The influence of enterprise dormitories on the urban integration of migrant workers in China: an exploration of two distinct migration stages of individual and family migration and the differences between them

China has experienced massive internal migration, with 376 million migrants in 2020, of whom rural-urban migrant workers constitute about two-thirds. High urban housing prices and limited access to public housing (constrained by the hukou system) push many migrant workers into low-cost housing options, notably enterprise dormitories and urban villages. The paper asks whether enterprise dormitories facilitate migrant workers' urban integration and how effects differ across migration stages. The authors posit a two-stage process: an initial individual migration stage focused on economic integration (jobs and income), where dormitories reduce living costs and risks; and a later family migration stage prioritizing spouse and child reunification, where dormitories' limited space hinders family integration. The study contributes by shifting attention to enterprise dormitories (often overlooked relative to urban villages) and by disentangling heterogeneous impacts across migration stages to inform housing policy.

The hukou system shapes migrants' access to welfare, employment, and housing, leading to residential patterns on city fringes and reliance on informal housing. Prior studies emphasize that housing tenure affects integration, but less attention is paid to how living patterns themselves influence integration. Enterprise dormitories are theorized within a dormitory labour regime that integrates production and reproduction spaces to extend control and reduce labor costs (Smith, 2003; Ren & Pan, 2006; Smith & Pun, 2006). While urban villages have been widely examined as low-cost housing, enterprise dormitories' role remains underexplored, often viewed negatively for living conditions and control. The paper conceptualizes urban integration as multidimensional (economic, social, psychological, identity/community), and frames migrant life cycles as progressing from individual to family migration stages. From this, two sets of hypotheses are formulated: Hypothesis 1 (supply-side/production logic)—dormitory provision relates to wage systems, labor time, and cost control (1a–1d); Hypothesis 2 (demand-side/life course)—dormitories support initial economic adaptation but impede family reunification, with specific expectations around marital status, initial employment, job stability, spouse reunification, and family reunification (2a–2e).

Design: The study combines two survey datasets to analyze supply-side (enterprise behavior) and demand-side (migrant life course) mechanisms. Datasets: - 2010 Pearl River Delta (PRD) and Yangtze River Delta (YRD) survey: Matched enterprise-employee dataset focused on migrant workers' labor rights and interests in two highly developed regions. Final analytic sample sizes vary by model (~3,934 to 4,152 observations). Strength: contains both enterprise and worker information, enabling linkage of dormitory provision to wage systems and labor time. Limitation: regional scope only. - 2017 China Migrants Dynamic Survey (CMDS2017): National individual-level dataset of migrants with agricultural hukou under age 65; after filtering, N=107,560. Strength: comprehensive coverage of family characteristics (spouse/child migration), employment, housing; reflects pre-COVID period. Limitation: lacks enterprise attributes and matching; fewer variables to test some hypotheses. Key variables: - Treatment: Enterprise dormitory (binary), indicating whether the enterprise provides dormitory housing (PRD/YRD) or whether the respondent lives in an enterprise dormitory (CMDS2017). - Outcomes (supply-side H1): wage system (hourly/time-rate vs basic/piece), weekly working hours, logarithm of hourly wage, logarithm or mean monthly wage. - Outcomes (demand-side H2): choice to live in a dormitory; marital status; initial employment (first-time entry to urban labor market, PRD/YRD only); job stability/loyalty; spouse reunification; family reunification (including minor children). Controls: - PRD/YRD: enterprise size (<100; 100–1000; >1000), sector (manufacturing, services, construction), enterprise property (state-owned, HK/Macau/Taiwan-funded, Europe/America/Japan-funded; reference: other private/joint-stock), worker gender, age (and squared), education, marital status, initial employment. - CMDS2017 (robustness): gender, age, marital status, education, industry, years living in city (enterprise-level variables unavailable). Models: - For binary dependent variables (e.g., wage system type, dormitory choice, spouse/family reunification): logit models; linear probability models (LPM) used as robustness checks. - For continuous dependent variables (e.g., weekly hours, log hourly wage, log/mean monthly wage): OLS regressions. Descriptive statistics: - PRD/YRD (N≈4,152): 51.4% of enterprises provide dormitories. Spouse migration rate 73.0%; family migration 35.9%. Compared to non-dorm enterprises, dorm enterprises show longer weekly hours (246.9 vs 235.8), lower log hourly wages (2.030 vs 2.077), with similar monthly wages; non-dorm enterprises more often use hourly wage systems (52.2% vs 42.7%). - CMDS2017 (N=107,560): 14.5% live in enterprise dormitories. Spouse migration 67.9%; family migration 47.1%. Dorm residents work overtime more, have lower log hourly and monthly wages, are younger, more often male, have shorter city tenure, and are less likely to be married.

- Supply-side (H1: production logic): - H1a (wage system): Enterprises providing dormitories are less likely to adopt hourly/time-rate systems and more likely to use basic/piece wage systems (logit coefficient for dormitory on piece/time indicator −0.550, p<0.01). - H1b (labor time): Dormitory provision is associated with longer working hours; employees in dormitory-providing firms work about 10.33 more hours per week (β=10.331, p<0.01, PRD/YRD). In CMDS2017, dormitory living also predicts higher working hours (β=0.662, p<0.01). - H1c (unit labor cost): Dormitory provision correlates with lower hourly wages (β=−0.048 in log hourly wage, p<0.01, PRD/YRD). Robustness (CMDS2017): log hourly wage β=−0.067 (p<0.01). - H1d (total wage costs): No significant difference in monthly wages in PRD/YRD (β=−0.009, ns), indicating cost control via longer hours and lower unit wages. Robustness (CMDS2017): log monthly wage β=−0.038 (p<0.01). - Demand-side (H2: migration stages): - Individual migration stage: - H2a: Married workers are less likely to live in dormitories (logit β=−0.769, p<0.01, PRD/YRD); nationally β=−0.886 (p<0.01, CMDS2017). - H2b: Initial employment (first-time urban workers) are more likely to select dormitory-providing enterprises (logit β=0.120, p<0.1 when controlling job stability, PRD/YRD). - H2c: Lower job stability is associated with dormitory living (logit β=−0.157, p<0.05, PRD/YRD), consistent with dorms as a short-term springboard. - Family migration stage: - H2d: Dormitory living reduces spouse reunification (logit β=−1.198, p<0.01, PRD/YRD); nationally β=−1.257 (p<0.01, CMDS2017). - H2e: Dormitory living reduces family reunification (logit β=−0.658, p<0.01, PRD/YRD); nationally β=−0.967 (p<0.01). - Descriptive contrasts: - PRD/YRD: Dorm enterprises show longer hours (246.9 vs 235.8), lower log hourly wages (2.030 vs 2.077), similar monthly wages; lower spouse (59.3% vs 84.2%) and family migration (27.4% vs 42.8%) rates than non-dorm firms. - CMDS2017: Dorm residents more often work overtime (0.789 vs 0.770), have lower log hourly (2.711 vs 2.768) and monthly wages (8.114 vs 8.152), and lower spouse (40.8% vs 72.4%) and family migration (27.2% vs 50.5%) rates.

Findings support a two-stage integration pathway shaped by housing arrangements. In the individual migration stage, enterprise dormitories lower living costs and adaptation risks, facilitating entry and economic integration for young, single or first-time urban workers. From the supply side, dormitories serve firms' production logic by extending labor time and reducing unit labor costs without increasing total wage bills. In the family migration stage, the spatial and institutional features of dormitories constrain spouse and child co-residence, thereby impeding family reunification and deeper social integration. This clarifies why dormitories can simultaneously aid initial adaptation yet hinder longer-term family settlement, aligning with broader patterns of low-cost urbanization in China where limited public housing and hukou-linked welfare push migrants into informal or employer-provided housing. The results highlight tensions between enterprise cost strategies and migrants' evolving life-cycle needs, with implications for housing policy, labor management, and social stability (e.g., left-behind children).

The study demonstrates that enterprise dormitories in China are embedded in a dormitory labour regime that primarily serves production by increasing working hours and lowering unit wages while keeping total wage expenditures stable. Dormitories support economic integration at the individual migration stage but impede spouse and family reunification at the family migration stage. Robustness checks with national CMDS2017 data corroborate these patterns. The paper broadens understanding of low-cost urbanization by showing how housing supply and migrants’ life-cycle demands interact. Future research should: (1) examine children’s development outcomes amid evolving family migration; (2) leverage updated datasets and measures to reassess dormitory roles given labor market and housing policy changes (robotics, new factories, housing affordability shifts); and (3) track how real estate market dynamics and dual-track housing (commercial plus affordable) reshape dormitory prevalence and migrant integration.

Key limitations include: (1) use of two datasets from different years and sampling designs (regional PRD/YRD 2010 vs national CMDS2017), limiting direct comparability; (2) regional scope of the PRD/YRD survey restricts generalization; (3) CMDS2017 lacks enterprise-level matching and detailed variables (wage system type, initial employment, job stability), preventing tests of some hypotheses; (4) data availability lags prevent depiction of conditions as of 2022; (5) findings are specific to China’s institutional context (hukou, housing policy), potentially limiting external generalizability.