by farm work, highlights the residents’ capacity to manage daily life, while “community support needs” underscore their reliance on external resources through shopping support. Furthermore, in Shimogo, increases in both self-reliance and reliance on community support are both associated with a reduction in the frequency of relative visits. In contrast, the dimensions for Nanmoku emphasize the motivations that drive the actions of out-migrated relatives: Family A is motivated to provide “daily support”, demonstrated through shopping support and yard work, and focuses on “community engagement”, exemplified by community activities, while Family B’s support motivations center on providing “on-demand support” through farm work and on “community engagement” through their own social and outdoor leisure activities. The results of the multinomial logistic regression demonstrate that, in Nanmoku, higher scores for Family A’s “daily support” and “community engagement” dimensions are associated with more frequent visits, while Family B’s focus on “on-demand support” is associated with a higher likelihood of less frequent (yearly) visits compared to weekly visits.

Figure 7 illustrates these support pathways by showing how key activities, such as “shopping support”, “farm work”, and “yard work”, contribute to different dimensions in Shimogo and Nanmoku. In Shimogo, these activities are structured around dimensions reflecting residents’ self-reliance or their need for external community support. In Nanmoku, the same activities reveal the different motivations of out-migrants who support their hometowns: the motivation for Family A is more consistent and direct support, and that for Family B is more flexible, need-based engagement.

By applying MCA, multinomial logistic regression, and Sankey visualization, this analysis demonstrates how motivation shapes contributions to rural support in depopulated areas. The dimensions identified in Shimogo are shaped by relatives’ perceptions of their residents’ needs and the need for both self-reliance and external support. In contrast, the dimensions for Nanmoku reveal how out-migrant relatives’ patterns of support are driven by the differing reasons motivating Family A to provide regular and direct support and those motivating Family B to offer more flexible engagement. These complex processes, made visible through a Sankey diagram (Figure 7), highlight that a complex interplay of factors influences how out-migrants both provide and maintain these support systems, emphasizing the need to acknowledge these factors when developing future revitalization strategies.

5.3. The Dynamic Interplay Between Isolation, Motivation, and Support Networks

In this section, we examine the diverse engagement patterns of out-migrated families—a key component of the “relationship population”—in Shimogo and Nanmoku, using cluster analysis (

Table 8. Comparative analysis of out-migrated family clusters: driving motivations, geographic context, and support activities.

Driving Motivation and Support Approach	Key Characteristics of the Clusters	Hometown Visit Frequency, Location, and Age	Support Activities
Shimogo Family: Obligation-Driven and Localized Support	Cluster 1 N = 107 Obligation-driven Visits
	Primarily younger, distant relatives, providing minimal support and community engagement.	Primarily annual visits; Kanto-based (35.6%); trips made by younger individuals (61.4% are under 50)	Limited practical or emotional support.
	Cluster 2 N = 47 High Support and Engagement
	Older Aizu-based relatives with extensive practical engagement and community involvement.	High rate (>monthly) of visits; predominantly Aizu (71.7%); mostly older (65.2% are over 50)	High involvement in chores, shopping, yard work, farm work, and community activities.
	Cluster 3 N = 46 Community-Maintenance-focused visits
	Mostly Aizu-based, with a focus on supporting farm work.	Moderate number (mostly annual) of visits; primarily Aizu (55.8%); primarily older individuals (over 50: 54.4%)	Strong focus on farm work.
	Cluster 4 (N = 113) Leisure-oriented visits
	Mix of Aizu and Kantō relatives prioritizing social and recreational activities over direct support.	Moderate number (mostly annual, sometimes > monthly) of visits; Aizu and Kanto (50%/32.1%); mixed ages (47.7% are under 50)	Primarily social and outdoor leisure activities.
Nanmoku Family A: Proactive Support and Engagement	Cluster 1 N = 32 Sustained support and multifaceted engagement
	Older, Seimo-based relatives providing comprehensive and consistent support.	Mostly monthly visits (68.7%); parents (16.1%) and children (71.0%) in Seimo (77.4%); ages: 50–65 (45.2%)	High involvement in shopping support, yard work, and farm work.
	Cluster 2 N = 34 Community-focused visits
	Oldest relatives with high visit frequencies, primarily focused on shopping, farm work, and community engagement.	Mostly weekly visits (87.9%); primarily children (78.8%) in Seimou (86.2%); oldest in the cohort, with most being over 50 (90.9%)	Focus is on shopping support and community activities.
	Cluster 3 N = 89 Obligation-driven visits
	Primarily younger individuals with limited direct support, driven by a sense of obligation.	Annual visits (40.7%); Kanto-based (27.6%); younger individuals (40.9%)	Main purposes of hometown visits are family reunions and grave visits.
Nanmoku Family B: On-Demand Support and Flexible Engagement	Cluster 1 N = 17 Farm work-driven Visits
	Older relatives with high visit frequencies, primarily focused on farm work.	Weekly visits to see children (70.6%); Seimou (88.2%); over 50 (41.2%)	Primarily farm work and outdoor leisure.
	Cluster 2 N = 46 Leisure-driven flexible support
	Varied relatives, with visits combining household assistance and leisure.	Monthly visits to see children (68.9%) and siblings (22.0%); Seimou (81.4%); mixed ages (all stages of life)	Balancing shopping support with outdoor leisure activities.
	Cluster 3 N = 56 Contact-driven visits
	A mix of ages and locations, with infrequent visits and less direct forms of support.	Annual visits to see children (70.9%) and siblings (25.5%); Kanto-based (34.5%); more varied ages (71.4% under 50)	Primary focus is on social activities.

Percentages represent within-cluster proportions based on valid responses; sample sizes vary across clusters. For Shimogo and Nanmoku Family A, k-means clustering was performed on object scores derived from MCA dimensions 1 and 2; clusters were selected based on a consideration of the explained variance, evenness of cluster sizes, and the interpretability of the resulting clusters. For Nanmoku Family B, a two-step cluster analysis was used, with a three-cluster solution chosen based on Schwarz’s Bayesian Criterion (BIC) (Shimogo data; see our previously published study [52]; Nanmoku data: see Table A8, Table A9, Table A13, Table A14 and Table A15).

) to synthesize the findings on geographic isolation (Section 5.1) and support motivations (Section 5.2). The visualization of these support pathways in Figure 6 can help one interpret the cluster analysis, highlighting that support strategies in depopulated rural areas of Japan are not simply a product of in- or out-migration but shaped by a combination of geographical context and individual agency, all of which influence their motivations and their subsequent contributions to rural resilience. Table 8 then summarizes the core findings, connecting the preceding analyses to show the influence of out-migrant preferences and motivations.

A comparative analysis of the clustered data (Table 8) reveals that while Shimogo’s out-migrant support is structured around perceived resident needs—ranging from the more distant, obligation-driven visits (Cluster 1, N = 107) to more consistent engagement prioritizing practical household support (Cluster 2, N = 47) or community-focused activities (Cluster 3, N = 46) or recreation (Cluster 4, N = 113)—Nanmoku Family A, with the most frequent visitors, demonstrates a more strategic approach, with visits structured through a combination of practical tasks (such as shopping and yard work, Cluster 1, N = 32) and community activities (Cluster 2, N = 34). This finding shows that this behavior is not simply a reaction to a need but an approach to support that reflects out-migrants’ motivations. In contrast, the results for Family B reveal that their support system ranges from those with highly specific farm work needs (Cluster 1, N = 17) and those who prioritize a more balanced approach that combines shopping with outdoor leisure activities (Cluster 2, N = 46) to others who focus on maintaining a more symbolic connection through less frequent annual visits (Cluster 3, N = 56). This analysis demonstrates how these varying choices highlight a diversity of approaches to support within rural communities and their relationship with the overall influence of out-migrants within this context.

The differing pathways of out-migration and hometown support are illustrated in Figure 8. This diagram illustrates that Shimogo’s support pathways are primarily localized, with visits often originating from nearby Aizuwakamatsu City. In contrast, while many of Nanmoku’s most frequent visitors (Family A) demonstrate a preference for the nearby cities of Tomioka and Takasaki, the less frequent visitors (Family B) are more likely to engage with their hometown from a more distant location, often in the Tokyo Metropolitan Region. This distribution highlights that the out-migrants, though connected to the Tokyo area, demonstrate a preference for the closer regional urban centers when structuring their engagements and that the location of their out-migration is directly connected to the forms of support that they can provide. As summarized in

Table 9. Contrasting rural support systems: Shimogo and Nanmoku.

Feature	Shimogo Resident-Driven Support	Nanmoku Out-Migrant-Driven Support
Geographic Context	Localized isolation; self-contained support	Isolation within metro region; strategic support networks
Primary Driver of Support	Resident needs as perceived by out-migrants	Out-migrant motivations shape support delivery
Focus of Support	Primarily meeting immediate local needs	Balancing family care with community engagement, strategic support
Out-Migrant Role	Supplements local resources; responds to specific needs	Family A provides direct support; Family B provides flexible and symbolic support
Limitations	Limited by proximity and family availability with respect to acting as a support network	Limited by out-migrants’ individual choices regarding their engagement and patterns of visiting
Flow Pathways	Localized networks; nearby cities and towns are the focus	Strategic use of local towns and regional hubs; limited direct support from Tokyo

This table summarizes the core findings of this study, contrasting the forms of support in Shimogo and Nanmoku.

, this analysis demonstrates that support systems in depopulated rural areas are shaped by a complex interplay of both geographic contexts and the decisions of out-migrants, all of which are structured by the needs of the residents and the agency of the out-migrant families. Shimogo is characterized by a system that has emerged to respond to the needs of the local population, creating a self-reliant and localized form of support, whereas Nanmoku has a more varied set of approaches to support. We studied how this is directly influenced by the motivations of out-migrant families and their personal choices regarding how to structure this support. These findings suggest that relying solely on a single framework may not fully capture the complexities of rural support, as the “relationship population” represents a diverse and heterogeneous group with varying dynamics and contributions.

5.4. Illustrative Case Studies: Lived Experiences in Nanmoku

This section presents two Nanmoku case studies illustrating how resident and out-migrant agency shapes support systems (

Table 10. Comparative analysis of the illustrative case studies in Nanmoku.

Resident Profile	Case 1: Isolated Elderly Single Resident	Case 2: Interconnected Couple
Resident Demographics	This individual is an 80-year-old woman with a strong desire to remain where she is. She lives alone in a geographically isolated area and has two out-migrated children.	This couple (over 65) is composed of a husband who is still working and a wife who manages household and farming activities, with the produce being primarily for family use. They have two out-migrated children (and grandchildren).
Support Network	This support network is multi-layered, with the local community (a mobile vendor, an on-demand car dealer, and a day service for bathing) and family (a son in Takasaki and a son in Tokyo) providing distinct forms of support.	Intergenerational support: their son in Tomioka provides regular practical assistance, and their daughter in Aichi will return during summer vacation, with this couple providing childcare during visits.
Out-Migrant Mobility	Son in Takasaki (Cluster A1, shopping support and community activities) and son in Tokyo (Cluster B1, farm work).	Son in Tomioka (Cluster A1, mainly shopping support), daughter in Aichi (Cluster B3), and son with variable locations and mixed support.
Community Context	Severely depopulated, limited infrastructure; reliant on community support.	An active, though aging, community with existing local infrastructure and where the husband remains engaged as a local organizer.
Resident Capacity	She is limited by her age and geographic isolation, but her desire for autonomy and resistance to moving is strong; her children respond with individually structured support.	This couple has a more active approach to daily living, with a focus on maintaining their home, their connections with their family, and their community through their own efforts.
Long-Term Prospects	Limited U-turn potential due to the remote location and the limited resources of the community.	Higher U-turn potential, with the son in Tomioka demonstrating ongoing engagement and a strong likelihood of maintaining multigenerational ties.
Limitations	Limited by proximity and the availability of this woman’s family to act as a support network.	Limited by out-migrants’ individual choices regarding their engagement and patterns of visiting.

These illustrative case studies are drawn from interviews with two Nanmoku residents. The sample is small (totaling 12 individuals), so it may not be fully representative. The details of the location within Nanmoku have been omitted to maintain privacy.

). Case Study 1 reveals how one elderly resident actively maintains her autonomy despite being geographically isolated, while Case Study 2 demonstrates how another family proactively structures support through intergenerational engagement and community participation. These examples offer critical insights into the diverse forms of engagement within the “relationship population” and how these forms are shaped by the interplay of individual motivations and geographic context, as further illustrated by the mobility pathways depicted in Figure 9.

Case Study 1: Autonomy and Multi-Layered Support in the Face of Isolation. This case study features an 80-year-old woman in Nanmoku who, despite being geographically isolated and elderly, maintains her autonomy through a combination of family support and community services (Table 10). Her strong desire to remain in her home shapes the support provided by her two sons: one based in nearby Takasaki (Cluster A1), who provides regular assistance, and another in Tokyo (Cluster B1), who focuses on less frequent but targeted farm support. This family support network is further supplemented by community resources, including a mobile vendor and a day service for bathing. This case highlights how individual agency, family support, and community services can be combined to create a viable support system, even in challenging circumstances.

Case Study 2 Active Aging and Intergenerational Reciprocity: The second case study features a couple in their late 60s residing in Nanmoku who, unlike the previous case, demonstrate a high degree of self-sufficiency while still benefiting from a strong intergenerational support network (Table 10). This couple’s lifestyle is characterized by active participation in their community, with the husband still working and playing a key role in community activities, and the wife managing the household and tending to their small farm. The couple’s support system involves their children and grandchildren. Their son in Tomioka (Cluster A1) provides regular practical assistance. Their daughter in Aichi (Cluster B3) visits periodically, providing a connection to the community while her children receive childcare from their grandparents. This case highlights the proactive nature of this family’s approach to support, where both generations contribute to a dynamic and mutually beneficial system. This dynamic, mutually beneficial system exemplifies a proactive approach to support, with the potential for future return migration via the couple’s more mobile unmarried son.

These case studies emphasize that support systems in depopulated rural areas are shaped not only by geographic context and resident needs but also by the agency and motivations of out-migrated family members. Case 1 illustrates how an elderly resident’s desire for autonomy can structure a multi-layered support system that draws on both family and community resources. Case 2 highlights that support can involve reciprocal relationships across generations, with out-migrants strategically balancing practical assistance, family ties, and personal preferences. These examples underscore the limitations of simplistic models of rural support and emphasize the need to recognize the diverse forms of engagement within the “relationship population”, whose members are often driven by a complex interplay of obligation, choice, and personal circumstances.

6. Discussion

This comparative study of Shimogo and Nanmoku revealed that support systems in depopulated rural areas are shaped by geographic isolation, resident needs, and the motivations of out-migrated family members. Our analysis progressed from examining geographic context (Section 5.1) to exploring motivations and support behaviors (Section 5.2), identifying diverse engagement pathways through cluster analysis (Section 5.3), and finally illustrating lived experiences through case studies (Section 5.4). This multi-faceted approach demonstrates the limitations of simplistic migration models and highlights the “relationship population’s” agency in structuring support. This discussion will explore the implications of these findings for rural mobility, strategic support choices, and policy interventions.

6.1. The Evolving Role of Local Towns in Depopulated Rural Areas

Contrary to the narrative depicting rural decline as a one-way trajectory, this research highlights the evolving role of local towns as critical anchors within dispersed support networks [2,64,65], demonstrating that the “relationship population’s” strategic engagement with these hubs is vital for rural sustainability in Japan [66]. The contrasting experiences of Shimogo and Nanmoku reveal that these towns are not merely passive bystanders in the process of depopulation. In Shimogo, as shown in the analysis of geographic context (Section 5.1) and out-migrant locations (Table 5, Figure 6), the nearest urban center, Aizuwakamatsu City, serves as a vital hub for the localized support network. This finding suggests that Aizuwakamatsu is more than just a destination for out-migrants; it functions as an anchor for maintaining connections within the Aizu region, facilitating the provision of support and sustaining the social fabric of the community through its connection to Shimogo [67].

Nanmoku presents a different dynamic, where the local towns of Tomioka and Takasaki act as strategic gateways, connecting this geographically isolated village to broader regional and metropolitan networks (Section 5.1, Figure 6). This highlights that out-migrants are making strategic choices about where to live, reflecting a balance between the desire for urban opportunities (“Tokaishikou”) and the pull to maintain strong ties with their hometowns (“Chimoto Teichaku”) (Section 5.3, Table 8 and Table 9) [68]. The decision to reside in these local towns, as illustrated by the prevalence of these locations among Family A members (Table 6), reflects a deliberate effort to balance proximity with access to wider opportunities and resources.

These contrasting patterns suggest that in the context of depopulation, local towns are evolving into critical nodes within an increasingly dispersed, yet interconnected, rural landscape. This finding indicates that policy interventions should not only focus on the villages themselves but also consider how best to support and strengthen these local towns, recognizing their increasingly important roles as centers for economic activity, social interaction, and the provision of essential services for the surrounding rural areas as well as in facilitating connection between out-migrants and their families [66]. Furthermore, these findings compel us to rethink the very nature of “Chiiki” (local region) revitalization policies [69]. Effective revitalization may require moving beyond a narrow focus on economic development to create a framework that incorporates the social and emotional ties of the “relationship population”, recognizing that the value of a community is not simply based on economic activity but also the social and emotional ties that bind it together.

6.2. Beyond Obligation: Agency, the “Peasant Economy”, and the “Relationship Population”

Traditional frameworks for understanding rural communities in Japan have often relied on concepts such as a “peasant economy” [70,71], a “moral economy” [72,73], and “sōgo fujo” (mutual aid) [35] to explain the social relationships and reciprocal obligations that underpin community life. However, the rapid depopulation of rural areas has challenged the traditional foundations of these systems, raising questions about their continued relevance in contemporary rural Japan [74,75]. While some studies have viewed the decline in “sōgo fujo” as a symptom of rural decline and the erosion of community ties [76], this research offers a different perspective, one that recognizes the enduring importance of family and community ties while acknowledging the emergence of new forms of engagement and support that are shaped by the agency and motivations of the “relationship population”.

The contrasting experiences of Shimogo and Nanmoku, as detailed in our analysis of geographic isolation (Section 5.1), resident needs, and out-migrant motivations (Section 5.2), as well as our analysis of behavioral patterns and clusters (Section 5.3), reveal that depopulation does not simply lead to a uniform erosion of traditional support systems but instead creates a shift in how these systems operate, based on both needs and motivations. In Shimogo, geographic isolation has fostered a localized support system where the actions of out-migrated relatives primarily reflect resident needs. The MCA dimensions of “resident self-reliance” and “community support needs” (Table 7) highlight how Shimogo’s out-migrants respond to the community’s capacity for self-sufficiency and its reliance on external resources. This is a system structured around a clear understanding of resident needs and driven by a need to maintain the existing social and community structures. The clusters identified in Shimogo (Table 8) further illustrate this pattern, with out-migrants engaging in a spectrum of activities that range from more infrequent, obligation-driven visits (Cluster 1) to more consistent support focused on practical household or farming needs (Cluster 2) or community engagement (Cluster 3). These patterns of support demonstrate that Shimogo’s out-migrants primarily respond to the perceived needs of residents within a localized network.

Nanmoku, however, presents a contrasting picture, where the “sōgo fujo” of support is more clearly driven by the motivations and agency of out-migrated relatives, who use a wider range of resources to maintain contact with their hometowns. The MCA dimensions for Family A and Family B (Table 7, Table A6 and Table A11; Figure A1 and Figure A3) capture distinct motivations for engagement: “daily support” and “community engagement” for Family A, and “on-demand support” combined with more symbolic forms of “community engagement” for Family B. The cluster analysis (Table 8 and Table 9) further revealed how these motivations translate into diverse patterns of engagement, with Family A demonstrating a clear commitment to providing both practical, everyday support and more symbolic forms of community participation, while Family B adopts a more flexible approach, responding to specific needs and maintaining connections through social and cultural activities. The differing patterns of mobility and support that are evident in Nanmoku are linked to the choices made by the out-migrants regarding where to live and how to maintain their connections with their families and communities. Our strategic use of both nearby towns and the more distant metropolitan region highlights that, in contrast to the more reactive approach in Shimogo, these out-migrants have agency in shaping support.

This analysis challenges the notion that rural depopulation inevitably leads to the demise of a “peasant economy” and “sōgo fujo”. Instead, it reveals that these concepts are being reinterpreted and adapted within the context of the “relationship population”, as out-migrants and residents negotiate new forms of engagement and support, demonstrating that individual choices, alongside deep-rooted cultural values, are crucial for understanding how rural communities are adapting to the challenges of demographic change. By acknowledging this diversity and recognizing the agency of both residents and out-migrants, we can develop more effective strategies for sustaining rural communities in the face of depopulation.

6.3. Beyond In-Migration: Empowering the “Relationship Population” for Rural Revitalization

This research demonstrates that rural support systems are not shaped primarily by in- or out-migration [76,77] but by a complex interplay of geographic context, resident needs, and the strategic choices of a diverse “relationship population”. This subsection moves beyond traditional migration models to explore the lived realities of support in depopulated areas and their implications for both policy and community revitalization.

While the desire for a more idyllic rural life and the inherent value of rural–urban mobility have been presented as driving forces for attracting residents to rural areas [30,45], the findings from Shimogo and Nanmoku highlight that support is, more often than not, the primary motivation for out-migrants to maintain connections with their hometowns. Although many may maintain a connection to place through patterns of leisure activities, most out-migrants are more driven by a need to provide support, whether through practical activities, such as farming, or through their routine engagements with community life [78]. We have also shown that there are strong limitations to the forms of support that can be delivered to highly isolated areas. As a result, policymakers must question whether promoting in-migration, particularly to severely depopulated areas such as the one seen in Case Study 1 in Section 5.4, is an effective solution for addressing the challenges of demographic decline. The complex networks of support in such areas also suggest that U-turn and J-turn migration must be understood in light of the broader social and practical realities of everyday life in rural communities and not simply as a straightforward means of addressing a demographic imbalance. These forms of rural mobility, as well as the limitations that they impose, must, therefore, be understood as a crucial component of what shapes the patterns of engagement that are visible throughout these communities [45]. It is not simply a case of promoting an urban-to-rural migration or even of encouraging former residents to return home, as these new forms of engagement also exist alongside deeply rooted bonds of family, tradition, community, and a sense of shared history [68,70,78].

These findings, particularly the contrasting examples from Shimogo and Nanmoku, highlight that simplistic migration policies fail to recognize the limitations imposed by geography, resources, and existing support networks. In Shimogo, with its self-reliant and localized structure, support is often structured as a reaction to resident needs. Therefore, direct policies promoting further migration may have a limited effect. In contrast, the data from Nanmoku highlight that out-migrant family members have developed a far more complex and strategic approach to their support, wherein they are making active choices regarding their participation and levels of engagement and the type of support that they will provide, shaped by their own capacity for action and personal preferences for engagement. The more nuanced patterns of support that have been structured by out-migrant communities in Nanmoku, and the strong ties to their hometowns that have resulted from this strategic behavior, highlight the importance of looking for ways to support J-turn migration as a means of strengthening communities while also recognizing that they do not always provide a magical solution to the challenges of rural depopulation. In cases like Nanmoku, it may be more productive to support the local hubs, such as Tomioka and Takasaki, which are already being used by out-migrants to structure their support systems and which provide easier access to a complex set of social and economic ties. Any policies must also be structured to reflect the needs of specific communities and the requirements and personal choices of those who are providing support.

This research emphasizes that rural sustainability cannot be achieved through a simple focus on migration alone but is instead the result of a complex interplay between geographic context, resident needs, and the ability of the “relationship population” to structure their own support systems in depopulated regions of Japan [79]. To ensure the long-term vitality of these areas, policymakers must move beyond simplistic models and develop strategies that explicitly value, support, and empower the diverse motivations and choices that structure the actions of both residents and their out-migrated relatives, creating an environment where both residents and the “relationship population” can actively contribute to the creation of sustainable rural communities.

6.4. Strengths and Limitations

This study has several key strengths. Its comparative design, analyzing distinct rural communities (Shimogo and Nanmoku), provides a nuanced understanding of how geographic context shapes support systems and out-migrant engagement. The mixed-methods approach employed, integrating quantitative and qualitative data, offers a rich perspective on the “relationship population”, revealing both broad patterns and individual experiences. This approach is also novel with regard to its focus on the agency and motivations of out-migrants, rather than solely resident needs, challenging conventional models of rural decline. Our detailed analysis of geographic context, motivations, and support pathways, combined with illustrative case studies, also provides a strong foundation for developing targeted policy recommendations.

Despite these strengths, this research has limitations. The relatively small sample size, particularly for qualitative data, restricts the generalizability of our findings. Reliance on cross-sectional data limits insights into the dynamic evolution of support systems over time. Future research should examine the long-term sustainability of these support models, the integration of formal and informal care systems, and the impact of evolving policies on rural networks. Comparative studies across different rural contexts could further validate these findings and expand their applicability. In addition, while the contrasting characteristics of Shimogo and Nanmoku provided a valuable framework for comparison, the significant difference in their population sizes should be acknowledged as a potential limitation. By addressing these issues, we can better understand how we can sustain aging, depopulated rural areas through innovative, inclusive strategies.

7. Conclusions

This study examined how geographic isolation and out-migrant motivations shape support systems in two depopulated mountainous rural areas of Japan: Shimogo Town and Nanmoku Village. By challenging the prevailing policy focus on internal migration, this research offers a new perspective by highlighting the “relationship population”—out-migrated family members who maintain ties to their hometowns—as a key component of rural sustainability. Through a mixed-methods approach, combining quantitative analyses of household surveys with qualitative analyses of interviews and case studies, we identified contrasting support patterns: Shimogo exhibits a localized system driven by resident needs, while Nanmoku reveals a strategic approach balancing practical support and community engagement.

Our findings underscore the limitations of one-size-fits-all migration policies and emphasize the need for context-specific strategies that recognize the diverse motivations and contributions of this “relationship population”. We have shown that while Shimogo’s support system is primarily a reflection of localized self-sufficiency and the expressed needs of its residents, the networks in Nanmoku are driven by the more diverse motivations and individual support preferences of its out-migrated relatives. These findings highlight the importance of mobility and dynamic exchanges, grounded in family ties, hometown bonds, and traditional cultural practices, as critical elements in fostering the long-term sustainability of rural communities. Furthermore, these findings suggest a need to re-evaluate top-down, migration-focused policies to acknowledge the power of existing family ties that are created and supported by local communities, especially in rural settings, where household-based, small-scale means of production (such as in Shimogo) have historically been the basis of social and economic life (as in a peasant economy) [70,71] and where such ties remain an important part of the local social structure.

This research contributes to the field by offering a new perspective on rural sustainability, one that moves beyond a purely demographic focus and embraces the complex social networks that underpin these communities. By using a mixed-methods approach, we were able to combine different types of data and show how they relate to each other. The framework we used and the approaches we have taken also provide a new way to study the complex interplay between mobility, support, and community resilience. Our findings have important implications for policymakers, local leaders, and researchers working on rural revitalization and aging in place and suggest that future strategies should empower the “relationship population” and support the diverse forms of engagement it offers.

Specifically, for Shimogo, policies should focus on strengthening localized support networks through investment in community resources and support for traditional activities while also facilitating practical support for out-migrants and encouraging farm-based return visits and participation in local cultural events, even for shorter periods, while also highlighting the importance of regional hubs such as Aizuwakamatsu City as a key area for a sustainable support system. Aizuwakamatsu is an important part of the Aizu area, and unlike Tomioka and Takasaki in Seimou, which are located within or close to the Tokyo metropolitan area, its own sustainability is also highly linked to the sustainability of the regions surrounding it. For Nanmoku, policies should support strategic mobility, remote living, and intergenerational reciprocity, recognizing the importance of family ties in a dynamic support system; improve transportation services for travel to nearby regional hubs (Tomioka and Takasaki) to facilitate the relationship population’s engagement; promote remote work opportunities; and support the development of local businesses and services that cater to both residents and out-migrated families.