The sustainable livelihood framework developed by DFID consists of five components: vulnerability context, livelihood capital, structural and process transformation, livelihood strategies, and livelihood outcomes (Yang et al., 2023; Zhang et al., 2024). In this framework, relocated households are seen as individuals making a living under the influence of vulnerability, using their livelihood capital to achieve survival and development. They are also subject to external policy support or constraints. Under the combined influence of these factors, relocated households choose different livelihood strategies to achieve optimal livelihood outcomes (Shang et al., 2023; Wu et al., 2022). The sustainable livelihoods theory integrates resources and capabilities from various domains—natural, material, human, social, and financial—and serves as the foundation for farmers to maintain and improve their livelihoods (Su et al., 2021; Zhang et al., 2024). Social integration of relocated households is a family-based migration behavior in which livelihood capital serves as a key resource. Differences in livelihood capital will inevitably affect the ability of relocated households to integrate into their new environments (Wang and Gao, 2022). On the one hand, livelihood capital provides economic stability and a sense of security, enabling relocated households to maintain a stable livelihood in their new environment. This economic stability is a prerequisite for social integration and reduces the risk of social exclusion and marginalization due to economic hardship (Mallick et al., 2020). Furthermore, a stable economic situation allows farmers to dedicate more time and energy to participating in community activities and establishing social relationships, thus enhancing social integration. On the other hand, livelihood capital enables relocated households to access information and resources, and through resource sharing, they are able to build a broader network of social relationships and obtain more social support, further promoting their integration into society (Azizi et al., 2017). In summary, the hypothesis is proposed:

According to sustainable livelihood theory, the livelihood process of relocated households is often accompanied by uncertainties, some of which stem from external factors such as environmental and social risks, while others arise from the individuals themselves, such as health risks. These risks are collectively referred to as livelihood risks (Bouahom et al., 2004; Ma and Zhao, 2022). Once a household experiences a risk shock, it can severely affect the living standards of its members, even causing some relocated households to fall back into poverty overnight, thereby constraining their social integration (Vivekananda et al., 2014). Sustainable livelihood theory posits that the resources and endowments held by farmers, i.e., livelihood capital, influence and represent their ability to sustain their livelihoods, form the foundation of the livelihood process, and serve as the basis for resisting risk shocks and maintaining livelihood stability (Su et al., 2018). Different forms of livelihood capital have complex relationships with various livelihood risks, and the extent of livelihood capital affects the severity of these risks. Generally, the more abundant a household’s livelihood capital, the more strategies it has to cope with livelihood risks (Siegel, 2005). The accumulation of livelihood capital in relocated households can mitigate and prevent livelihood risks, which might otherwise negatively impact social integration by affecting psychological wellbeing, access to resources, social capital, and adaptive capacities.

Livelihood risks are multidimensional, and this paper analyzes the role of livelihood risks in the process of social integration as influenced by livelihood capital, focusing on environmental, employment, and health risks. First, environmental risks refer to unfavorable factors and uncertainties that relocated households may encounter in their new environment, such as natural disasters, environmental pollution, and poor housing conditions (Su, 2017). Livelihood capital can impact environmental risks in multiple ways: the higher the livelihood capital, the stronger the economic and resource acquisition capabilities of relocated households, allowing them to improve living conditions and mitigate environmental risks (Su et al., 2019). Environmental risks can lead to health problems, economic losses, and psychological stress, which can hinder social interactions and participation, thus affecting social integration.

Secondly, employment risk refers to the unstable employment, uncertainty and insufficient employment opportunities that farmers face in the new urban environment (Kuang et al., 2020). On the one hand, livelihood capital enhances the adaptive capacity of relocated households, enabling them to cope with and mitigate employment risks more effectively (Zeng et al., 2021a). On the other hand, employment risks may lead to unstable economic income and restricted career development, limiting relocated households’ economic capital accumulation and career development, which in turn affects their survival and development in towns and cities.

Finally, health risks refer to the health problems and medical challenges that relocated households may face in the new environment, which may include, but are not limited to, an increased incidence of diseases, lack of medical resources, insufficient health protection, and health problems brought about by changes in lifestyles (Zeng et al., 2021b). The accumulation of livelihood capital is conducive to increasing the health awareness of relocated households, enabling them to prevent and cope with health problems more effectively, and relocated households with higher financial capital are able to pay for healthcare and receive timely healthcare services (Zeng et al., 2021a); at the same time, health risks may lead to a decline in relocated households’ labor capacity and a reduction in employment opportunities, limiting relocated households’ capital accumulation and career development, which in turn affects their survival and development in towns and cities. In summary, the following hypotheses are proposed (Figure 1):

The data used in this study are derived from field research conducted in Liangshan Prefecture, Sichuan Province, China, between January and March 2024. To ensure representative coverage of resettlement patterns, we employed a stratified purposive sampling method to select five counties from Liangshan Prefecture’s 17 administrative units. The selection criteria prioritized counties with resettlement rates exceeding 15%, aiming to capture regions where demographic shifts most significantly impacted socioeconomic dynamics. This approach balanced geographic diversity and data richness while aligning with the study’s focus on high-impact resettlement areas. The final sample included Ganluo, Butuo, Yuexi, Meigu, and Zhaojue counties. Next, a combination of stratified and random sampling methods was applied to account for the differences in economic development levels across various regions, as well as the varying characteristics of relocated farmers. This approach aimed to select a diverse range of relocated farmer samples to ensure that the survey results would more closely reflect reality. In each county, 1–3 resettlement areas were randomly selected, with 50–80 households chosen from each area. A total of 661 questionnaires were collected, and after excluding invalid responses, 610 valid questionnaires were retained, yielding a validity rate of 92%.

Table 4 presents the results of estimating the impact of livelihood capital on the social integration of relocated households using both the OLS model and the quantile regression model. According to the OLS regression results, livelihood capital passes the 1% significance level test, with a positive coefficient, indicating that livelihood capital significantly increases the social integration level of relocated households. Specifically, for each one-unit increase in livelihood capital, the social integration level of relocated households rises by 55.32%. This confirms Hypothesis 1.

The quantile regression results also show that livelihood capital significantly improves the social integration of relocated households, regardless of the quantile point used for the sample regression. At the 25th, 50th, 75th, and 95th quantiles, the effects of livelihood capital on the social integration of relocated households are 0.538, 0.421, 0.408, and 0.346, respectively. These findings suggest that the effect of livelihood capital on social integration diminishes as the level of social integration increases. The likely reason is that relocated households with lower social integration levels have a more urgent need for livelihood capital. An increase in livelihood capital significantly improves their economic situation and social relationships, thus fulfilling their basic needs and substantially enhancing their social integration (Peng and Ling, 2019). As their level of social integration improves, their basic needs are better met, and further increases in livelihood capital result in diminishing marginal effects on social integration, leading to a weaker impact.

The empirical model discussed above may suffer from endogeneity issues, such as mutual causality or omitted variable bias, when estimating the impact of livelihood capital on social integration. To address these potential endogeneity problems, suitable instrumental variables need to be selected. Drawing on previous studies (Wang et al., 2021), we choose “the mean livelihood capital of other relocated households in the same resettlement community” and “the altitude of the relocation site” as instrumental variables for social integration. In a community, where social interactions are often based on acquaintances, the “peer effect” is prevalent. Therefore, the livelihood capital of relocated households is influenced by the mean livelihood capital of other relocated households in the same community, fulfilling the correlation requirement for instrumental variables. Additionally, the livelihood capital of other relocated households does not directly affect the social integration of a given household, satisfying the exogeneity condition. High-altitude areas generally face harsh climates, poor transportation, limited resources, and underdeveloped infrastructure, all of which constrain local economic activities and development opportunities, thus affecting the accumulation of livelihood capital for farmers. Consequently, the altitude of the relocation site is relevant as an instrumental variable. At the same time, the altitude itself does not directly impact the social integration level of relocated households in their new residence, thus fulfilling the exogeneity requirement of the instrumental variable.

Prior to conducting the regression, we tested the validity of the selected instrumental variables. The results of the non-identification test show that the Kleibergen-Paap rk LM statistic is 27.21, with a corresponding p-value of 0.000, rejecting the null hypothesis of “non-identification.” For the weak instrument test, the Cragg-Donald Wald statistic is 16.89, which exceeds the critical value of 11.59 at the 10% bias level, thereby rejecting the hypothesis of “weak instruments.” The Hansen J statistic for the over-identification test gives a p-value greater than 0.1, meaning we fail to reject the null hypothesis that “all instrumental variables are exogenous.” This confirms that the instrumental variables meet the exogeneity requirement.

Table 5 presents the model parameters after including the instrumental variables in the two-stage ordinary least squares estimation. The second-stage regression results show that after addressing potential endogeneity issues, the coefficient of livelihood capital on the social integration of relocated households is 1.519, which is statistically significant at the 1% level. Although this estimated coefficient is larger than the one obtained from the baseline regression, the sign and significance remain consistent. This suggests that the conclusion that increased livelihood capital promotes the social integration of relocated households is robust, even after accounting for omitted variables and potential reverse causality between livelihood capital and social integration.

Three methods are used to conduct robustness tests on the baseline regression results. First, we replace the dependent variable, following the approach of Qian et al. (2021), by using “whether they intend to live permanently at the relocation site” instead of social integration. Since the dependent variable is binary, we use a binary logit model for empirical analysis. Model (1) in Table 6 reports the robustness test results using the alternative dependent variable. The results show that livelihood capital is significantly and positively associated with social integration after the substitution, indicating that the main findings of this study are robust.

Second, we modify the measurement method for the variables. In the baseline regression, the level of social integration is calculated using factor analysis. To avoid potential biases due to differences in measurement methods, we re-estimate the social integration level using the entropy method. Model (2) in Table 6 presents the regression results based on this alternative measurement. The results are consistent with those of the baseline regression, with livelihood capital continuing to have a significant positive impact on social integration at the 1% level, further confirming the robustness of the main findings.

Third, we alter the sample by excluding elderly individuals, who may face additional health issues, lower labor participation, and different social engagement patterns (Zhang and Sun, 2024). We exclude households where the head of the household is over 60 years old and re-estimate the regression. The results, shown in Model (3) of Table 6, indicate that the relationship between the dependent and independent variables remain significantly positive, and the results are consistent with those of the baseline regression, confirming the robustness of the original model.

As discussed in the previous section, the livelihood risks faced by relocated households include three main aspects: environmental risk, employment risk, and health risk. Therefore, to further analyze the mediating role of livelihood risks, this paper first examines the mechanism through which livelihood capital affects the social integration of relocated households from the overall livelihood risk perspective. It then delves deeper into the mediating roles of environmental risk, employment risk, and health risk.

Table 7 presents the estimated impact coefficients after incorporating the mechanism variables. Model (1) shows that the impact coefficient of livelihood capital on livelihood risk is −0.378, which is statistically significant at the 1% level. This indicates that livelihood capital has a significant inhibitory effect on the livelihood risk of relocated households. Model (2) reveals that both livelihood capital and livelihood risk have significant impacts on the social integration of relocated households at the 1% level. Moreover, the coefficient for livelihood risk is negative, suggesting that livelihood risk plays a significant mediating role in the effect of livelihood capital on the social integration of relocated households. In other words, livelihood capital promotes the social integration of relocated households by reducing livelihood risks. In terms of sub-dimensions, environmental risk, employment risk, and health risk each play a partial mediating role in the relationship between livelihood capital and social integration.

To ensure the robustness and reliability of the mediation results, this paper further employs the Bootstrap and Sobel tests to assess the mediating role of livelihood risk (Gong et al., 2024). The results are shown in Table 8. According to the Sobel test, the direct and indirect effects of livelihood risk, environmental risk, employment risk, and health risk are all significant at the 1% level. This confirms that livelihood risk plays a significant mediating role in the process by which livelihood capital affects the social integration of relocated households. It is important to note that the Bootstrap test for mediation effects is based on whether the 95% confidence interval contains 0. If the interval does not contain 0, the mediation effect is present; if it does contain 0, the mediation effect does not exist. The Bootstrap results indicate that the confidence interval for livelihood risk does not include 0, which further supports the conclusion that livelihood risk has a significant mediating effect in the relationship between livelihood capital and the social integration of relocated households. Thus, hypotheses H2, H2a, H2b, and H2c are validated.

From a comprehensive perspective, livelihood risk plays a crucial role in the process through which livelihood capital influences the social integration of relocated households. It is a key factor in evaluating whether migrants can achieve “stability, integration, and gradual prosperity” after relocation. Among these, environmental risk, employment risk, and health risk each play a partial mediating role, with their mediation effects accounting for 10.32, 31.42, and 23.42%, respectively. This suggests that, in addition to directly affecting the social integration of relocated households, livelihood capital also improves social integration by reducing environmental, employment, and health risks.