What Drives Migrants Back to Set up Firms? Return-home Entrepreneurial Intention of Rural Migrant Workers in China

Migrants’ entrepreneurship has been discussed in the Western literature on international immigrants. Due to low-skilled labor and market discrimination, transnational migrants have encountered barriers in the mainstream job market in host countries, which has led to their entrepreneurial activities (Dana and Morris, 2007). With the emergence of the new generation of knowledge-based international immigrants, there was a ‘revolutionary return’, the return of such high-skilled immigrants to their home countries, which received academic attention (Démurger and Xu, 2011; Leibbrand et al., 2019). Some new return immigrants tend to engage in knowledge-based entrepreneurial activities, making use of the skills learned and social and human capital acquired abroad (Pîrvu and Axinte, 2012). However, the literature has paid little attention to the determinants that inspire immigrants’ desire or passion to set up a new business venture and consciously plan to do so in the future in their home countries, namely, return-home entrepreneurial intention (RHEI).

Entrepreneurial intention is an individual’s self-acknowledged conviction (Thompson, 2009). It also reflects a state of ‘developing/potential entrepreneurship’, which can be regarded as the budding stage of entrepreneurial activities (Vivarelli, 2004). According to planned behavior theory, entrepreneurial intention is a reliable predictor of entrepreneurial behavior (Tognazzo et al., 2016; Bilgiseven and Kasımoğlu, 2019), especially for these groups who have potential for future entrepreneurship, such as students (Solesvik et al., 2014; Buli and Yesuf, 2015), international return migrants (Autio et al., 2014), and return rural migrant workers (RMWs, the rural registered permanent residence who has been employed for more than 6 months away from the township, town or street where his or her registered permanent residence is located) in China (Duan et al., 2020).

RMWs in China are characterized by similarities with early international migration (Zhu et al., 2019a). The majority of RMWs are low-skilled workers (Akay et al., 2012; Chen et al., 2014), and work in low-paid jobs due to the urban-rural dual structure, household registration systems, and low labor skills (Duan et al., 2020). After 2000, RMWs gradually returned to their hometowns to seek new career development. The annual increase in the number of RMWs returning to their hometowns remained at more than 10% (Duan et al., 2020). RMWs are likely to engage in entrepreneurial activities upon their return due to accumulating human, social and financial capital (Démurger and Xu, 2011; Hessels et al., 2017). Those returning RMWs as a/an (potential) entrepreneur are called ‘back phoenixes (to the chicken nest)’ as a metaphor. However, what initially drives them to set up firms in their hometowns has not been discussed fully. Their return-home entrepreneurial intention (RHEI) is interpreted as a decision after weighing the unfair treatment in the urban labor market against the less constrained career choices in their hometowns (Zhao, 2002), which implies return-home entrepreneurship is survival-oriented but not opportunity-oriented. However, Duan et al. (2020) believed that job demand and entrepreneurial resources together stimulated RHEI. Moreover, Démurger and Xu (2011) point out that the urban experience makes RMWs more likely to start entrepreneurial activities than local farmers, which implies that the positive effects of migration destinations on RHEI are undervalued. As a consequence, it is urgent to use an integrating analytical framework to take into consideration the effects of hometowns and migration destinations, which RMWs were embedded into. In combination of ‘mixed embeddedness’ with ‘double-layered embeddedness’, this paper took RMWs in Anhui Province, China, as an example and analyzed the impacts of the social, economic and institutional context in the two types of regions, RMWs’ hometowns and migration destinations, on RMWs’ RHEI. The findings aim to supplement the idea advocated by Pike et al. (2000) that the use of embeddedness requires the interpretation of ‘who is embedded in what’ and better local environments for rural sustainable development by promoting RMWs’ return-home entrepreneurship (Démurger and Xu, 2011).

Early entrepreneurial intention research addressed individuals’ entrepreneurial traits and revealed that personalities, educational experiences, and self-efficacy influenced individuals’ entrepreneurial cognition and intention, as well as their ability to invoke entrepreneurial resources (Buli and Mohammed, 2015; Law and Breznik, 2017). Subsequently, another line of literature found that entrepreneurial intention was also influenced by environmental factors (Thompson, 2009; Autio and Acs, 2010; Autio et al., 2014), such as institutional and economic (Minniti and Bygrave, 1999), educational (Buli and Yesuf, 2015), and cultural (Solesvik et al., 2014) contexts, which means that it is embedded in a specific context (Jack and Anderson, 2002; Solesvik et al., 2014). To some extent, embeddedness is considered as a confusingly fuzzy notion (Markusen, 1999) and polyvalent concept, since there are many definitions of what embeddedness might consist of, such as political, cultural, structural, cognitive, social, temporal, technological, etc. (Hess, 2004). Additionally, Szkudlarek and Wu (2018) referred to mixed embeddedness and Wigren-Kristofersen et al. (2019) called for research on the multilayered perspectives of embeddedness.

Changes in a specific context or localities with different contexts bring about the identification and development of entrepreneurial opportunities (Ardichvili et al., 2003). However, Duan et al. (2020) mentioned that much of the literature focused on macroscale policies and little was concerned about mesoscale (such as regional) factors. Kloosterman et al. (1999) determined the concept of mixed embeddedness to discuss how international immigrants successfully undertake entrepreneurial activities in the context of ethnic communities, market opportunities and macrolevel institutions in host countries. However, they care little about the effects of home countries or regions. The cultural, institutional and socioeconomic context of home countries also shapes the perception of migrants at the individual and national levels (Azmat, 2010); thus, Greenberg et al. (2018) view double-layered embeddedness as important for many rural business owners, which includes embeddedness in local-traditional networks based on long-term familiarity, trust and professionalism and embeddedness in extra-regional and national professional networks in urban population centers where they build up their business partnerships. However, meaning in the host and home countries is helpful for international migrants to identify international opportunities (Ren and Liu, 2015) and offer them competitive advantages over non-migrants (Abramitzky et al., 2019).

Migrant workers in China keep close ties with their hometowns. The annual Spring Festival travel movement from the southeastern coastal areas to the central and western regions to a large extent shows the indissoluble social and economic ties between these people and their hometowns. Meanwhile, in many places, the establishment of start-up parks for migrant workers returning to their hometowns shows the support provided by the formal system of their hometowns. Moreover, some China’s cases indicated that both RMWs’ hometowns, and migration destinations influence their entrepreneurial activities in the host regions (Démurger and Xu, 2011; Zhu et al., 2019a). Return-home RMWs are anticipated to act as a bridge between their hometowns and host regions and set up industrial linkages between the two places, which also demonstrates the importance of double-layered embeddedness. This is especially worth further examination. As a consequence, the authors extend mix embeddedness to a broader meaning, in integration with double-layered embeddedness, and discuss the effects of the three-dimensional contexts in the two regions on RHEI.

Granovetter (1985) used the concept of social embeddedness to measure the effects of social networks on economic behavior. According to Hisrich (1990), it is the interaction of potential entrepreneurs’ role (in social networks) and social context that leads to their high entrepreneurial awareness. Social networks, as a kind of social capital, can be transformed into other forms of capital (Davidsson and Honig, 2003) and provide individuals with facilitation to access scarce resources. Especially in rural areas, farmers with high-quality social connections are more likely to discover new entrepreneurial opportunities and increase the likelihood of entrepreneurship (Ma, 2002). Moreover, embeddedness in the social relationships of their hometowns and migration destinations links the resources of internal and external networks (Bosworth and Atterton, 2012), which contributes to the success of RMWs’ returning entrepreneurship. From this perspective, social embeddedness into the two places is probably related to RMWs’ RHEI.

Generally, kinship-based connections in rural areas of China provide RMWs physical and mental support to set up a firm; however, high-quality social relationships beyond kinships are supposed to be constructed and sustained with purposes. The experience of leaving home weakens the originally stable embeddedness of social and economic systems and leads more or less to de-embeddedness and a low frequency of interaction with their original communities (Davanzo, 1976). It is not surprising that RMWs may feel like ‘missing persons’ in their hometown, and the context of their economic activities and social relations change after migration (Ghezzi and Mingione, 2007). In addition, due to the differences in job interests and life rhythm, it is difficult for RMWs to integrate into the local social network (Gmelch, 1980). As a consequence, rebuilding and maintaining crucial connections can provide RMWs with necessary support for their return plan (Eccles and Nohria, 1992), especially for entrepreneurial choice. In Ma’s view, human capital accumulated by RMWs during their migration can only be fully utilized through the social network of their hometowns (Ma, 2002). From this point of view, RMWs who own intensive connections to their hometowns and thus obtain access to more local information and resources, tend to have RHEI.

Rural entrepreneurship is not purely endogenous (Deller et al., 2019) and is supposed to need extra-local social and economic networks instead of internal networks within the local rural context. The migration enables RMWs to expand their social connections to local employers in migration destinations and workmates (especially with entrepreneurial intention) from other cities or provinces. As the lack of entrepreneurial resources is considered as the main constraint of entrepreneurial intention and successful entrepreneurial activities, such social connections would affect RMWs’ entrepreneurial cognition, and provide them with professional skills, business experiences, and the identification of market opportunities (Démurger and Xu, 2011). Moreover, maintaining social contacts with the outside is conducive to bringing resources back to the countryside and linking their resources in urban areas with those in their hometowns upon returning as well (Jones et al., 2014). Therefore, Hypothesis 1 (H1) is proposed:

Hypothesis 1a (H1a): The social connections rooted in RMWs’ hometowns stimulate their RHEI.

Hypothesis 1b (H1b): The social connections built in RMWs’ migration destinations benefit their RHEI.

In Kloosterman et al.’s (1999) view, economic embeddedness emphasized the role of the political and economic structure of the market in international immigrants’ entrepreneurial activities. The accessibility of resources determines individuals’ opportunity discovery and entrepreneurial intention. Economic embeddedness means that individuals are alert to gaps in the market and messages from the economy, and are attentive to how to exploit a gap in the market (Solesvik et al., 2014). RMWs become familiar with the local economic condition and have sensitivity to entrepreneurial resources (such as human and financial capital) and subjective perception of entrepreneurial opportunities, which affect their entrepreneurial alertness and entail their entrepreneurial intention (Ardichvili et al., 2003).

In the migration destination market, due to a lack of resources (such as human and financial capital), there are few entrepreneurial opportunities for RMWs, and their opportunity structure is essentially reflected in the barriers to entering the mainstream market (Sassen, 1996). However, this does not mean that economic embeddedness has no relation to RHEI. The experience in migration destinations gives them access to market information. Some RMWs could find market gaps by matching new firms with existing firms and then recognize entrepreneurial opportunities and transplant the existing entrepreneurial examples to their hometowns, as Sarasvathy et al. (2010) pointed out. RMWs can establish economic and technological links and find market gaps between their hometowns and migration destinations. Even after returning, they can reach new market opportunities by connecting resources between the places where they move in and out (Bosworth and Atterton, 2012).

Gaddefors and Cronsell (2009) analyzed successful entrepreneurs’ returning and found that if entrepreneurs embed their companies into the local context, they can reduce their dependence on the external economy and save costs (Jack and Anderson, 2002). Entrepreneurs can learn about their local industrial situation and start their own businesses by contacting local stakeholders and listening to local interest groups (Gaddefors and Cronsell, 2009). In this way, they are more easily exposed to local residents’ needs and market gaps. Thus, they are inclined to choose self-employment or enter a higher level of work after returning (Ilahi, 1999).

Therefore, Hypothesis 2 (H2) is proposed:

Hypothesis 2a (H2a): The market opportunities that RMWs identify in their hometowns promote their RHEI.

Hypothesis 2b (H2b): The market opportunities that RMWs identify based on the gap between their hometowns and migration destinations inspire their RHEI.

The institution includes formal institutions and informal institutions (Scott, 2014). Formal institutions, such as laws and regulations, stipulate the rules for economic activities and may affect an individual’s entrepreneurial intention and behavior. Informal institutions, such as values and cultural cognition, affect the nature and process of entrepreneurship and the way entrepreneurs behave (Welter and Smallbone, 2011).

Formal encouragement policies of entrepreneurial activities are believed to decrease resistance in practicing business activities and increase the accessibility of scarce resources and economic returns, which motivates potential entrepreneurship (Blanchflower, 2000). For example, the Anhui Department of Human Resources and Social Security issued the ‘Notice on Further Supporting and Promoting the RMWs to Return and Start Their Business’ (http://hrss.ah.gov.cn/public/6595721/8439905.html, accessed on 2020-09-16) in January 2020. Specifically, the notice proposes to give entrepreneurs a one-off subsidy of 5000 yuan (RMB) and support the construction of 150 demonstration parks for returning RMWs’ entrepreneurship. Furthermore, the government organizes entrepreneurship-oriented training to promote entrepreneurship. It was found that education and training are conducive to impart entrepreneurial skills so that individuals can be more active in engaging with their partners’ entrepreneurship and confident in carrying out their entrepreneurial intention (Davidsson and Honig, 2003; Buli and Yesuf, 2015). However, there are two questions. First, it is uncertain whether the government policies work truly well, as in China, new policies tend to be tested on a small scale (Yang, 2004), and only a small number of entrepreneurs may benefit from these policies. Second, in RMWs’ migration destinations, due to the relatively good innovation level and economic foundation, cities prefer knowledge-based innovation and entrepreneurship. Therefore, formal institutions usually encourage knowledge-based entrepreneurship rather than labor-intensive entrepreneurship, and RMWs’ entrepreneurial activities cannot meet the requirements (Penco et al., 2020).

In addition, the informal institution reflects the institution from a cultural-cognitive dimension, including local traditions, customs, social norms and culture, and refers to the shared conceptions that constitute the nature of social reality and the frames through which meaning is made (Scott, 2014). If a region upholds the values of entrepreneurialism, the burgeoning entrepreneurial culture stimulates new business start-ups (Beugelsdijk, 2010). In the context of migration, during working outside, the model of successful entrepreneurs in RMWs’ individual social networks has important impacts on their entrepreneurial intentions (Auken et al., 2006; Nowiński and Haddoud, 2019). The value orientation brought by the surrounding entrepreneurial success affects their entrepreneurial psychology. At the same time, entrepreneurial activities in their hometown create local entrepreneurial culture, and the increasing entrepreneurial spirit will drive reforms and the legalization of entrepreneurship-oriented informal institutions.A cultural environment encouraging entrepreneurship created in communities is conducive to the occurrence of entrepreneurial activities (Zhu et al., 2019b).

Therefore, Hypothesis 3 (H3) is proposed:

Hypothesis 3a (H3a): The entrepreneurship-oriented institutions in RMWs’ hometowns stimulate their RHEI.

Hypothesis 3b (H3b): Informal institutions in migration destinations stimulate RMWs’ RHEI.

Along with the urban industrial structure adjustment and the development of the rural economy, a large-scale backflow of interprovincial migrant workers gradually emerged within China. Older migrant workers with lower education levels or without stable marital status seem to have stronger backflow intentions (Zhang et al., 2021). The distribution of backflow intention presents a spatial differentiation, and Anhui Province is a typical region. This paper takes Anhui Province as the research area, which is located in central China (Fig. 1) and is a major labor exporting province. In 2019, the total number of migrant workers from Anhui Province reached 19.8 million, of which 14.0 million were RMWs, accounting for 70.2% of the total migrant workers (http://ahzd.stats.gov.cn/web/view?strId=a179253898f148a6b916b33a2ddd862e&strColId=dffe6cc990a74da9904117911e9112aa&strWebSiteId=8ced4cf95fd8406fba69284370440ab8). In 2000, to attract and encourage RMWs to return to their hometowns, the counties where their household registration locations were, the provincial government officially launched the ‘Back Phoenix: Return of Talent’ Project with funds and technology. The project encouraged local registered RMWs to return and engage in entrepreneurial activities in townships or rural areas, guiding the evolution of the local ‘outsourcing work wave’ toward the ‘back to home business wave’. For example, Hefei economy developed rapidly, which brought more employment and entrepreneurial opportunities for returning RMWs. In 2018, Anhui Province’s population indicated a net inflow of 282.3 × 10³ , being the third largest in all the provinces (National Bureau of Statistics of the People’s Republic of China, 2019). Therefore, Anhui Province is the typical case for examining the determinants of RMWs’ RHEI in labor-exporting provinces of China. the third largest in all the provinces. Therefore, Anhui Province is the typical case for examining the determinants of RMWs’ RHEI in labor-exporting provinces of China.

Figure 1.  Location of study area

This study used a questionnaire to obtain first-hand data, which included information ranging from basic personal information (age, marriage, education, etc.), work status (wage level, work experience, etc.) to entrepreneurial intention, as well as the social, economic, and cultural contexts of their hometowns and migration destinations. According to the proportion of RMWs with return-home intention in Anhui Province, China, the sample size of each prefecture-level city was determined. Then, a random sample in the city at the prefecture level was taken. The survey was conducted in 2015 and 2020, and questionnaires were randomly distributed in 16 prefecture-level cities of Anhui Province. A total of 429 samples were recovered, of which 402 were valid samples, and 27 were excluded for missing some information. The distribution of sample sources is shown in Table 1. The sample proportion of each prefecture-level city was basically consistent with the proportion of its RMWs, so it was considered that the sample distribution was reasonable.

RHEI refers to the intention to engage in entrepreneurial activities for return migrant workers. In the effective sample of this paper, 61.69% of RMWs had RHEI, accounting for more than half of the sample. The basic statistics of the 402 valid samples show that those who are male, young, unmarried or have high education are more likely to have RHEI (Table 2).

In this paper, the independent variable is RMWs’ RHEI, which is reflected in the answer to the question ‘Do you plan to go back to your hometown (county-level administrative units) and set up a firm/shop or engage in self-employment business in the future’? Following the literature review and hypotheses presented in Section 2, eight indicators were selected as independent variables, shown in Table 3, to characterize the local and external social, economic, and institutional environments in which RMWs are embedded. In addition, as RHEI is affected by individual characteristics, considering RMWs’ mixed embeddedness in two places, gender, age, education and other indicators were included as control variables in the model. The specific index selected and the definitions of the indicators are shown in Table 3. In addition, the geographical locations of the hometown and migration destination (in other provinces) are related to the environment in which the RMWs are embedded. Therefore, in Model II, the dummy variables of the hometown and the migration destination were included. Taking the characteristics of the social and economic development of China and Anhui Province into consideration, the surveyed destinations to which RMWs migrated were divided into three categories, eastern China, central China and western China, and their hometowns were divided into three categories, northern Anhui, central Anhui, and southern Anhui (Fig. 1).

The results of the multicollinearity test for the independent variables selected in Table 3 are shown in Table 4. The results show that the minimum tolerance was 0.544. The variance inflation factors (VIFs) of all the indicators were less than 2.000, indicating that there is no multicollinear relationship between the indicators, and regression analysis can be performed. In addition, the P (Chi-square) of all categories was less than 0.05, indicating significant differences among all items.

Regression analysis is commonly used to analyze the extent of dependence between objects. Binary logistic regression (BLR) is a regression model established for binary variables (0 or 1) to study the relationship between dependent variables and their influencing factors. The independent variable can be qualitative data or quantitative data (Xiao et al., 2018).

In this paper, the RMWs’ RHEI (Yes or No) is a dichotomous variable, so the BLR model is used, and the basic formula is as follows:

where P_i represents the probability that sample i has RHEI, j represents the jth influencing factor, m represents the number of influencing factors, z_i is a random variable that follows the logistic distribution, β_j represents the regression coefficient of influencing factor j, X_ij represents the value of the jth influencing factor of sample i, α is the intercept of the model, and μ_i is the error term.

In this paper, the results of the BLR model fitted by the entry method are shown in Table 5. The variables of Model I include the environmental indicators that characterize the embeddedness framework and personal characteristics of RMWs, and the dummy variables location of the hometown (HLO) and the migration destinations (DLO) are further introduced in Model II. The prediction accuracy of Model I is 87.1%, and that of Model II is 86.8%, both greater than 85.0%, and the Hosmer and Lemeshow Test of Model II is 0.058, greater than 0.050, which indicates a relatively good model fit and accurate prediction.

In Model I and Model II, the independent variables HII, HECO, DII, DFI, DSOC, DECO and the control variables EDU and TEC pass the significance test, indicating that both environmental factors and individual characteristics affect RMWs’ RHEI. Among the independent variables, the OR of HECO is the largest in both models, 10.561 in Model I and 10.849 in Model II. The OR of TEC in the personal dimension is the second largest, 8.705, in Model I. This shows that compared with personal characteristics, environmental factors have greater impacts on RMWs, and it is necessary to pay attention to the impact of embeddedness on individuals’ entrepreneurial intention. The embeddedness framework does not exaggerate the importance of context analysis but is highly applicable to research on entrepreneurship. In this way, when describing an individual’s mental motivation and behavior in a specific situation, it is necessary to comprehensively reflect on how an individual’s economic activities will be stimulated and restricted in the social network and institutional context.

Furthermore, HII and HECO represent RMWs’ hometowns and DII, DFI, DSOC and DECO represent their migration destinations, indicating that the environment of the hometown and the migration destination both affect RMWs’ RHEI. Moreover, these variables represent the social, economic and institutional dimensions of the two places, which shows that the environment of RMWs’ hometowns and the migration destinations plays a role in their RHEI. This shows that the modification of the mixed embeddedness framework in this paper is effective. In the research on migrants’ entrepreneurship, the connection of the geographical location during their migration should be considered, and RMWs are regarded as an intermediary linking the environment of the migration origin (hometown) and destination. In this way, their entrepreneurial intention, behavior and results can be better understood. In contrast, focusing only on the impact of a migrant’s location at a given time ignores the hidden influences of other environments during their migration.

In Model I, comparing the indicators of RMWs’ hometowns and the migration destinations, the impacts of HECO (OR = 10.561) and HII (OR = 3.198) representing economic and institutional dimensions from their hometowns are greater than most of the corresponding indicators from their migration destinations, DII (OR = 4.118), DFI (OR = 0.371), DSOC (OR = 2.320) and DECO (OR = 1.353). This shows that the impact of the economic environment of RMWs’ hometowns is greater than the impact of the environment of their migration destinations. In Todaro’s model (Todaro, 1969), the institutional effects of urban biases and high expected income in developing cities would induce excessive rural-urban migration and increase unemployment and urban poverty. Thus, favorable conditions in the destinations are unlikely to have a positive impact on RHEI. However, in Stark’s theory (Stark and Taylor, 1991), labor migration is a family strategy for small rural households to spread risks across sectors and accumulate capital for a possible RHEI. Thus, favorable conditions in the destinations tend to have positive effects on RHEI. The uncertainty of the direction of influence leads to the smaller impact of destinations on RHEI. When RMWs consider returning home to engage in entrepreneurial activities, their first consideration is often the development situation of their hometown. RMWs’ RHEI is motivated by the ease of embedding in their hometown. If RMWs want to ensure that their RHEI can be implemented, they need to overcome their de-embeddedness after leaving their hometowns and become re-embedded in their hometown by invoking their capital and resources there (Greenberg et al., 2018). Additionally, embedding in hometowns and migration destinations helps RMWs accumulate resources. Embedding in the migration destinations helps RMWs expand their resources, such as knowledge, skills and social networks, outside the local rural areas (the hometown where entrepreneurial activities take place). Embedding in their rural hometown is directly related to the realization of RHEI, and intuitively, this embeddedness is stronger than that in the migration destinations.

Regarding the dimension of economic embeddedness, HECO and DECO pass the significance test at the 0.01 level in Model I, and the coefficients (B) of both variables are positive, indicating that the economic environment in the two locations has a positive effect on RMWs’ RHEI, which supports Hypothesis 2a and Hypothesis 2b. HECO (OR = 10.561) has the strongest influence on RMWs’ RHEI. The result shows that it is conducive to stimulating RMWs’ RHEI if they embed the economic environment of their hometowns and are sensitive to the market opportunities there. The more positive the RMWs’ judgment of the economic environment and more sensitivity to the market opportunities in their hometown, the more likely they are to present RHEI. When RMWs consider selecting their hometown as the location choice for entrepreneurial activities, in most cases, they initially recognize the market gap there. The entrepreneurial intention itself is an indicator of the psychological dimension, so RMWs’ cognition of the economic environment and potential in their hometown affects their entrepreneurial psychology and decision. In migration destinations, the greater the ‘demand for workers like me (RMWs)’ is, the more likely RMWs are to present RHEI. Actually, the urgent demand for migrant workers at present is an indicator that shows the trend of the RMWs’ backflow as a result of the development of their hometowns and means the increasing cost in labor-intensive industries of migration destinations. As a consequence, such a situation encourages RMWs to return with the tide to engage in entrepreneurship if they have entrepreneurial intention and catch the gaps and opportunities.

In the dimension of institutional embeddedness, HII, DII and DFI pass the significance test at the 0.01 level, and HFI fails the significance test. The coefficients of HII and DII are positive, and the coefficient of DFI is negative. This partly confirms Hypothesis 3, which indicates that formal and informal incentive institutions for entrepreneurial activities have a positive relationship with RHEI. The impact of DII (OR = 4.118) and HII (OR = 3.198) on RHEI ranks second to HECO and TEC, indicating that the informal institutional environment has a strong incentive effect on their RHEI, especially the informal institution formed by the acquaintance with a successful entrepreneurial role model and the entrepreneurs around.

The phenomena of ‘In my hometown, there are many people who have engaged in entrepreneurial activities’, ‘There are many people around me who have engaged in entrepreneurial activities successfully, and I want to do the same’ reflect the entrepreneurial atmosphere and the entrepreneurship-oriented culture in RMWs’ hometowns and migration destinations. These informal institutions give RMWs positive hints and guidance, stimulating their RHEI. The fact of peer RMWs’ successful entrepreneurial activities enhances RMWs’ self-confidence in entrepreneurship and stimulates their entrepreneurial psychology. Although this informal institution is different from the norms and coercion of the formal institution, its influence on the will of the psychological dimension is strong, even stronger than the function of the formal institution.

The negative impact of DFI (OR = 0.371) indicates that the effect of ‘entrepreneurship-oriented training or implemented policies for migrants’ entrepreneurship’ organized by government is negative to RMWs’ RHEI, which may emphasize the norms and skills suitable for knowledge-based entrepreneurship in cities. Such an institution is usually designed to improve the vocational skills of RMWs through training, making them more responsive to the needs of the urban market, and is not conducive to their RHEI. The incentive effect of HFI in the model is not significant, which shows that the effect of local formal institutions on encouraging entrepreneurial activities in the hometown is limited. Anhui Province launched the project ‘Back Phoenix: Return of Talent’ to attract and encourage RMWs to start their own businesses in the hometowns, and few RMWs enjoyed the benefits of such encouragement policies in the early stages, except for a couple of role models and successful returners who obtained timely access to the project. It is no doubt that such encouragement policies and industrial infrastructures (such as industrial parks for returning entrepreneurship) set up role models and better the informal entrepreneurial atmosphere in the local.

In the dimension of social embeddedness, DSOC (OR=2.320) passes the 0.01 level of significance, and the coefficient is positive, but HSOC fails the significance test. This shows that the social relationship of RMWs in migration destinations is positively related to their RHEI, and Hypothesis 1b is supported. This indicates that the expanding and maintaining social connections that RMWs owned helped them form RHEI. DSOC has a positive incentive effect on RMWs’ RHEI, which shows that the social connections set up or maintained after moving to migration destinations are conducive to RHEI. Social connections are opportunities to expand social relations at their destinations, helping them establish broader social ties, such as workers’ relations, and accumulate more social resources. The exchange of information through these ties can bring more possibilities for entrepreneurship but also opportunities for RMWs to accumulate capital and have a positive impact on their RHEI. The result of the HSOC differs from Hypothesis 1a, indicating that RMWs’ maintenance of social connections in their hometown does not significantly affect their RHEI. Long-term migration leads to RMWs’ de-embeddedness in their hometown, and their quondam social relations are not significant.

The predictive accuracy of Model II is 86.8%, 0.3% lower than that of Model I. The result indicates that after geo-location variables are introduced, the model is not improved. Neither HLO nor DLO pass the significance test in Model II, showing that the location of neither the rural hometown nor the migration destination is statistically related to RMWs’ RHEI.

However, we cannot completely deny the impact of location. The P (Chi-square) of HLO and DLO are both 0.000, indicating a significant difference between different sample categories. Therefore, further distinguishing between the samples based on location is made, as shown in Table 6. From the perspective of their hometowns, 61.72%, 60.45%, and 63.92% of RMWs in northern Anhui, central Anhui, and southern Anhui, respectively, have RHEI. Compared with the RMWs from southern Anhui, those from central Anhui and northern Anhui had lower RHEI. Social and economic development is better in southern Anhui than in central Anhui and northern Anhui, with better industrial development and more enterprises. Therefore, RMWs may be more willing to return and engage in entrepreneurial activities. RMWs from central Anhui and northern Anhui have fewer opportunities to return for entrepreneurship or work, and the situation in their hometowns is probably not as good as that in their migration destinations. In Anhui Province, the difference in the location of RMWs’ hometowns entails differences in economic level and resources available, which affect the economic environment for entrepreneurial activities after their return. These aspects affect the psychological expectations of migrant workers regarding occupation.

From the perspective of the migration destinations, on the one hand, RMWs tend to choose cities in eastern China when they go out for work, which accounts for 59.70% of the total sample (Table 6), representing a higher proportion than those working in central and western China. On the other hand, RMWs working in eastern China have the highest proportion of RHEI (63.33%). In eastern China, RMWs mainly migrate to four provinces and cities, namely, Zhejiang Province, Jiangsu Province, Shanghai City, and Guangdong Province. Among them, the proportion (accounting for the same category) of RMWs with RHEI who move to Jiangsu, Shanghai and Guangdong is higher than the overall level in eastern China, at 66.67%, 76.00% and 70.83%, respectively. In addition, the coefficient of DLO is negative in Model II, indicating that a higher proportion of RMWs with RHEI is higher among those who work in eastern China, while the proportion is lower in central and western China. More specifically, although RMWs’ RHEI is not statistically related to the economic level of eastern China, central China and western China, it may be related to specific cities. There are still differences in the development of different regions in China. The better the social, economic, and institutional environment of destination cities RMWs move into, the higher their entrepreneurial intention will be.

Therefore, when considering the geographical location, attention should be paid to the transformation of different scales. In this paper, the environmental impact is significant at the small scale of the city (county) but not at the regional scale.

Much of the literature on international migrant entrepreneurship focuses on mixed embeddedness in the social, economic, and institutional contexts of host countries while neglecting their entrepreneurial motivation and embeddedness in the context of their home countries, especially at the mesoscale of regions. The research on rural entrepreneurship in Western countries attaches importance to external linkages with other regions in integration with those within local communities in terms of double-layer embeddedness; however, it does not concern migration and does not discuss institutional factors. The case of Anhui, China, proved that the modified framework in combination of mixed embeddedness with double-layer embeddedness was suitable to analyze RMWs’ RHEI, but it did not make a comparison of RMWs between different job types. The job types affect the quality and quantity of RMWs’ established networks and the social, financial and institutional support they can obtain, which in turn affect their embeddedness in the migration destinations and even in their original places. Therefore, a wider survey including more provinces and RMWs’ classification is needed for further research to test the robustness of this result, and in-depth interviews are worth further explaining the interaction of three dimensions of mixed embeddedness into original places and migration destinations, which probably enriches the connotation of ‘mixed embeddedness’ in immigrant entrepreneurship research.

Moreover, this article indicates that the increasing demand for skills in migrant destinations as a result of regional industrial upgrading is conducive to the accumulation of migrant workers’ human capital, which also drives some RMWs with lower professional skills back home for decreasing job opportunities or limited career development. It is still unclear whether necessity-oriented entrepreneurship would increase. In addition, the increase of market opportunities in the central and western regions attracts RMWs to return home and inspires some RMWs’ entrepreneurial intentions. Under such circumstances, it is unsuitable to use the dichotomy of ‘winner’ and ‘loser’ to treat return-home RMWs and to evaluate their intention to start businesses at home. On the whole, such great changes in the regional social and economic development of China are also supposed to be an important context of entrepreneurial intentions and activities, which is also a response to the contextualization of Chinese entrepreneurship research, as Kloosterman et al. (1999) suggested in terms of entrepreneurship research.

Migration is beneficial to accumulating RMWs’ professional skills or human capital and is thus more conducive to their entrepreneurial intention. On the basis of the case study of Anhui, a middle province of China, this article attaches the importance of mixed embeddedness with double-layer embeddedness to RMWs’ entrepreneurial intention. The following conclusions can be drawn. First, compared with migration destinations, RMWs’ hometowns, original places, exert stronger effects on their RHEI. This is partly because RMWs can more easily obtain access to entrepreneurial resources (such as financial capital and market information) in their hometowns than in their migration destinations, which also implies that it is not easy for RMWs to embed themselves into migration destinations. However, the role of migration destinations can not be ignored. Actually, the better the social, economic, and institutional environment of migration destinations RMWs moved into, the higher their entrepreneurial intention they will have when returning to their hometowns, probably for greater benefit for their accumulation of social, financial, and human capital and their cognitive proximity to the entrepreneurial culture there.

Second, the market opportunity in hometowns is the most important factor affecting RMWs’ RHEI. It shows that burgeoning business opportunities in the local market encourage RMWs’ returning decision-making and RHEI, and it also cultivates potential local entrepreneurs. In this case, RMWs with RHEI are supposed to be potential opportunity entrepreneurs instead of necessity ones, and they tend to seek business opportunities in the local market. In this sense, the new generation of return-home RMWs with strong entrepreneurial intention and the capability to identify market opportunity are supposed to bring about the prosperity of local economies in underdeveloped areas in China.

Third, the institutional dimension, especially entrepreneurial atmosphere and culture as the informal institution in the two places, strongly promotes RMWs’ entrepreneurial intention. Successful role models as entrepreneurs in the two places are sources of positive entrepreneurial perception for RMWs, and peers’ and friends’ entrepreneurial activities positively increase the effect of entrepreneurial culture on inspiring RMWs’ RHEI. This suggests that successful entrepreneurs magnify the positive incentive effect of local entrepreneurial culture, especially that of migrant communities, and further enhance RMWs’ entrepreneurial intention. Moreover, local encouragement policies as formal institutions in the hometown are supposed to provide RMWs with economic and social support to set up firms and improve the entrepreneurial atmosphere or culture. However, compared with informal institutions, formal institutions have limited effects.