The impact of social exclusion and identity on migrant workers' willingness to return to their hometown: micro-empirical evidence from rural China

Rapid industrialization and urbanization in China have attracted large numbers of rural surplus labor to cities, contributing to urban non-agricultural sectors. However, rural-urban migration has left many rural communities with the elderly and children, while rural revitalization strategies emphasize attracting and retaining rural talents. Despite policy support to encourage return entrepreneurship, migrant workers often face cultural and psychological barriers to urban integration, including social exclusion and discrimination by urban residents and labor market institutions. These challenges may shape migrant workers’ willingness to return to their hometowns (MWRH). This paper investigates, from the perspective of identity, how social discrimination (used interchangeably with social exclusion) affects MWRH, and whether identity serves as a mediating mechanism. The study aims to inform integrated urban–rural development and rural revitalization by clarifying how discrimination and identity influence migrants’ return intentions.

Influencing factors of MWRH: Prior research identifies rural "pull" factors (hometown environment, supportive policies, non-agricultural development, lower living costs) and urban "push" factors (household registration thresholds, children’s education burden, adaptation pressure, lack of non-agricultural experience, social security arrangements, social network exclusion, urban identity, and income). Education, working hours, and marital status also matter. Social exclusion and MWRH: Migrants’ urban integration depends on two-way interactions with locals; discrimination stemming from interests, culture, and institutional factors can hinder integration and promote return intentions. The Migrant Assimilation perspective suggests discrimination weakens settlement willingness, leading to return. Identity and MWRH: Identity reflects belonging and behavioral standards guiding choices. Strong urban identity facilitates social integration and equal access to rights and services, reducing return intentions. Social exclusion and identity: Conflicts between natives and newcomers and dual labor market segmentation foster exclusion, reducing migrants’ sense of belonging and identity. Better neighborhood environments reduce perceived discrimination and strengthen belonging. Hypotheses: - H1: Social discrimination has a significant positive effect on MWRH. - H2: Identity has a significant negative effect on MWRH. - H3: Social discrimination has a significant positive effect on MWRH through identity (identity mediates).

Data: China Migrants Dynamic Survey (CMDS) 2017 by the National Health and Family Planning Commission, covering 31 provinces, ages 15–59, PPS sampling. After removing inapplicable responses, N=47,011. Variables: - Dependent: MWRH (1=will return home when not intending to stay in the city; 0 otherwise), derived from two CMDS questions on settlement intention and alternative plans. - Key independent: Social discrimination index (0–5) summing five equally weighted binary dimensions: 1) Interpersonal interaction discrimination (social network primarily non-locals/rarely interacts=1; interacting with locals=0). 2) Labor market discrimination (no labor contract/one-off task=1; otherwise=0). 3) Occupational discrimination (informal job such as courier/cleaner/builder/housekeeper=1; otherwise=0). 4) Cultural discrimination (agree it is important to follow hometown customs and/or that health habits differ from locals=1; otherwise=0). 5) Regional discrimination (agree locals look down on strangers=1; otherwise=0). - Mediator: Identity measured by four items (e.g., “I feel I am a local,” “Locals would accept me,” “I like the current city,” “I would like to integrate and be one of the locals”) on a 4-point scale, higher values indicating higher identity. - Controls: Individual (gender, ethnicity, political affiliation, marital status, education, age, homeownership), family (log monthly household income, number living together, contracted land, homestead, household difficulties), mobility (number of cities moved, flow range: cross-provincial/city/county). Models: - Probit for MWRH with robust standard errors. - Mediation analysis via Baron and Kenny stepwise approach, plus Sobel and Bootstrap tests. Equations: Y=cX+e; M=aX+e; Y=c′X+bM+e; and empirical specifications: Willingness=α1+β1Discrimination+β2Controls+ε; Identity=α2+γ1Discrimination+γ2Controls+ε; Willingness=α3+δ1Discrimination+δ2Identity+δ3Controls+ε. - Endogeneity: Instrumental variable Probit (IV Probit) using “whether applied for a residence permit” as IV for social discrimination (relevance: reduces discrimination via access to services; exogeneity: not directly related to MWRH). First-stage F=786.11. - Robustness: Alternative estimator (Logit), alternative construction of discrimination via entropy evaluation method (EEM), and sample trimming (exclude ages ≤20 and/or ≥60). Goodness-of-fit, Wald, Hosmer-Lemeshow tests reported.

- Descriptives: N=47,011; mean age ~36; Han 90%; married 84%; education mean 2.26 (between junior and senior high); homestead 74%; contracted land 60%; mean social discrimination 2.48. - Main effect (Probit): Social discrimination positively predicts MWRH at 1% significance across model specifications. With full controls, coefficient ~0.142 (SE=0.013). Reported marginal effect: 0.96%, and described as a 0.83% increase in MWRH probability per unit increase in discrimination. Model fit: goodness-of-fit 96.84%; Wald 594.88***; Hosmer-Lemeshow p=0.052. - Controls: Age and political affiliation (Party membership) positively associated with MWRH; Party members 1.51% more likely to return. Higher income reduces MWRH; home ownership reduces MWRH; availability of homestead/contracted land increases MWRH; larger mobility range reduces MWRH; more cities moved increases MWRH. - Mediation (Identity): Discrimination negatively affects identity at 1% significance; identity negatively affects MWRH at 1% significance. After adding identity, discrimination’s effect on MWRH remains positive but reduced, indicating partial mediation. Mediation share: 50.77%. Sobel test Z=23.63 (p<0.01). Bootstrap (1,000 iterations) CI for mediated effect (0.0032, 0.0067), excluding 0. - Endogeneity (IV Probit): IV (“applied for residence permit”) strongly relevant (first-stage F=786.11). Over-identification/strength tests (AR=116.36; Wald=14.54***) support instrument validity. Second-stage confirms positive effect of discrimination on MWRH (Chi(2)=107.74***). - Heterogeneity: • Region (Probit coefficients): Eastern 0.141***; Middle 0.125***; Western 0.146***; Northeastern 0.260***. Mediation holds across regions; identity negative and significant where reported. • Age: New generation (born ≥1980) 0.139***; Older generation 0.149***; identity mediates in both groups. • Flow range: Inter-provincial 0.135***; Intra-provincial 0.151***; identity mediates in both. • Income groups (monthly household): <3000 yuan 0.206***; 3000–5000 0.167***; 5000–8000 0.107 (reported significant in text); >8000 0.106**. Mediation present across income groups with identity negative and significant where reported. - Robustness: Results consistent using Logit; using EEM-based discrimination index; and after excluding ages ≤20 and/or ≥60; mediation by identity remains significant.

Findings show that social discrimination in interpersonal, labor market, occupational, cultural, and regional domains significantly increases migrant workers’ willingness to return home, aligning with migrant assimilation perspectives and highlighting barriers to urban integration. Identity—reflecting belonging and acceptance—reduces return intentions and serves as a key mechanism: discrimination lowers identity, which in turn raises the likelihood of returning. The effects are stronger in less developed regions (western, northeastern), among older migrants, those with lower incomes, and intra-provincial movers, suggesting that local economic context, life-course stage, and resource constraints condition how exclusion translates into return intentions. The robustness and IV analyses support a causal interpretation consistent with the proposed framework. These results underscore the importance of reducing discrimination and fostering urban identity to support migrants’ urban settlement where desired and inform targeted interventions for groups most affected.

The study demonstrates that: (1) social discrimination significantly increases migrant workers’ willingness to return home (MWRH), with results robust to endogeneity controls; (2) identity significantly suppresses MWRH and partially mediates the discrimination–MWRH link, explaining about 50.77% of the total effect; and (3) the discrimination effect is heterogeneous—stronger in western and northeastern regions, among older, lower-income, and intra-provincial migrants. Policy implications include: promoting a positive image of migrant workers to reduce prejudice; enhancing migrants’ legal literacy, professional ethics, and integration into urban life; and balancing urban expansion with rural revitalization by improving labor mobility systems and equalizing public services, alongside supportive platforms for returnees to contribute to rural development. Theoretically, the paper enriches understanding of psychological factors (discrimination and identity) in shaping MWRH and clarifies identity’s mediating role.