Relationships among rural tourism participation, livelihood capital structure, and household income outcomes: a cross-sectional survey of tourism-participating households in the Yunqiu Mountain Area, Yellow River Basin

Abstract

Purpose

Against the backdrop of China’s rural revitalization strategy and the national strategy for ecological conservation and high-quality development in the Yellow River Basin, rural tourism has become an important avenue for expanding livelihood opportunities in some resource-transformation mountain areas. Existing studies have largely examined the relationship between tourism development and household income or livelihood transformation, while paying relatively little attention to the statistical associations among tourism participation, livelihood capital structure, and income outcomes within households already participating in tourism, particularly in resource-transformation mountain areas of the Yellow River Basin.

Methods

Drawing on the Sustainable Livelihoods Framework, this study extends the conventional five-capital framework by incorporating psychological capital and develops a six-dimensional analytical framework comprising natural, physical/material, financial, human, social, and psychological capital. Using 374 valid household questionnaires from tourism-participating households in Yunqiu Mountain in the Yellow River Basin, we employ an ordered logit model, regression analyses, and partial least squares structural equation modeling (PLS-SEM) to examine the statistical associations among tourism participation, livelihood capital, and household income outcomes.

Results

Tourism participation is significantly and positively associated with household income outcomes. It is also positively associated with all six dimensions of livelihood capital, with comparatively larger standardized associations for psychological and financial capital. All six livelihood capital dimensions are positively associated with household income outcomes; the coefficients for psychological and financial capital are numerically larger, although their difference is not statistically significant. Supplementary PLS-SEM analysis indicates that all six livelihood capital dimensions form statistically significant indirect association pathways linking tourism participation and household income outcomes.

Conclusion

The study extends the conventional livelihood capital framework by incorporating household psychological agency and provides micro-level empirical evidence on how tourism participation, livelihood capital, and income outcomes are statistically related in resource-transformation mountain communities in the Yellow River Basin.

Introduction

Comprehensively advancing rural revitalization and promoting common prosperity in rural areas are important tasks in China’s contemporary modernization agenda. Industrial development can support rural development by facilitating resource use and absorbing labor (Wang and Liu, 2023; Sun and Wang, 2023). Rural tourism, drawing on diverse rural resources, expands household livelihood choices through employment, business operation, agricultural product sales, and industrial integration, and has therefore been regarded as an important pathway for rural transformation and household income generation (Gao and Wu, 2017; Zhang et al., 2021). In recent years, China has continued to promote the integrated development of agriculture, culture, and tourism, as well as the expansion of rural leisure industries. The Rural Revitalization Promotion Plan (2024–2027) further emphasizes developing distinctive rural cultural and tourism industries and the role of rural tourism in increasing rural residents’ incomes. According to estimates from the Data Center of the Ministry of Culture and Tourism, rural tourism in China received 3.157 billion visits in 2025, an increase of 9.3% year on year; annual tourism revenue reached RMB 1.81 trillion, up 9.6%; and the number of people employed in rural tourism reached 7.83 million (Dai, 2026). These figures underscore the importance of rural tourism for rural industrial development and employment and highlight the need to examine its relationship with household income more closely.

The relationship between rural tourism and household income, livelihood transformation, and community development has long been an important topic in tourism and rural development research. Existing studies suggest that rural tourism can provide rural households with employment and business opportunities beyond agriculture and is associated with household income and livelihood sustainability (Su M. et al., 2019; He et al., 2021). For example, Su M et al., (2019), studying Hetu Town in Anhui Province, found that combining tourism activities with traditional livelihoods can enhance livelihood diversity and livelihood freedom. However, unequal tourism participation may also generate social risks. Using provincial panel data from China, He et al. (2021) found a significant association between rural tourism development and sustainable farmer income. However, the strength of the association varied across income types. Tourism benefits, however, are not necessarily distributed evenly across households. Household resource endowments, modes of participation, and positions within the tourism industry may shape their ability to capture tourism benefits and improve their livelihoods (Gao and Wu, 2017; Su M. et al., 2019). Tourism is also characterized by seasonality and market fluctuations, requiring tourism households to adjust livelihood strategies over the year and potentially generating differentiated livelihood outcomes (Su Z. et al., 2019; Su et al., 2022). Thus, tourism participation and household income should not be viewed as a simple one-to-one relationship; household resources, participation modes, and the external operating environment may all be associated with income outcomes. Moreover, a binary distinction between participation and non-participation does not adequately capture tourism participation. It also encompasses different modes and degrees of participation. Rural households may participate through tourism businesses, employment, and other tourism-related livelihood activities, and different tourism livelihood strategies are associated with household resources and livelihood sustainability (Su Z. et al., 2019). Research on traditional village tourism in China has likewise shown that tourism businesses, farmer cooperation, and benefit-sharing arrangements are closely related to community participation and the distribution of tourism benefits (Gao and Wu, 2017). These findings suggest that research should move beyond whether households participate in tourism to examine differences among households that already participate. Studies of the relationship between tourism participation and community power further indicate that participation in tourism-related economic activities does not necessarily imply equal positions in resource control and decision-making. For example, Abdalla et al. (2022), examining women’s participation in Zanzibar’s tourism industry, found that socioeconomic factors, including the nature of tourism activities, were associated with women’s participation in decision-making. Collectively, these studies indicate that tourism participation may entail differentiated benefits, resource control, and decision-making opportunities.

At the theoretical level, the Sustainable Livelihoods Framework (SLF) provides an important tool for analyzing the relationships among household resource endowments, livelihood strategies, and livelihood outcomes. Existing tourism livelihood studies have examined tourism development and rural livelihoods through five forms of capital: natural, human, social, financial, and physical (Su Z. et al., 2019; Yu et al., 2020). Some studies have also contextualized the framework by incorporating dimensions such as cultural resources and psychological capital (Ma et al., 2021; Chen and Wang, 2023), providing a basis for examining heterogeneity among tourism-participating households. The Yellow River Basin is rich in natural and cultural resources, and rural tourism has become an important component of rural transformation and industrial development in parts of the basin. Its development varies substantially across locations owing to differences in historical and cultural resources, accessibility, economic development, and topography (Zhang et al., 2021). In resource-transformation mountain areas, the relationships among tourism development, household livelihood capital, participation modes, and livelihood strategies are particularly complex, yet micro-level evidence remains limited (Liu et al., 2025). Against this background, three areas warrant further attention. First, researchers should differentiate subjects more carefully: many studies examine tourism participation and income or livelihood change at the aggregate level, with less attention to heterogeneity among households already participating in tourism. Second, the dimensions of livelihood capital require further extension: the conventional SLF has paid relatively limited attention to psychological capital as an independent dimension, and more empirical evidence is needed on its association with tourism participation and income outcomes. Third, more contextual evidence is needed from resource-transformation mountain areas of the Yellow River Basin, where household livelihood transformation has distinctive characteristics. Accordingly, this study focuses on tourism-participating households in Yunqiu Mountain and incorporates psychological capital into a six-dimensional livelihood capital framework. It addresses three questions: (1) Is tourism participation statistically associated with income outcomes among tourism-participating households? (2) Is tourism participation associated with the six dimensions of livelihood capital, and do these associations differ across dimensions? (3) Are the six livelihood capital dimensions associated with income outcomes, and do they form indirect association pathways between tourism participation and income outcomes?

The study contributes in three respects. First, by focusing on tourism-participating households, it complements research on heterogeneity among tourism participants. Second, by incorporating psychological capital into the conventional SLF, it provides additional evidence on the role of household psychological resources in tourism-related livelihoods. Third, by examining a resource-transformation mountain village in the Yellow River Basin, it provides micro-level contextual evidence on the statistical relationships among tourism participation, livelihood capital, and income outcomes. The remainder of the paper is organized as follows. Section 2 presents the theoretical framework and hypotheses; Section 3 describes the research design; Section 4 reports the empirical results; Section 5 discusses the findings; and Section 6 concludes with management implications, limitations, and directions for future research.

Theoretical framework and hypotheses

From the sustainable livelihoods framework to an analytical framework for tourism participation and income associations

The Sustainable Livelihoods Framework (SLF) is an important theoretical tool for analyzing household resource endowments, livelihood strategies, and livelihood outcomes. The classic DFID framework distinguishes five forms of livelihood capital—human, natural, physical, financial, and social capital—and situates them within a system linking vulnerability contexts, institutions and processes, livelihood strategies, and livelihood outcomes (DFID, 1999). However, this classification focuses primarily on objective forms of capital and gives limited attention to individual psychological resources and their relationship with livelihood behavior. Bebbington (1999); emphasized that livelihood capital involves not only possessing resources but also accessing, using, and converting them into livelihood objectives. Thus, households with similar objective resource endowments may still pursue different livelihood strategies, while risk perceptions, confidence in action, and expectations about the future may shape livelihood choices under uncertainty. Psychological capital theory identifies self-efficacy, hope, optimism, and resilience as core positive psychological resources (Luthans et al., 2007; Luthans and Youssef-Morgan, 2017). Previous studies have linked psychological capital to individual performance, satisfaction, and the quality of life of rural families (Mohammadi-Mehr et al., 2025), but its independent role in tourism-related livelihoods remains insufficiently examined. Accordingly, this study extends the SLF by incorporating psychological capital into a six-dimensional framework comprising natural, physical/material, financial, human, social, and psychological capital.

The inclusion of psychological capital is motivated by the research question and the empirical context. Households in Yunqiu Mountain combine tourism employment, business participation, and traditional agricultural production. Under tourism-market volatility and employment uncertainty, self-efficacy, development expectations, and coping capacity may be associated with both tourism participation and income outcomes. Psychological capital differs conceptually from human and social capital. Human capital refers to production-related individual resources such as education, professional knowledge, and labor skills; social capital concerns the capacity to access resources through social networks, trust, and organizational ties; psychological capital refers to positive psychological resources that individuals draw upon when facing tasks, challenges, and uncertainty. Although these dimensions may be related in practice—for example, training may improve skills and self-efficacy simultaneously—they are not interchangeable. We therefore model psychological capital as a distinct latent construct and assess its discriminant validity against the other capital dimensions to reduce potential concerns about conceptual overlap.

In this study, livelihood capital structure refers to the overall configuration of household resource endowments, accessibility, capabilities, and psychological resources across six dimensions: natural, physical/material, financial, human, social, and psychological capital. Empirically, the six forms of capital are operationalized as corresponding latent constructs. Measurement-model assessment and structural-path analysis examine and compare their statistical associations with tourism participation and income outcomes. Given the cross-sectional design, the estimated relationships represent contemporaneous statistical associations among the six livelihood capital dimensions, tourism participation, and income outcomes at the time of the survey.

Hypotheses

Association between tourism participation and household income outcomes

Rural tourism participation represents households’ involvement in the tourism industry through tourism-related activities. Compared with a simple participation/non-participation indicator, examining participation by mode, breadth, and degree can better capture heterogeneity among households engaged in tourism. Previous research indicates that rural households participate in tourism through diverse livelihood strategies that are associated with household resource endowments (Su Z. et al., 2019; Su et al., 2022). Participation may take the form of employment in scenic areas, seasonal wage work, homestay or agritainment businesses, agricultural-product supply, or equity participation through land or housing. These modes differ in labor requirements, asset needs, returns, managerial autonomy, and risk exposure. For example, employment in a scenic area is generally associated with wage income and job allocation; self-employment may involve business income, asset investment, and market risk; while land or housing equity participation may involve rights of asset use and dividend arrangements (Su Z. et al., 2019; Yu et al., 2020). Accordingly, the association between tourism participation and household income outcomes may reflect participation mode, participation intensity, household resources, and institutional conditions. Tourism participation does not necessarily imply higher total household income, nor does it necessarily increase the share of tourism income in total income. Household income may also depend on agricultural production, non-farm employment, household labor structure, asset conditions, and other income sources. Existing studies suggest that tourism development can diversify rural livelihoods, but tourism benefits and their distribution are not necessarily equal across households; differences in resources and participation capacity may be associated with different income outcomes (Su M. et al., 2019; Ma et al., 2021). Conversely, households’ pre-existing income and resource conditions may shape their opportunities and modes of tourism participation. Households with greater financial resources, business premises, or social connections may find it easier to enter self-employment or asset-based participation, while labor capacity, skills, and time availability may shape participation choices (Yu et al., 2020). Therefore, we propose:

H1: Tourism participation is positively associated with household income outcomes.

Association between tourism participation and livelihood capital structure

Household participation in tourism may be associated with multiple dimensions of livelihood capital, but such associations do not necessarily imply that tourism participation directly causes capital accumulation or optimization. The SLF suggests that households mobilize and allocate different forms of livelihood capital in developing livelihood strategies and outcomes (Su Z. et al., 2019; Yu et al., 2020). Human capital may be associated with opportunities to acquire service experience, market information, and business practices through tourism employment and business activities. Such experiences may relate to skill use, knowledge accumulation, and perceived capabilities. Social capital may be associated with opportunities to interact with enterprises, cooperatives, tourists, suppliers, and other market actors. Participation may also be related to cooperation, information channels, and community interaction, although social networks may simultaneously facilitate tourism participation and reflect pre-existing differences in social resources (Su Z. et al., 2019; Ma et al., 2021). Financial capital may relate to household cash flow, savings, and financing needs, while pre-existing financial resources may shape participation modes. Physical/material capital may be associated with business premises, equipment, and household asset utilization. Previous research has linked tourism development to changes in rural households’ financial and physical capital (Yu et al., 2020; Yi, 2022). Natural capital is closely related to rural landscapes, land resources, and ecological conditions that underpin tourism. However, households may differ in their actual control over and ability to benefit from such resources (Ma et al., 2021). Psychological capital—including self-efficacy, hope, resilience, and optimism—may be associated with willingness to participate, business choices, and risk responses. Existing research has incorporated psychological capital into rural tourism livelihood analysis and suggests that it can be an important dimension of tourism households’ livelihood capital (Su et al., 2022). At the same time, market conditions and business experience may also relate to respondents’ psychological states. We therefore hypothesize that tourism participation is positively associated with each of the six livelihood capital dimensions.

H2a: Tourism participation is positively associated with household human capital.

H2b: Tourism participation is positively associated with household social capital.

H2c: Tourism participation is positively associated with household financial capital.

H2d: Tourism participation is positively associated with household physical/material capital.

H2e: Tourism participation is positively associated with household natural capital.

H2f: Tourism participation is positively associated with household psychological capital.

Association between livelihood capital structure and household income outcomes

Household income is an important indicator of livelihood conditions, but it reflects multiple factors. Different forms of livelihood capital may affect income outcomes through different channels. Existing research suggests heterogeneous relationships between livelihood capital and rural household income and income structure (Yang et al., 2025; Han et al., 2025; Zhang et al., 2025). Human capital—including education, knowledge, skills, and labor capacity—may be associated with households’ ability to access employment, operate businesses, and adapt to market demand. Social capital, including social networks, reciprocal relationships, and cooperation channels, may be related to access to market information, employment opportunities, and business resources. Financial capital may be associated with households’ capacity to operate businesses, purchase equipment, cope with fluctuations, and adjust livelihood activities. At the same time, household income may also affect financial accumulation and financing capacity. Physical/material capital, including housing, business premises, equipment, and other tangible assets, may be associated with the ability to engage in tourism and other livelihood activities, with its relevance potentially varying by participation mode. Natural capital, including land, natural resources, and ecological conditions, is related to agricultural production, tourism-resource utilization, and other livelihood activities. Psychological capital, including self-efficacy, hope, resilience, and optimism, is associated with goal pursuit, coping with difficulties, and business decision-making. As a positive psychological resource, it has been linked to behavioral performance and adaptability (Luthans et al., 2007; Luthans et al., 2010), and studies of farmers have also reported associations between psychological capital and entrepreneurial behavior (Li et al., 2022). Nevertheless, associations between psychological capital and income outcomes may also reflect pre-existing income, household resources, business environments, and unobserved factors. We therefore propose:

H3a: Household human capital is positively associated with household income outcomes.

H3b: Household social capital is positively associated with household income outcomes.

H3c: Household financial capital is positively associated with household income outcomes.

H3d: Household physical/material capital is positively associated with household income outcomes.

H3e: Household natural capital is positively associated with household income outcomes.

H3f: Household psychological capital is positively associated with household income outcomes.

Indirect association pathways through livelihood capital structure

Tourism participation, livelihood capital structure, and household income outcomes are interconnected. Tourism participation relates to employment, business activity, resource use, and social interaction, while the resources, capabilities, and psychological conditions involved in these activities are also associated with household income outcomes. Existing studies also indicate complex relationships among tourism participation, livelihood capital, and livelihood outcomes (Su M. et al., 2019; Yu et al., 2020; Su et al., 2022). Accordingly:

H4: Livelihood capital structure constitutes indirect association pathways between tourism participation and household income outcomes.

Based on these hypotheses, we develop a research model incorporating one direct association and six indirect association pathways (Figure 1).

Hypothesized research model.

Research design

Study area and case selection

The study area comprises villages surrounding Yunqiu Mountain Scenic Area in Guanwangmiao Township, Xiangning County, Linfen City, Shanxi Province. Located in the Yellow River Basin and designated as a national 5A-level tourist attraction, the area exhibits typical mountain-rural characteristics, with local industrial development closely connected to natural-resource use, rural cultural heritage, and tourism. Yunqiu Mountain has developed tourism around local natural landscapes and traditional cultural resources. As tourism has expanded, the surrounding area has gradually developed a diversified industrial pattern centered on scenic-area operations, supplemented by household employment and business participation, with agricultural production combined with tourism services. Tourism development is therefore connected not only to operations within the scenic area but also to employment, agricultural-product supply, household businesses, and community development in surrounding villages.

The villages surrounding Yunqiu Mountain display distinctive characteristics of resource transformation and agriculture-tourism integration. Scenic-area development is closely associated with adjustments in local resource use and industrial transformation, and tourism has gradually become an important option for households seeking to expand livelihood opportunities. Scenic-area operations, village development, and household livelihood activities have interacted. Local tourism-related activities include employment in the scenic area, tourism services, household businesses, agricultural-product supply, and other tourism-related economic activities. These modes differ in labor requirements, financial conditions, business assets, and how they generate returns. This study focuses on heterogeneity among households that already participate in tourism and examines the statistical associations among tourism participation, livelihood capital structure, and household income outcomes.

Data sources

The data were collected through a household survey conducted from August 13 to 20, 2025. The questionnaire covered household characteristics, tourism participation, livelihood capital, and income outcomes. A pilot survey of 50 households was conducted in Dahe Village in July 2025 to assess scale reliability and item suitability; semantically ambiguous items were revised based on respondents’ feedback. We excluded pilot respondents from the formal sample. The formal survey covered Dahe, Banershang, Donggou, and Xiachuan villages under Yunqiu Mountain. The sampling frame was a roster of resident households provided by village committees, supplemented by information on scenic-area employment, business registration, and cooperative participation to identify tourism-participating households. We used stratified random sampling, with the four villages as the first stratum and tourism-participating households within each village as the second. The number of questionnaires allocated to each village reflected the number of tourism-participating households and fieldwork conditions. The approximate numbers of tourism-participating households were 180 in Dahe, 110 in Banershang, 100 in Donggou, and 80 in Xiachuan; 145, 95, 90, and 70 questionnaires were therefore distributed, respectively. Within each village, we randomly selected households using a random-number table. During fieldwork, 7, 6, 4, and 7 households in the four villages, respectively, were unavailable or refused to participate and were replaced with subsequent households on the roster. This replacement procedure may introduce non-random response bias; accordingly, the sample is not treated as strictly probability-representative, and the findings are limited to statistical associations among tourism-participating households in the study area.

We distributed 400 questionnaires. Trained interviewers conducted face-to-face household interviews in Mandarin, with dialect explanations provided when necessary. Before each interview, the researchers explained the study purpose, survey content, intended data use, confidentiality arrangements, and the voluntary nature of participation, and documented informed consent. Respondents could decline participation, refuse to answer any question, or terminate the interview at any time without adverse consequences. After collection, we screened questionnaires according to pre-specified data-quality criteria. We excluded 26 questionnaires: 11 for patterned responding, 6 for internal inconsistency, 5 for missing key items, and 4 for implausibly short completion times. The final sample comprised 374 valid questionnaires: 138 from Dahe, 88 from Banershang, 85 from Donggou, and 63 from Xiachuan, yielding a valid response rate of 93.5%.

To strengthen contextual interpretation, the study also draws on qualitative materials accumulated through the research team’s long-term fieldwork. Since 2014, the team has conducted sustained field research on tourism development and rural revitalization in Yunqiu Mountain, including multiple rounds of semi-structured interviews with 62 individuals, including scenic-area managers, village committee officials, and different types of households, together with participant-observation records, village documents, and local policy texts. These materials were used only to understand the local tourism-development process, household participation modes, and livelihood-transformation context and to support contextual interpretation of the case and empirical findings. They were not entered into the quantitative models or pooled with the 374 household questionnaires.

Variable measurement and operationalization

The core variables comprise rural tourism participation, six dimensions of livelihood capital, and multiple income-related outcome variables. Measurement indicators were primarily adapted from the Sustainable Livelihoods Framework and prior research on household livelihood capital and rural tourism participation, with contextual modifications reflecting tourism development in Yunqiu Mountain (Table 1).

Variable measurement, operationalization, and theoretical basis.

Core construct | Dimension | Code | Measurement item | Measurement scale and coding | Theoretical and measurement basis
Tourism participation (TP) | Economic participation | TP1 | I provide tourism-related services, such as management and visitor guiding, for the scenic area. | 5-point Likert: 1 = strongly disagree; 5 = strongly agree | Chen and Wang (2023); Yu et al. (2020); Yi (2022)
TP2 | I supply agricultural products, handicrafts, or other goods to the scenic area or other tourism businesses.
TP3 | I independently operate, or make use of local resources to operate, tourism-related businesses such as specialty homestays and cultural tourism experiences.
TP4 | I participate in tourism projects operated by the scenic area, cooperatives, or other tourism businesses through land, housing, capital, or other forms of equity contribution and participate in the distribution of returns.
Participation breadth | TP5 | My tourism activities span multiple tourism segments, including accommodation, dining, and educational or study-oriented tourism, with participation in two or more types. | He et al., (2021); Su et al., (2022)
TP6 | In addition to providing services directly to tourists, I provide products or services to other cultural and tourism businesses.
TP7 | I use digital platforms, such as short-video or livestreaming platforms, to showcase or promote my tourism products and services.
Decision-making agency | TP8 | I participate in meetings concerning tourism development in the village or scenic area and can express my views. | Abdalla et al. (2022); Yang et al. (2025)
TP9 | I proactively propose suggestions to upgrade tourism products or improve tourism services.
Human capital (HC) | Skills and knowledge | HC1 | I actively participate in tourism-related skills training provided by the government, scenic area, or other organizations. | 5-point Likert: 1 = strongly disagree; 5 = strongly agree | Su M. M. et al., (2019); Li et al. (2020); Chen and Wang (2023)
HC2 | I possess the communication, marketing, financial management, and other skills required for tourism-related business activities.
HC3 | I can communicate with visitors from outside the area and have relatively broad knowledge and perspectives on development opportunities.
Employment and labor allocation | HC4 | Tourism development has created more opportunities for household members to return home or find employment locally.
Social capital (SC) | Organizational networks | SC1 | I am a member of a tourism-related specialized cooperative and have corresponding opportunities to participate and provide input on decision-making. | 5-point Likert: 1 = strongly disagree; 5 = strongly agree | Su M. M. et al., (2019); Chen and Wang (2023)
Social trust | SC2 | I trust the scenic area's tourism development plans and management approach and am willing to cooperate with the scenic area over the long term.
SC3 | I trust village officials to play a constructive role in tourism development and dispute resolution.
Community identification | SC4 | I identify with the outcomes of local tourism development and am willing to actively help safeguard the reputation of the village and scenic area.
Financial capital (FC) | Financial reserves | FC1 | My household has sufficient disposable funds to support tourism-related business activities or cover unexpected expenditures. | 5-point Likert: 1 = strongly disagree; 5 = strongly agree | Su M. M. et al. (2019); Li et al. (2020)
Access to finance | FC2 | My household can obtain loans for business or living expenses relatively easily through banks, credit cooperatives, or other formal financial channels.
Physical/material capital (PC) | Living facilities | PC1 | Housing, roads, water and electricity supply, internet access, and other living and public facilities in the area are relatively well developed. | 5-point Likert: 1 = strongly disagree; 5 = strongly agree | Chen et al. (2017); Yu et al. (2020); Su Z. et al., (2019)
Productive assets | PC2 | My household owns modern transportation and communication facilities and the equipment needed for tourism-related business activities.
Use of cultural resources | PC3 | Local tangible cultural heritage is well preserved and appropriately utilized.
Natural capital (NC) | Resource endowment | NC1 | The village has sufficient natural resources, such as cultivated land and forestland. | 5-point Likert: 1 = strongly disagree; 5 = strongly agree | Yu et al. (2020); Su Z. et al., (2019)
Ecological protection | NC2 | Adequate attention is given to environmental protection during local tourism development.
Environmental practices | NC3 | I actively engage in environmentally responsible practices in my everyday production and household activities.
Psychological capital (PsyC) | Self-efficacy | PsyC1 | I am confident in my ability to adjust household labor and business arrangements in response to tourism employment opportunities or changes in the business environment. | 5-point Likert: 1 = strongly disagree; 5 = strongly agree | Luthans et al. (2007); Li et al. (2020); Su et al. (2022)
Hope | PsyC2 | I have positive expectations about my household's future development direction and prospects.
Resilience | PsyC3 | When facing the tourism off-season or employment fluctuations, I can persevere and actively seek new opportunities.
Optimism | PsyC4 | I am optimistic about the future development prospects of the local tourism industry.
Income outcomes (IL) | Objective income level | IL1 | Total annual household income in 2024, including wage income, agricultural and non-agricultural business income, property income, and transfer income. | Six ordered categories: ① < RMB 30,000; ② 30,000–<50,000; ③ 50,000–<100,000; ④ 100,000–<150,000; ⑤ 150,000–<250,000; ⑥ ≥250,000 | He et al. (2021); Su M. M. et al., (2019)
Income structure | IL2 | Tourism-related income as a share of total household income (wages, business income, dividends, rental income, etc.). | 0–100%
Income diversification | IL3 | Number of household income sources (agricultural production, off-farm employment, scenic-area employment, self-employment, cooperative dividends, etc.). | Count variable: 1–6 sources | Su Z. et al. (2019); Su et al. (2022)
Subjective income perceptions | IL4 | My household's income rarely experiences substantial fluctuations. | 5-point Likert: 1 = strongly disagree; 5 = strongly agree | Adapted to the local study context
IL5 | Compared with other households in the village, I consider my household's income level to be relatively high.
IL6 | Given our current livelihood arrangements, my household's income will not decline over the next 3 years. | He et al. (2021); Su et al. (2022)
IL7 | I am satisfied with the tourism industry's overall contribution to increasing my household income.
Control variables | Demographic and household characteristics | C1 | Gender | Male = 1; Female = 0 | –
C2 | Age | Continuous (years) | –
C3 | Educational attainment | Number of years of formal education | –
C4 | Number of permanent household residents | Continuous (number of persons) | –
C5 | Number of household laborers | Continuous (number of persons) | –
C6 | Years of tourism participation | Continuous (years) | –
C7 | Tourism activity type/primary participation mode | Categorical | –

Tourism participation

Tourism participation (TP) is operationalized as a composite index comprising three content dimensions: economic participation, participation breadth, and decision agency. Table 1 reports the corresponding items and their theoretical and measurement bases. We classified a household as a tourism participant if at least one household member engaged in a tourism-related activity during the previous 6 months. We measured all items on five-point Likert scales. We first calculated item means within each dimension, then z-standardized each dimension, and averaged the three standardized dimension scores with equal weights to form the composite tourism participation index. TP is an analytical composite rather than a first-order reflective latent variable; TP1–TP9 are therefore not assumed to reflect a single underlying trait. Households were also classified by their main participation mode; we used this classification only for sample description and robustness analyses and did not include it as a predictor in the main model.

Livelihood capital structure

Building on the Sustainable Livelihoods Framework and the rural tourism context, household livelihood capital structure comprises six dimensions: natural, physical/material, financial, human, social, and psychological capital. Table 1 reports the measurement indicators. Psychological capital comprises positive psychological resources including self-efficacy, hope, resilience, and optimism. It is treated as a psychological-resource state reported by households at the time of the survey, rather than as a stable personality trait or a resource necessarily generated by tourism participation. Livelihood capital is measured primarily in terms of household-level possession, perceived availability, and accessibility. Some physical and natural capital indicators concern shared village conditions, but the measures capture households’ assessments of these conditions and their household-level accessibility. The household is therefore the basic unit of analysis.

Household income outcomes

Household income outcomes (IL) are divided into objective income level, income structure and diversification, and subjective income perceptions. IL1 is the primary objective income outcome and refers to total pre-tax household income in 2024, including wage income, agricultural and non-agricultural business income, property income, and transfer income. Because rural households often lack precise income records, IL1 was measured using income intervals and treated as an ordered categorical variable. During the survey, respondents’ reports were cross-checked where possible against bankbooks or transaction records, business records, homestay bookings, and cooperative dividends. IL2 is the share of tourism-related income in total household income and reflects dependence on tourism in the income structure rather than income level itself. IL3 is the number of household income-source categories and reflects income diversification. IL4–IL7 represent perceived income stability, relative income position, income sustainability, and satisfaction with income improvement, respectively. Because objective income and subjective income perceptions have different measurement properties, we analyze them separately rather than combining them into a single income latent variable.

Control variables and missing-data treatment

Depending on model requirements, the analyses include gender, age, educational attainment, number of permanent household residents, and number of household laborers as control variables. Gender is coded as male = 1 and female = 0; the remaining variables are coded according to the survey. “Refused to answer” and “don’t know” responses are coded as NA. We treat “Not applicable” as missing rather than assigning a neutral or zero value. The main models use complete-case estimation, with the effective sample size determined separately for each model and reported in the results tables. All 374 valid questionnaires have complete TP items and IL1; only a small number of individual items in other measures contain missing values, and the overall proportion of missing observations is low.

Model specification and estimation

Ordered logit model for objective income level

Because IL1 is measured in six ordered income intervals, we treat it as an ordinal outcome. Let Y i denote the income category of household i , taking values 1, 2, …, 6. We specify a latent continuous variable as a linear function of the tourism participation index or dimension-specific indicators, the six livelihood capital dimensions, and the control variables, with an error term. We determine the observed ordered income category using threshold parameters. We report ordered-logit coefficients, odds ratios, and 95% confidence intervals and interpret them as statistical associations rather than causal effects.

Income structure, diversification, and subjective income perceptions

We analyze IL2–IL7 separately as supplementary income outcomes. We estimate IL2, the share of tourism income in total household income, using linear regression after standardizing tourism participation and the six livelihood capital measures; we report unstandardized coefficients and robust standard errors. We model IL3, the number of income-source categories, using negative binomial regression and report incidence-rate ratios (IRRs). We analyze IL4–IL7, representing income stability, relative income position, income sustainability, and satisfaction with income improvement, using linear regression. These outcomes represent distinct dimensions of household income and are not combined into a single income latent variable.

Supplementary PLS-SEM decomposition of statistical associations

Following the regression analyses, we used partial least squares structural equation modeling (PLS-SEM) as a supplementary approach to decompose the association structure among tourism participation, livelihood capital, and income outcomes. PLS-SEM suits exploratory and explanatory models with multiple constructs and association pathways. Given the cross-sectional design, we used PLS-SEM only to statistically decompose associations. We estimated the models in SmartPLS 4.0 using the path-weighting scheme, with a maximum of 300 iterations and a convergence criterion of 1 × 10–7. We performed bootstrap resampling 5,000 times and reported 95% confidence intervals. The measurement model assesses internal consistency, convergent validity, and discriminant validity, while the structural model examines explanatory power, predictive performance, and collinearity. Predictive assessment uses PLSpredict with 10-fold cross-validation repeated 10 times. The baseline model follows the theoretical structure; an adjusted model additionally includes gender, age, education, number of household laborers, years of tourism participation, and village dummy variables to assess the robustness of the core associations. We handle missing observations using complete-case analysis.

Results and analysis

Sample characteristics and descriptive statistics

Demographic characteristics of the sample

Table 2 presents the sample characteristics. Among the 374 valid observations, males accounted for 53.5% and females for 46.5%. Respondents aged 31–54 years accounted for 75.1% of the sample, indicating that the sample was predominantly composed of young and middle-aged adults. In terms of educational attainment, respondents with junior high school education or below and those with high school or secondary vocational education accounted for 34.8% and 34.0%, respectively. Annual household income was mainly concentrated in the RMB 150,000–<250,000 category, accounting for 33.4%, and by main tourism participation type, employment-oriented, business-oriented, supply-oriented, and equity-oriented participation accounted for 37.9%, 28.9%, 20.3%, and 12.9%, respectively. This classification is for descriptive purposes only and is not included as a predictor in the main models. The main participation type reported in the table represents each household's dominant participation type. It does not exclude the possibility that a household simultaneously engages in other forms of tourism participation.

Sample characteristics (N = 374).

Variable category | Subcategory | Frequency | Percentage
Gender | Male | 200 | 53.5%
Female | 174 | 46.5%
Age | ≤30 years | 37 | 9.9%
31–45 years | 163 | 43.5%
46–54 years | 118 | 31.6%
≥55 years | 56 | 15.0%
Educational attainment | Junior high school or below | 130 | 34.8%
High school or secondary vocational school | 127 | 34.0%
Junior college or bachelor’s degree | 100 | 26.7%
Master’s degree or above | 17 | 4.5%
Annual household income | < RMB 30,000 | 14 | 3.7%
RMB 30,000–<50,000 | 48 | 12.8%
RMB 50,000–<100,000 | 62 | 16.6%
RMB 100,000–<150,000 | 95 | 25.4%
RMB 150,000–<250,000 | 125 | 33.4%
≥ RMB 250,000 | 30 | 8.0%
Main tourism participation type | Employment-oriented | 142 | 37.9%
Business-oriented | 108 | 28.9%
Supply-oriented | 76 | 20.3%
Equity-oriented | 48 | 12.9%

Percentages may differ by 0.1 percentage point because of rounding.

Descriptive statistics of income outcomes

As shown in Table 3, the mean share of tourism-related income in total household income (IL2) was 45.6%, with a standard deviation of 24.3% and a median of 42.0%. The 40%–<60% group accounted for the largest proportion of households (28.1%). The mean number of household income sources (IL3) was 2.35 (SD = 0.92), and two income sources were most common, accounting for 35.3% of the sample. The mean values of IL4–IL7 were 3.92, 4.15, 3.88, and 4.21, respectively. We analyzed these indicators separately according to their measurement properties rather than combining them into a single latent income construct.

Descriptive distribution of income-related indicators (N = 374).

Indicator | Statistic | Value
IL2 | Mean | 45.6%
Standard deviation | 24.3
Median | 42.0%
0–<20% | 68 (18.2%)
20%–<40% | 92 (24.6%)
40%–<60% | 105 (28.1%)
60%–<80% | 72 (19.3%)
80%–100% | 37 (9.9%)
IL3 | Mean | 2.35
Standard deviation | 0.92
1 source | 86 (23.0%)
2 sources | 132 (35.3%)
3 sources | 101 (27.0%)
≥4 sources | 55 (14.7%)
IL4 | Mean ± SD | 3.92 ± 0.85
IL5 | Mean ± SD | 4.15 ± 0.78
IL6 | Mean ± SD | 3.88 ± 0.91
IL7 | Mean ± SD | 4.21 ± 0.75

IL3 has a minimum value of 1; no zero category was specified.

Measurement model assessment

Assessment of the tourism participation measure

Tourism participation (TP) consists of three dimensions: economic participation, participation breadth, and decision-making agency. The variance inflation factors (VIFs) among the three dimensions ranged from 1.18 to 1.46, all below 3.3, indicating no evident multicollinearity. TP was not directly specified as a first-order reflective latent variable. Instead, the composite index was calculated as described above.

Reliability and validity of the livelihood capital structure

Cronbach’s α values for the livelihood capital constructs ranged from 0.76 to 0.87, composite reliability (CR) values ranged from 0.80 to 0.91, and average variance extracted (AVE) values ranged from 0.65 to 0.77, all meeting commonly used criteria. All heterotrait–monotrait ratios (HTMT) were below 0.85, indicating satisfactory internal consistency, convergent validity, and discriminant validity (Table 4).

Reliability and convergent validity of the measurement model.

Latent construct | No. of items | Cronbach’s α | CR | AVE | HTMT
NC | 3 | 0.78 | 0.80 | 0.67 | <0.85
PC | 3 | 0.82 | 0.89 | 0.72 | <0.85
FC | 2 | 0.76 | 0.87 | 0.77 | <0.85
HC | 4 | 0.84 | 0.89 | 0.67 | <0.85
SC | 4 | 0.83 | 0.88 | 0.65 | <0.85
PsyC | 4 | 0.87 | 0.91 | 0.72 | <0.85

Common method bias and multicollinearity

Harman’s single-factor test showed that the first common factor explained 32.74% of the variance. In the unmeasured latent method construct (ULMC) test, the mean squared loading of the method factor was 0.031, and changes in individual item loadings after incorporating the method factor were all below 0.2. The TP–IL direct association changed by 0.005, while the indirect association between TP and IL through psychological capital changed by 0.003. These results did not indicate clear evidence that a single method factor dominated the measurements. The VIFs of the observed indicators ranged from 1.23 to 2.89, and the VIFs of the predictors in the structural model were within acceptable ranges, indicating no severe multicollinearity.

Direct association paths

Association between tourism participation and household total income

Because IL1 was measured using six ordered income intervals, we used an ordered logit model. Table 5 reports the statistical associations of tourism participation, livelihood capital, and control variables with the household annual total income category. After controlling for gender, age, years of education, household size, and the number of household laborers, the composite tourism participation index was significantly and positively associated with the household annual total income category (β = 0.182, SE = 0.042, OR = 1.200, 95% CI = [1.110, 1.300], p < 0.001). This indicates that, holding other variables constant, a one-unit increase in the tourism participation index was associated with approximately a 20.0% increase in the odds of a household being in a higher income category.

Ordered logit regression results for household total income.

Variable | β | SE | Wald | p | OR | 95% CI
TP | 0.182 | 0.042 | 18.77 | <0.001 | 1.200 | 1.110–1.300
NC | 0.075 | 0.028 | 7.18 | 0.007 | 1.078 | 1.022–1.140
PC | 0.088 | 0.031 | 8.06 | 0.005 | 1.092 | 1.029–1.161
FC | 0.219 | 0.039 | 31.54 | <0.001 | 1.245 | 1.155–1.344
HC | 0.114 | 0.036 | 10.03 | 0.002 | 1.121 | 1.046–1.204
SC | 0.097 | 0.033 | 8.64 | 0.003 | 1.102 | 1.035–1.176
PsyC | 0.247 | 0.036 | 47.08 | <0.001 | 1.280 | 1.195–1.374
Gender | −0.062 | 0.107 | 0.34 | 0.562 | 0.940 | 0.762–1.160
Age | 0.010 | 0.008 | 1.56 | 0.214 | 1.010 | 0.994–1.026
Years of education | 0.058 | 0.028 | 4.29 | 0.038 | 1.060 | 1.003–1.120
Household size | 0.030 | 0.036 | 0.69 | 0.401 | 1.030 | 0.960–1.106
Number of household laborers | 0.077 | 0.049 | 2.47 | 0.112 | 1.080 | 0.982–1.188

The dependent variable IL1 consists of six ordered income categories denotes the ordered logit regression coefficient, and OR denotes the odds ratio. The 95% CI is the 95% confidence interval for OR. The model controls for gender, age, years of education, household size, and the number of household laborers. The results represent statistical associations rather than causal effects.

Associations of tourism participation with income structure, diversification, and subjective income perceptions

IL2, IL3, and IL4–IL7 were further used as supplementary outcome variables. As shown in Table 6, tourism participation was significantly and positively associated with the share of tourism-related income (IL2), with B = 0.156 (95% CI: 0.082–0.230, p < 0.001). It was also significantly and positively associated with the number of income sources (IL3), with an incidence rate ratio (IRR) of 1.12 (95% CI: 1.064–1.179, p < 0.001). In addition, the coefficients of tourism participation for income stability, relative income level, income sustainability, and satisfaction with income improvement were 0.132, 0.118, 0.145, and 0.127, respectively, all significant at p < 0.01.

Associations with income structure, diversification, and subjective income perceptions.

Outcome | Model | Focal variable | Estimate | SE | 95% CI | p | N
IL2 | Linear regression | TP | 0.156 | 0.038 | 0.082–0.230 | <0.001 | 374
FC | 0.118 | 0.035 | 0.050–0.186 | 0.001 | 374
PsyC | 0.104 | 0.033 | 0.040–0.168 | 0.002 | 374
IL3 | Negative binomial regression | TP | IRR = 1.12 | 0.029 | 1.064–1.179 | <0.001 | 374
Negative binomial regression | FC | IRR = 1.08 | 0.033 | 1.017–1.147 | 0.012 | 374
Negative binomial regression | PsyC | IRR = 1.07 | 0.031 | 1.012–1.132 | 0.018 | 374
IL4 | Linear regression | TP | 0.132 | 0.041 | 0.052–0.212 | 0.002 | 374
IL5 | 0.118 | 0.039 | 0.042–0.194 | 0.003 | 374
IL6 | 0.145 | 0.043 | 0.061–0.229 | 0.001 | 374
IL7 | 0.127 | 0.040 | 0.049–0.205 | 0.002 | 374

Association analysis of the livelihood capital structure

Direct association paths

Building on the preceding ordered logit analysis, this study further employed partial least squares structural equation modeling (PLS-SEM) to statistically decompose the direct association structure among tourism participation, livelihood capital structure, and household total income. As shown in Table 7, the composite tourism participation index was significantly and positively associated with all six dimensions of livelihood capital. The path coefficient from TP to PsyC was 0.675, followed by 0.649 for TP → FC, 0.631 for TP → PC, 0.622 for TP → HC, 0.593 for TP → SC, and 0.561 for TP → NC; all were statistically significant at p < 0.001. The six livelihood capital dimensions were also significantly and positively associated with IL1. The path coefficients were 0.247 for PsyC → IL1, 0.219 for FC → IL1, 0.114 for HC → IL1, 0.097 for SC → IL1, 0.088 for PC → IL1, and 0.075 for NC → IL1, all statistically significant at p < 0.01. In addition, the direct path from TP to IL1 was 0.182 (p < 0.001), indicating a significant direct statistical association between tourism participation and the income outcome. H1, H2a–H2f, and H3a–H3f were supported.

Direct association path coefficients in the PLS-SEM structural model.

Path | Hypothesis | Standardized path coefficient β | SE | t | p | 95% CI
TP → IL1 | H1 | 0.182 | 0.042 | 4.327 | <0.001 | [0.104, 0.265]
TP → HC | H2a | 0.622 | 0.038 | 16.368 | <0.001 | [0.549, 0.696]
TP → SC | H2b | 0.593 | 0.041 | 14.463 | <0.001 | [0.514, 0.672]
TP → FC | H2c | 0.649 | 0.035 | 18.543 | <0.001 | [0.581, 0.717]
TP → PC | H2d | 0.631 | 0.039 | 16.179 | <0.001 | [0.556, 0.706]
TP → NC | H2e | 0.561 | 0.044 | 12.750 | <0.001 | [0.476, 0.646]
TP → PsyC | H2f | 0.675 | 0.032 | 21.094 | <0.001 | [0.613, 0.737]
HC → IL1 | H3a | 0.114 | 0.036 | 3.167 | 0.002 | [0.045, 0.186]
SC → IL1 | H3b | 0.097 | 0.033 | 2.939 | 0.003 | [0.034, 0.162]
FC → IL1 | H3c | 0.219 | 0.039 | 5.674 | <0.001 | [0.144, 0.296]
PC → IL1 | H3d | 0.088 | 0.031 | 2.839 | 0.005 | [0.029, 0.149]
NC → IL1 | H3e | 0.075 | 0.028 | 2.679 | 0.007 | [0.022, 0.131]
PsyC → IL1 | H3f | 0.247 | 0.036 | 6.892 | <0.001 | [0.178, 0.318]

Table 5 and this table use different estimation methods. Table 5 reports ordered logit regression coefficients, whereas this table reports standardized PLS-SEM path coefficients; do not compare the two sets of coefficients numerically. Interpret all paths as statistical associations rather than causal effects. Bootstrap resampling was conducted 5,000 times.

Indirect association paths

The product-of-coefficients approach, based on PLS-SEM path coefficients, was further used to decompose the indirect statistical association between tourism participation and household annual total income through the six dimensions of livelihood capital. As shown in Table 8, the indirect association for TP → PsyC → IL1 was approximately 0.167, followed by 0.142 for TP → FC → IL1, 0.071 for TP → HC → IL1, 0.058 for TP → SC → IL1, 0.056 for TP → PC → IL1, and 0.042 for TP → NC → IL1. The sum of the six path products was approximately 0.536, consistent with the total indirect association reported in the original model. Among the estimated statistical indirect associations, the paths involving psychological capital and financial capital had relatively larger numerical values, accounting for 31.2% and 26.5% of the total indirect association, respectively. This finding indicates that, within the association structure specified in this study, these two capital dimensions showed relatively stronger statistical associations. H4 was supported.

Decomposition of indirect association paths of livelihood capital between tourism participation and household total income.

Indirect association path | Indirect association | SE | T | p | 95% CI | Share of indirect association (%)
TP → PsyC → IL1 | 0.167 | 0.030 | 5.567 | <0.001 | [0.112, 0.228] | 31.2%
TP → FC → IL1 | 0.142 | 0.028 | 5.071 | <0.001 | [0.091, 0.199] | 26.5%
TP → HC → IL1 | 0.071 | 0.024 | 2.958 | 0.003 | [0.027, 0.121] | 13.2%
TP → SC → IL1 | 0.058 | 0.021 | 2.762 | 0.006 | [0.019, 0.102] | 10.8%
TP → PC → IL1 | 0.056 | 0.020 | 2.800 | 0.005 | [0.020, 0.099] | 10.4%
TP → NC → IL1 | 0.042 | 0.017 | 2.471 | 0.013 | [0.012, 0.080] | 7.8%
Total indirect paths | 0.536 | 0.034 | 15.765 | <0.001 | [0.472, 0.602] | 100%
Direct association: TP → IL1 | 0.182 | 0.044 | 4.327 | <0.001 | [0.104, 0.265] | –
Total path | 0.718 | 0.035 | 20.514 | <0.001 | [0.652, 0.781] | –

Comparison of indirect association paths

To determine whether differences among the indirect association paths involving the six capital dimensions were statistically significant, we conducted pairwise comparisons using bootstrap resampling with 5,000 repetitions. As shown in Table 9, the difference between the indirect paths TP → PsyC → IL1 and TP → FC → IL1 was 0.025, with a 95% CI of [−0.044, 0.092], which was not statistically significant (p = 0.482). Thus, although the indirect association for psychological capital was numerically slightly larger than that for financial capital, the difference was not statistically significant. By contrast, the differences between psychological capital and human capital (difference = 0.096, p = 0.012), social capital (difference = 0.109, p = 0.008), physical/material capital (difference = 0.111, p = 0.006), and natural capital (difference = 0.125, p = 0.001) were statistically significant. The differences between financial capital and each of these four capital dimensions were also statistically significant (p < 0.05). However, most pairwise comparisons among human, social, physical/material, and natural capital were not statistically significant (p ≥ 0.386), indicating that although these four capital dimensions differed numerically in their indirect associations, their differences could not generally be statistically distinguished from one another.

Bootstrap tests of differences in indirect association paths.

Path comparison | Difference in indirect association | 95% CI lower | 95% CI upper | p
PsyC–FC | 0.025 | −0.044 | 0.092 | 0.482
PsyC–HC | 0.096 | 0.028 | 0.168 | 0.012*
PsyC–SC | 0.109 | 0.039 | 0.184 | 0.008**
PsyC–PC | 0.111 | 0.040 | 0.186 | 0.006**
PsyC–NC | 0.125 | 0.058 | 0.196 | 0.001***
FC–HC | 0.071 | 0.005 | 0.142 | 0.041*
FC–SC | 0.084 | 0.018 | 0.156 | 0.024*
FC–PC | 0.086 | 0.020 | 0.158 | 0.020*
FC–NC | 0.100 | 0.034 | 0.172 | 0.008**
HC–SC | 0.013 | −0.052 | 0.078 | 0.682
HC–PC | 0.015 | −0.048 | 0.080 | 0.624
HC–NC | 0.029 | −0.038 | 0.098 | 0.386
SC–PC | 0.002 | −0.052 | 0.056 | 0.944
SC–NC | 0.016 | −0.039 | 0.074 | 0.548
PC–NC | 0.014 | −0.042 | 0.072 | 0.612

***p < 0.001, **p < 0.010, *p < 0.050. Bootstrap resampling was conducted 5,000 times.

Taken together, the indirect association values corresponding to PsyC and FC were relatively large. However, the difference between the two paths was not statistically significant (difference = 0.025, 95% CI = [−0.044, 0.092], p = 0.482). Therefore, the path involving PsyC is not statistically significantly stronger than the path involving FC. The indirect association values for the other four capital dimensions were relatively lower, and most differences among them were not statistically significant.

Robustness checks

This study conducted robustness and sensitivity analyses on item overlap, random permutation, sample composition, income coding, and control-variable adjustment (Table 10). First, after removing potentially overlapping measurement items, the direct association coefficient between TP and IL1 decreased from 0.182 to 0.155 (95% CI = [0.078, 0.232], p < 0.001), while the direction and statistical significance remained unchanged. This suggests that the core association was not entirely attributable to overlap in measurement content.

Robustness and sensitivity analysis results.

Test | Model | Outcome and coding | Core association | Estimator | Estimate | 95% CI | p | N | Comparison with baseline
De-overlap model | PLS-SEM | IL1 (1–6) | TP → IL1 direct association | Standardized path coefficient (β) | 0.155 | [0.078, 0.232] | <0.001 | 374 | Decreased from 0.182 to 0.155
Random permutation | Random permutation | IL1 (1–6) | TP → IL1 total association | Mean of permutation distribution | 0.024 | — | <0.001 | 374 | Lower than baseline total association of 0.718
Excluding short-term participants | PLS-SEM | IL1 (1–6) | TP → IL1 total association | Total association coefficient | 0.706 | [0.640, 0.772] | <0.001 | 336 | Approximately 1.7% change from baseline 0.718
Excluding self-employed households | PLS-SEM | IL1 (1–6) | TP → IL1 direct association | Standardized path coefficient (β) | 0.166 | [0.088, 0.244] | <0.01 | 322 | Approximately 8.8% decrease from baseline 0.182
Income-coding sensitivity analysis | Linear regression | Ordinal coding of IL1 | TP → IL1 | Unstandardized coefficient (B) | 0.172 | [0.110, 0.234] | <0.001 | 374 | Alternative coding analysis
Highest income category = RMB 250,000 | Linear regression | Numeric IL1 | TP → IL1 | Unstandardized coefficient (B) | 0.175 | [0.113, 0.237] | <0.001 | 374 | Same direction
Highest income category = RMB 300,000 | Linear regression | Numeric IL1 | TP → IL1 | Unstandardized coefficient (B) | 0.178 | [0.116, 0.240] | <0.001 | 374 | Same direction
Highest income category = RMB 500,000 | Linear regression | Numeric IL1 | TP → IL1 | Unstandardized coefficient (B) | 0.181 | [0.119, 0.243] | <0.001 | 374 | Same direction
Control-variable model | PLS-SEM | IL1 (1–6) | TP → IL1 total association | Total association coefficient | 0.174 | [0.096, 0.252] | <0.001 | 374 | Core direction and significance remain consistent

As all p-values are reported directly in the dedicated “p” column (e.g., <0.001 and <0.01), The asterisk (*) previously appearing in the table footnote has been removed.

Second, a random permutation test with 5,000 permutations yielded a mean of 0.024 and a standard deviation of 0.048 for the permutation distribution of the total TP–IL association (p < 0.001). This test targets the null-association scenario under random matching and cannot rule out alternative explanations from omitted variables, sample-selection bias, spatial factors, or other sources. Moreover, we did not further stratify the permutation procedure by village; therefore, we treat this result as a supplementary diagnostic rather than evidence of causal identification.

Third, to test the sensitivity of the core results to changes in sample composition, we excluded the 38 households with less than 1 year of tourism participation. Among the remaining 336 households, the total association between TP and IL1 was 0.706 (95% CI = [0.640, 0.772], p < 0.001), representing a change of approximately 1.7% relative to the baseline total association of 0.718. In addition, after excluding 52 self-employed households, the direct association between TP and IL1 was 0.166 (95% CI = [0.088, 0.244], p < 0.01). Because this latter analysis changes the study sample composition, we use its results primarily to assess the sensitivity of the core association to different participation-type compositions rather than to replace the baseline model.

Regarding income measurement, this study further examined the results’ sensitivity to alternative income coding schemes. Because IL1 consists of six ordered income intervals, the baseline analysis employed an ordered logit model. In a supplementary analysis, we treated the ordinal coding of IL1 as approximately continuous and estimated linear regression models under alternative assignments for the highest income category. When we assigned the highest income category values of RMB 250,000, RMB 300,000, and RMB 500,000, respectively, the estimates for TP → IL1 were 0.175, 0.178, and 0.181, all with p < 0.001. These results were consistent in direction with the baseline analysis. Because this procedure changes both the treatment of IL1 and the estimation model, it is not statistically equivalent to the baseline ordered logit model. It should therefore be regarded as an income-coding sensitivity analysis rather than a robustness test fully equivalent to the baseline specification. Its purpose is solely to examine whether the direction of the core association is sensitive to the numerical treatment of the income intervals, rather than to replace the baseline ordered logit results.

Finally, after further incorporating control variables, the total association between TP and IL1 was 0.174 (95% CI = [0.096, 0.252], p < 0.001), with the direction and statistical significance of the core association remaining stable.

Discussion

The findings show that, among farmers who already participate in tourism, the degree of tourism participation is significantly and positively associated with income outcomes, consistent with previous findings (Su M.M. et al., 2019; He et al., 2021). Given the cross-sectional design and limited sample coverage, this result reflects a contemporaneous association among participating households and does not rule out unobserved factors. It should therefore not be interpreted as evidence that tourism participation necessarily increases household income. At the same time, tourism participation was significantly and positively associated with all six dimensions of livelihood capital, suggesting that tourism participation among households in Yunqiu Mountain is not merely a single economic activity but is closely connected with multiple aspects of livelihood activities (Su Z. et al., 2019; Yu et al., 2020). Farmers with different levels of participation have different capital endowments and access to resources. These associations may be bidirectional; therefore, tourism participation cannot be identified as the sole source or determinant of changes in livelihood capital.

Psychological capital was positively associated with both tourism participation and income outcomes, consistent with previous studies (Luthans et al., 2007; Li et al., 2022; Su et al., 2022). The psychological capital measure used in this study encompasses self-efficacy, hope, resilience, and optimism, thereby providing a supplementary perspective for extending the conventional livelihood capital framework. However, these associations cannot be directly interpreted as indicating that psychological capital “causes” an increase in income. Psychological capital may operate jointly with education, household resources, business experience, and other factors. The present study therefore identifies psychological capital as an important analytical dimension in tourism livelihood research rather than attributing an independent causal effect to it. The indirect association analysis showed relatively larger numerical values for psychological and financial capital, but the Bootstrap pairwise comparison indicated that the difference between them was not statistically significant. Accordingly, psychological capital is not a statistically superior pathway to financial capital, avoiding an unsupported ranking of pathways.

All six dimensions of livelihood capital were significantly and positively associated with the household annual total income category, with financial capital showing a relatively larger association coefficient. This may reflect the need for financial investment and cash-flow support in tourism businesses and non-farm livelihood activities (Su Z. et al., 2019; Yu et al., 2020). However, the cross-sectional design does not allow the causal direction between the two to be determined, and the possibility of reverse association cannot be excluded, whereby higher-income households may have greater capacity to accumulate financial resources. Human and social capital were also significantly and positively associated with income outcomes, consistent with previous findings (Yang et al., 2020; Han et al., 2025). These associations may likewise be affected by differences in households' pre-existing resources and participation modes. Although natural capital had a relatively smaller association coefficient, its value as a foundation for agricultural production and a core tourism resource remains relevant. Multiple factors shape the conversion of these resources into household income, including resource property rights, utilization patterns, participation channels, and benefit-sharing mechanisms. Therefore, do not infer the relative importance of different forms of capital solely from model coefficients.

The PLS-SEM decomposition of indirect associations indicates that all six dimensions of livelihood capital correspond to significant indirect pathways. The indirect associations involving psychological and financial capital were relatively large, whereas those involving the other four capital dimensions were smaller. Pairwise comparisons showed that the difference between psychological and financial capital was not statistically significant, while most differences among the latter four capital dimensions were also statistically insignificant. These findings indicate that the indirect association strengths of different capital dimensions differ numerically but do not establish a clear and stable priority order. The association between tourism participation and income outcomes therefore links to multiple dimensions of livelihood capital and should not be explained by a single resource endowment.

Rural tourism can expand livelihood opportunities but may also bring seasonality, market fluctuations, and adjustments in resource allocation. Household labor allocation among tourism activities may vary across peak and off-peak seasons, and households may reallocate resources among agriculture, tourism employment, and self-employment (Su et al., 2022). Thus, higher tourism participation does not necessarily mean a complete withdrawal from traditional agricultural livelihoods; instead, it may involve recombining multiple livelihood activities. For business activities requiring relatively high initial investment, financing needs and market uncertainty may also increase household business risks (Cui et al., 2025; Kan and Sun, 2024). Therefore, when interpreting the association between tourism participation and income outcomes, consider both income opportunities and potential risks. Income indicators alone are insufficient to determine the effects of tourism development on overall household welfare. This study's discussion is limited to the sample of tourism-participating households in Yunqiu Mountain. The findings demonstrate statistical associations among different levels of tourism participation, livelihood capital, and income outcomes within this sample. However, they cannot be used to infer that tourism participation has the same implications for all households or to generalize directly to non-participating households or other regions.

This study has several limitations. First, it uses a cross-sectional survey design, with all variables measured at the same time; therefore, it cannot establish temporal ordering or causal effects. Future research could employ longitudinal surveys or panel data to examine dynamic patterns. Second, the sample includes only households already participating in tourism and excludes non-participants, households that have exited tourism activities, and households whose tourism businesses have failed. Consequently, the selection process underlying tourism participation cannot be fully examined. Future research could expand the sample to address this issue. Third, the livelihood outcomes considered in this study focus primarily on income and livelihood capital indicators and do not adequately cover other dimensions of household risk and welfare, such as debt, land rights, and consumption welfare. Future research could construct a more comprehensive indicator system and combine quantitative and qualitative evidence to examine the dynamic processes of livelihood adjustment.

Conclusions and implications

Conclusion

This study focuses on farmers in Yunqiu Mountain, Shanxi Province, in the Yellow River Basin who had already participated in tourism-related activities. Based on 374 valid questionnaires, ordered logit models, regression analyses, and PLS-SEM were employed to examine the statistical associations among tourism participation, livelihood capital, and household income outcomes. The main conclusions are as follows.

First, the composite tourism participation index was significantly and positively associated with the household annual total income category. Tourism participation was also positively associated with the share of tourism-related income, the number of income sources, and indicators of income stability, relative income advantage, income sustainability, and satisfaction with income improvement.

Second, the composite tourism participation index was significantly and positively associated with natural capital, physical/material capital, financial capital, human capital, social capital, and psychological capital. Among these dimensions, the standardized path coefficients linking tourism participation with psychological capital and financial capital were numerically relatively large.

Third, all six dimensions of livelihood capital were significantly and positively associated with household income outcomes. The coefficients corresponding to psychological and financial capital were numerically relatively large; however, pairwise Bootstrap comparisons did not identify a statistically significant difference in their indirect associations. Therefore, this study does not establish a definitive ranking of the six livelihood capital dimensions by strength. Instead, they show differentiated statistical associations with household income outcomes.

Finally, tourism participation was associated with household annual total income through significant statistical indirect pathways involving natural capital, physical/material capital, financial capital, human capital, social capital, and psychological capital. The indirect association values for psychological and financial capital were numerically large, but the difference between them was not statistically significant.

Management implications

Based on the above findings, this study proposes the following management implications.

First, support for multidimensional livelihood capital should be strengthened, with attention to coordinating psychological capital with physical/material capital. For households with limited skills and access to information, strengthen training in tourism services, business management, and digital marketing. For households with limited financial resources, financial services may be improved under conditions of manageable risk and voluntary participation; for households with limited access to cooperative resources, cooperative organizations, information platforms, and community-based collaboration can broaden access to resources. Experience sharing, demonstration-based learning, and entrepreneurship guidance can provide development-related information and capacity support; however, interventions targeting psychological capital should not be regarded as a deterministic means of increasing income.

Second, diversified tourism participation and benefit-sharing mechanisms should be improved. Participation channels in employment, business operation, and product supply can be expanded according to households' resource conditions. Cooperative organizations, benefit-distribution mechanisms, and information disclosure systems should be strengthened to safeguard households' rights to information and participation in cooperative arrangements. For households with relatively weak access to resources, attention should be given to both their opportunities to participate and the conditions under which they can obtain benefits, thereby reducing participation disparities across groups.

Third, promote greater coordination between ecological protection and household livelihood opportunities. Tourism development is closely connected with natural landscapes, ecological environments, and rural cultural resources. In ecologically sensitive areas, explore livelihood opportunities that integrate ecological protection, cultural-resource utilization, and tourism services, encouraging farmers to participate in ecological maintenance, green business activities, and cultural presentation. At the same time, strengthen benefit-sharing mechanisms that link tourism revenues to community public services and ecological protection to promote coordination between resource use and ecological conservation.

Fourth, strengthen the capacity of tourism-participating households to respond to livelihood risks. Seasonal fluctuations and market uncertainty characterize rural tourism, and different participation modes involve different types of business risks. Accordingly, develop differentiated measures for employment-oriented, business-oriented, supply-oriented, and asset-based participation, including employment safeguards, business guidance, supply-chain coordination, and standardized contractual and benefit-sharing arrangements. Diversified employment, information services, and necessary risk-protection mechanisms can also reduce households' dependence on a single tourism-related income source. For business projects involving relatively large financial investments, pay closer attention to business-risk warnings and contractual standardization, so tourism participation is not simply equated with stable income growth.

References

References

AbdallaF. H. SamahA. A. HashimA. H. RosnonM. R. (2022). Relationship between socio-economic factors and participation in decision making in tourism industry among women in Zanzibar. Econ. Bus. Q. Rev. 5, 124–132. doi: 10.31014/aior.1992.05.02.419

BebbingtonA. (1999). Capitals and capabilities: a framework for analyzing peasant viability, rural livelihoods and poverty. World Dev. 27, 2021–2044. doi: 10.1016/S0305-750X(99)00104-7

ChenH.-Z. WangW.-L. (2023). Influence of sustainable livelihood capital on farmers' participation behavior in rural tourism. J. Yunnan Agric. Univ. (Soc. Sci.) 17, 88–95. doi: 10.12371/j.ynau(s).202210047

ChenJ. ZhangL. YangX. LiG. (2017). Impact of rural tourism development on farmers' livelihoods and community tourism effects: a case study from the perspective of tourism development models. Geogr. Res. 36, 1709–1724.

CuiH. ZhengL. WangY. (2025). The impact of changes in land transfer decisions on rural livelihood transitions: evidence from dynamic panel data in China. Appl. Geogr. 176:103515. doi: 10.1016/j.apgeog.2025.103515

DaiB. (2026). Promoting the integrated development of agriculture, culture and tourism to create a new space for rural better living. Newsl. About Work. Rural. Areas 9, 35–37. doi: 10.3969/j.issn.0546-9503.2026.09.014

DFID (1999). Sustainable Livelihoods Guidance Sheets. London: Department for International Development.

GaoJ. WuB. (2017). Revitalizing traditional villages through rural tourism: a case study of Yuanjia Village, Shaanxi Province, China. Tour. Manag. 63, 223–233. doi: 10.1016/j.tourman.2017.04.003

HanW. ZhangZ. ZhangX. (2025). Farmland rental participation, livelihood capital, and rural income inequality: evidence from rural China. Appl. Econ. 57, 6346–6366. doi: 10.1080/00036846.2024.2383797

HeX. WangJ. GaoX. WangY. ChoiB. R. (2021). Rural tourism: does it matter for sustainable farmers' income? Sustainability 13:10440. doi: 10.3390/su131810440

KanK. SunJ. (2024). Dispossession and beyond: politics of rural land conversion in China's tourism villages. J. Rural. Stud. 112:103429. doi: 10.1016/j.jrurstud.2024.103429

LiH. QiN. LiZ. MaW. (2022). The relationship between farmers' entrepreneurial behavior and macroeconomics based on the probit regression model and entrepreneurial psychological capital. Front. Psychol. 13:954874. doi: 10.3389/fpsyg.2022.954874, 36211850.

LiW. ShuaiC. ShuaiY. ChengX. LiuY. HuangF. (2020). How livelihood assets contribute to sustainable development of smallholder farmers. J. Int. Dev. 32, 408–429. doi: 10.1002/jid.3461

LiuX. ZhangX. YangH. JiaoX. ChenP. HuY. (2025). Spatial-temporal characteristics and coupled coordination mechanism of sustainable agriculture and rural development of resource-based areas in the Yellow River Basin: an empirical analysis based on 42 prefecture-level cities. J. Northwest Univ. (Nat. Sci. Ed.) 55, 726–739. doi: 10.16152/j.cnki.xdxbzr.2025-04-004

LuthansF. AveyJ. B. AvolioB. J. PetersonS. J. (2010). The development and resulting performance impact of positive psychological capital. Human Resour. Dev. Q. 21, 41–67. doi: 10.1002/hrdq.20034

LuthansF. AvolioB. J. AveyJ. B. NormanS. M. (2007). Positive psychological capital: measurement and relationship with performance and satisfaction. Pers. Psychol. 60, 541–572. doi: 10.1111/j.1744-6570.2007.00083.x

LuthansF. Youssef-MorganC. M. (2017). Psychological capital: an evidence-based positive approach. Annu. Rev. Organ. Psychol. Organ. Behav. 4, 339–366. doi: 10.1146/annurev-orgpsych-032516-113324

MaX. WangR. DaiM. OuY. (2021). The influence of culture on the sustainable livelihoods of households in rural tourism destinations. J. Sustain. Tour. 29, 1235–1252. doi: 10.1080/09669582.2020.1826497

Mohammadi-MehrS. HayatiD. KaramiE. (2025). Psychological capital and farmer families' quality of life under the climate variability conditions. Sci. Rep. 15:5320. doi: 10.1038/s41598-024-78750-3, 39939617.

SuZ. AaronJ. R. GuanY. WangH. (2019). Sustainable livelihood capital and strategy in rural tourism households: a seasonality perspective. Sustainability 11:4833. doi: 10.3390/su11184833

SuM. M. WallG. WangY. JinM. (2019). Livelihood sustainability in a rural tourism destination—Hetu town, Anhui province, China. Tour. Manag. 71, 272–281. doi: 10.1016/j.tourman.2018.10.019

SuZ. WenR. ZengY. YeK. KhotphatT. (2022). The influence of seasonality on the sustainability of livelihoods of households in rural tourism destinations. Sustainability 14:10572. doi: 10.3390/su141710572

SunJ. WangS. (2023). Deconstruction of the path of rural revitalization promoting common prosperity based on hypercycle theory. J. Nat. Resour. 38, 1955–1967. doi: 10.31497/zrzyxb.20230803

WangY. LiuY. (2023). Models and classification of rural revitalization in China: from the perspective of lucid waters and lush mountains. Geogr. Res. 42, 2005–2017. doi: 10.11821/dlyj020230093

YangF. JiangY. ZengW. (2020). Livelihood capitals on income inequality among rural households: Evidence from China. Ciência Rural, 50:e20190461. doi: 10.1590/0103-8478cr20190461

YangY. AlamM. KunasekaranP. BalasubramaniamK. RosnonM. R. (2025). The influencing factors of sustainable agriculture tourism in the farmhouses of Hainan, China. Pertanika J. Soc. Sci. Humanit. 33, 55–72. doi: 10.47836/pjssh.33.S1.04

YiX. (2022). Difference of farmers' livelihood capital before and after rural tourism development. Comput. Intell. Neurosci. 2022:4138220. doi: 10.1155/2022/4138220, 35387244.

YuP. ZhangJ. WangY. WangC. ZhangH. (2020). Can tourism development enhance livelihood capitals of rural households? Evidence from Huangshan National Park adjacent communities, China. Sci. Total Environ. 748:141099. doi: 10.1016/j.scitotenv.2020.141099, 32814283.

ZhangH. DuanY. HanZ. (2021). Research on spatial patterns and sustainable development of rural tourism destinations in the Yellow River Basin of China. Land 10:849. doi: 10.3390/land10080849

ZhangJ. FanZ. LiuJ. AhmadF. CaoZ. (2025). Livelihood capital, risk response, and rural household poverty vulnerability: an empirical experience based in China. Rev. Dev. Econ. 29, 1693–1711. doi: 10.1111/rode.13184