Introduction

After the Chinese labor market transited into the Lewis turning point, although the absolute income (i.e., wages without considering the cost of living in the city) of migrant workersFootnote 1 has increased (Yang & Zhang, 2014), their relative income and satisfaction level has continued to decline (Tian, 2017). Thus, researchers observe an inverse growth trend; thus, a proportion of the migrant workers are subjected to relative poverty and possess relatively low capital stock (Guo et al., 2018). Consequently, in the Chinese society, justice and equality is compromised (Li & Sicular, 2014). Therefore, for the government, how to effectively enhance the income of migrant workers is a crucial concern.

To maintain social justice and narrow the gap between urban and rural areas, the Chinese government has widely promoted vocational training and retraining (Kinglun, 2012). The State Council has issued the Guiding Opinions on Further Improving the Training of Migrant Workers and the Vocational Skills Upgrading Program for the New Generation of Migrant Workers (2019–2022), whereas the Spring Tide Action and other training and service programs for migrant workers have been implemented in succession, with training courses that target a variety of trades (e.g., Chinese cooks, housekeepers, couriers, and electricians). The training time is determined by the number of lessons, and the length of study required varies for different types of work and different grades (beginner–medium–advanced), ranging from 80–700 lessons (45 min/lesson), usually for approximately 1 month. These vocational trainings in the form of government endorsement have enabled more migrant workers to become highly skilled laborers and industrial workers with stable employment (Kettunen, 1997; Li, 2010), and labor income is more secure. Therefore, does the GPVST exert an impact on the income of migrant workers? If so, what is the impact mechanism? The answers to these questions have a direct bearing on the formulation and implementation of subsequent relevant policies.

With regard to the assessment of the income effect of training, domestic and foreign scholars have performed immense research. Foreign scholars have been discussing this phenomenon since the 1980s and 1990s, and some studies have measured the return on training at approximately 5% (Blundell et al., 1999). When refined, it has been observed that training effectiveness has been related to gender, training duration, and training content (Booth, 1991; Green et al., 1996). However, most of the foreign studies have been conducted on unemployed groups or general laborers (Biewen et al., 2014; Riphahn & Zibrowius, 2016), and less on immigrant groups; therefore, it is not possible to apply their conclusions to extrapolate the effectiveness pertaining to the implementation of the Chinese government’s training.

Conversely, in the Chinese study, there are two views on whether vocational training was able to raise the income of migrant workers. On one hand, scholars postulate that vocational training positively affects the income of migrant workers. Luan (2022), who utilized data from the 2019 China Five Provinces Migrant Worker Survey and unconditional quantile regression, confirmed that vocational training increased the wage rate of non-farm employment for migrant workers, and this observation has been confirmed by other scholars (Greenhalgh & Stewart, 2010; Wang, 2023). On the other hand, it has also been stated that the content and quality of vocational training are often constrained by the employment characteristics of migrant workers as well as the budget, and that the return to training is low (Li et al., 2017), or even insignificant, thereby impacting adults’ hourly wages (White et al., 1996; Zhou et al., 2016).

The controversy relates both to differences in theoretical understanding of the reality of how vocational training affects the income of migrant workers and to differences in empirical research methodology and data selection. Specifically, in terms of theoretical frameworks, different scholars have various understandings of how vocational training affects the income of migrant workers, not only using different theoretical elements (e.g., Life-cycle Theory, Human Capital Theory, etc.; Shi et al., 2021; Wang et al., 2022) to construct theoretical frameworks but also distinguishing between different types of training and different ages of migrant workers in terms of research perspectives (Liu et al. 2015; Yin 2021). Further, from an empirical point of view, the migrant workers’ participation or non-participation in GPVST is a choice effected by individuals based on the comprehensive judgment of various factors, which is typically a self-selective behavior, and will be affected by some unobservable factors, such as the family’s social relations and personal development motivation (Faltermeier & Abdulai, 2010), and whether or not to deal with the issue of self-selection of the sample also has an important impact on the results. Meanwhile, these studies exhibit both nationwide or regional secondary data and primary data from the researchers’ surveys (Zhao & Geng, 2020); both the studies target the older generation of migrant workers and focus on the newer generation of migrant workers (Lu & Cui, 2023).

For a long time, existing studies focus more on the whether or not GPVST has an income-enhancing effects (Yang, 2017; Zhang et al., 2022), with more attention dedicated to the role exerted by human capital between training and IL traditionally. IS is also a crucial factor for judging the following: the effectiveness pertaining to the implementation of vocational training. However, scholars possess limited knowledge on the role of vocational training in influencing IS. On the mechanism of influence, the role exerted by other livelihood capitals between training and income is also ignored, failing to explore the multidimensional income-enhancing mechanisms involved (Chen & Liu, 2016).

Therefore, this study takes migrant workers as the research object, utilizes the 2016 and 2018 CLDS panel data to construct two periods of panel date, and applies PSM–DID to analyze the role of GPVST in influencing the IL and IS of migrant workers from the subjective and objective levels. Based on the livelihood capital theory and using the mediation effect model to analyze the influence mechanism of GPVST on the income of migrant workers, it is expected to reflect the real value and significance of GPVST in a more comprehensive manner, and to provide theoretical support and empirical evidence for enhancing the income of migrant workers and narrowing the urban area–rural area income gap.

Theoretical framework

For most workers, especially disadvantaged groups such as migrant workers, the level of individual income, which refers to the labor remuneration received by an individual after a labor relationship with an employer, mainly depends on the level of their human capital (Mincer, 1974; Ali & Ahmad, 2013), and education and training are the most crucial methods of enhancing human capital (Wang & Li, 2009). However, due to specific constraints, namely economic conditions and working hours, the difficulty with which individuals systematically receive formal education is increased, and vocational training, with its flexible schedule and varied forms, has become the most realistic and feasible option for individuals to enhance their human capital.

Therefore, why can GPVST raise the IL of migrant workers? The human capital theory, in neoclassical economics, postulates that differences in the process and outcome of employment pertaining to the labor market depend to a large extent on the amount of a worker’s human capital. An increase in human capital can significantly enhance a worker’s productive capacity, thereby creating higher marginal returns in regard to economic earnings (Becker, 2009). GPVST, as an educational remedy, not only possesses the relevance and immediacy to reduce employment search time (Alfonsi et al. 2020), but will also incentivize or constrain mechanisms to effectively promote the upward mobility of migrant workers in the internal labor market, with higher returns to their labor (Ibarrarán et al., 2012). Therefore, this study proposes the following hypothesis:

H1a: GPVST exerts a positive effect on the IL of migrant workers.

Compared to the absolute IL, IS represents the subjective evaluation of an individual who compares his or her income with the required income or subjective poverty line, the income of the reference group, the deserved or expected income, and the past income (Zhang & Zhang, 2016). In a develo** country like China, where the system is in transition, the rapid development of industrialization and urbanization has kept away from the social reproduction model that relies on intergenerational relationships to pass on knowledge and skills (Catherine & Marieke, 1997), and the acquisition of relevant skills and qualification certificates through training releases effective human capital signals and enhances their bargaining power and ability to safeguard their own rights and interests, through which they can obtain wages and remuneration consistent with their expected returns (i.e., the enhancement of IS). Based on the preceding analysis, the following research hypotheses are proposed.

H1b: GPVST exerts a positive effect on the IS of migrant workers.

Based on the analysis of the internal logic of GPVST and migrant workers’ income, the mechanism of influence therein requires further attention. Since the mid-1990s, a number of international development research institutions and non-governmental organizations, on the basis of summarizing the theory and practice of poverty alleviation, have proposed the livelihoods concept, with empowerment and capacity as the main elements. Livelihoods is a relatively broad concept, and the currently widely adopted livelihoods concept is the three-module definition of capabilities, capitals, and activities proposed by Chambers, Conway (1992), which focuses on the link between assets and choices possessed in practice, and on the basis of which different actions are pursued to generate the level of income required for survival (Ellis, 2000).

Based on Chambers’ definition of livelihoods, the Sustainable Livelihoods Analytical Framework (Fig. 1) developed by the UK’s Department for International Development (DFID) articulates how individuals and households can pursue different livelihood strategies through a range of livelihood capitals in the context of a given formal and informal system, which in turn influences different livelihood outcomes towards sustainable livelihoods. The framework comprises five components: vulnerability context, livelihood capital, structural and institutional transformation, livelihood strategies, and livelihood outputs. The livelihood pentagon, consisting of livelihood capital, is at the center of the framework and is subdivided into five types: human capital, physical capital, financial capital, social capital, and natural capital. The framework perceives individuals as follows: objects that survive or make a living in a context of vulnerability in which they have access to a certain amount of capital as a means of achieving desired outcomes and meeting their livelihood goals. Currently, the framework has been widely applied to the research and implementation of rural sustainable development issues such as livelihood diversification (Smith et al. 2001), livelihood security (Singh and Hiremath, 2010) and livelihood resilience (Sun et al. 2023) for farming households.

Fig. 1
figure 1

Sustainable livelihood framework.

Full size image

Chinese migrant worker households are farming households that have opted for the following important livelihood choice: going to work in the city. Therefore, the analytical framework depicted in Fig. 1 is principally also applicable to the analysis of migrant workers’ issues, and can provide the basic concepts and methods for analyzing the internal mechanisms of how GPVST affects migrant workers’ incomes. Based on the DFID Sustainable Livelihoods Framework, this study optimizes and integrates the GPVST framework—livelihood capital—income change (Fig. 2). As illustrated in Fig. 2, after receiving GPVST, the five livelihood capitals are directly or indirectly affected to varying degrees, further affecting the income status of migrant workers (i.e., changes in the livelihood capitals ultimately affect the livelihood output outcomes of migrant workers). The specific analysis is depicted as follows.

Fig. 2
figure 2

Analytical framework for GPVST–Livelihood Capital–Income Change.

Full size image

(1) Natural capital refers to the reserves of natural resources that farmers own or may own, such as the area and the quality of farmland, and the degree of fine fragmentation of arable land (Zhai et al. 2012), and the factor responds to the individual migrant worker’s ability to acquire information, agricultural production capacity, ability to choose a job, and ability to allocate resources. Generally, when the education level is higher, the individual possesses more skills and more optimal health, and they are more likely to receive higher labor compensation (Gong et al. 2019; Hu & Kang, 2021; Fahad et al. 2022). To compensate for the innate gap between education and its returns, GPVST has become a crucial method of compensating for the unequal treatment of migrant workers with low educational attainment, which can effectively strengthen the human capital signals of migrant workers and alleviate the problem of asymmetry in the labor market. Based on the preceding analysis, the following research hypotheses are proposed.

H2d: GPVST can enhance the income of migrant workers by increasing their human capital in the inflow area.

(5) Social capital refers to the available social resources through which different livelihood strategies, including the social networks and associations in which individuals participate, can be realized. Existing research indicates that social capital exerts a significant impact on the increase in occupational choices and the probability of identifying relatively higher-paying jobs (Montgomery, 1991; Gao et al. 2021; Yang et al. 2022). Integrative social capital (geographic ties) facilitates the formation of mutual social support (Bian & Huang, 2009) and the transfer of information (Zhao, 2003), factors for obtaining social and economic support, which in turn leads to income enhancement; by contrast, transversal social capital (the extension of social networks) can lead to cooperation between groups that formerly belonged to different social networks, and enhance the overall wage negotiation ability and IL of migrant workers. As a socialized collective activity, GPVST not only reinforces the existing integrated social capital of migrant workers, but also encourages migrant workers to cooperate more with local residents, and to cross over between different social networks and, thus, accumulate transversal capital. Therefore, the following research hypotheses are proposed.

H2e: GPVST can enhance the income of migrant workers by increasing their social capital in the inflow area.

In summary, GPVST affects the natural, physical, financial, human, and social capital of migrant workers by facilitating the transfer of agricultural land, enhancing employment stability, expanding financial financing channels, compensating for the inherent deficiencies of education, and integrating social networks; In regard to the relationship between types of livelihood capital and income, human capital is the most critical factor that contributes to income inequality (Sehrawat & Singh, 2019), whereas social capital such as social networks and resource sharing (Zhang & Gao, 2020; Huang & Fang, 2021) and the financial capital that can be mobilized for disposable income (Chen et al. 2022) also significantly affect income. Both material wealth accumulation and natural resource endowment exhibit a diversified tendency to influence the income of migrant workers (Wanma & Yang, 2011). Consequently, GPVST affects changes in the different types of livelihood capital of migrant workers, which in turn deeply affects their income status, ultimately resulting in different impacts of GPVST on IL and IS.

The research contribution of this study, which differs from traditional research perspectives, the theoretical level, the research not only focuses on the impact of GPVST on the objective IL of migrant workers, but also notes the crucial role of subjective IS, and aims to accurately and comprehensively measure the effect of GPVST implementation from both the subjective and objective perspectives; furthermore, on the traditional basis of human capital, in the research content according to the DFID sustainable livelihood framework through which the analytical framework utilized herein is constructed, the study focuses more on other livelihood capital changes occasioned by the impact of the role; thus, the role of the GPVST on the mechanism of the migrant workers’ income is clarified. At the application level, GPVST exerts a direct effect on income, but also from the dimension of livelihood capital structure change in-depth examination of the indirect effect between them, should attract the attention of policymakers; we should focus on not only the income effect of GPVST but also the enhancement of the multi-dimensional capital of migrant workers; thus, with regard to the income-generating effect, a more direct and effective transformation is expected.

Data and methodology

Methods and models

PSM–DID

As analyzed in the previous section, migrant workers’ participation or non-participation in GPVST is a self-selective behavior, and there may be problems of reverse causation and omitted variables with respect to income, which can lead to biased results. Therefore, this study assumes that the omitted variables are non-time-varying, and selects two-period panel data for empirical estimation using the fixed-effects method.

Since 2010, the pilot vocational training program for migrant workers has been widely promoted within China, and has achieved remarkable results; however, there are still problems such as the lack of integrated planning of training programs, the efficiency pertaining to the utilization of funds, the low quality of training, and the insufficiently perfect monitoring and control mechanisms (Zhang et al. 2011; Du & Zhang, 2012).

To enhance the skills level and employability of migrant workers and promote the transfer of rural labour to non-agricultural industries and cities and towns, under the guidance of the central policy, the Chinese government formulated a new round of the National Migrant Worker Training Program in 2015, and provinces and municipalities launched government-provided or government-subsidized vocational skills training programs for migrant workers.

Herein, 2016 is set as the initial period and 2018 as the intervention period, and the treatment group comprises the migrant workers who received GPVST in 2016, whereas the reference group consists of the migrant workers who did not receive GPVST. To fulfill the basic premise pertaining to the common trend assumption of DID (Liu and Zhao, 2015), this study adopts the PSM–DID proposed by Heckman et al. (1997) to assess the impact effect of GPVST on the income of migrant workers. The model is set up as follows.

$${Y}_{{it}}={\beta }_{0}+{\beta }_{1}{D}_{i}+{\beta }_{2}T+{\beta }_{1}{D}_{i}\times T+\sum {\beta }_{x}{Control}+{Z}_{i}+{\varepsilon }_{{it}}$$
(1)

where Yit represents an explanatory variable measuring the income status of migrant workers, with subscripts i and t denoting the ith migrant worker and year t, respectively. Di distinguishes between the control and treatment groups, T distinguishes whether or not they participate in GPVST, and the cross-multiplier term, Di × T, denotes the core explanatory variable that measures whether or not migrant worker i participates in GPVST. Control denotes a series of control variables including gender, age, marital status, household registration, education, and mobility experience. Zi controls for individual fixed effects that do not vary over time, and εit denotes a random error term.

The average treatment effect according to PSM–DID can be estimated using Eq. (2).

$${{ATT}}_{{PSM}-{DID}}\,=\,E\left({Y}_{18}^{T}-{Y}_{16}^{T}|{C}_{16},D=1\right)-E\left({Y}_{18}^{C}-{Y}_{16}^{C}|{C}_{14},D=1\right)$$
(2)

where T and C denote the treatment and control groups, respectively. Y16 and Y18 denote the values of the explanatory variables before and after the participation in GPVST, respectively. D denotes the dummy variable for whether or not to participate in GPVST, and C14 denotes the initial characterization variables pertaining to the sample of migrant workers controlled in the PSM process with the provincial dummy variable. Meanwhile, we analyzed the sampling scheme of the CLDS data survey, mastered the sampling process, determined the final sample structure, and introduced the sample weights into the model test of this study.

Parallel multiple mediation model

According to the theoretical mechanism analysis, GPVST may affect the income status of migrant workers by enhancing their livelihood capitals. Therefore, we set up parallel multiple mediation effects to verify the mechanism of GPVST affecting the income of migrant workers, and the model is constructed as follows.

$${Y}_{i}={\delta }_{0}+{\delta }_{1}{X}_{i}+{\delta }_{2}{C}_{i}+{\varepsilon }_{1}$$
(3)
$${Z}_{i}={\gamma }_{0}+{\gamma }_{1}{X}_{i}+{\gamma }_{2}{C}_{i}+{\varepsilon }_{2}$$
(4)
$${Y}_{i}={\lambda }_{0}+{\lambda }_{1}{X}_{i}+{\lambda }_{2}{Z}_{i}+{\lambda }_{3}{C}_{i}+{\varepsilon }_{3}$$
(5)

where Yi denotes the income status of the ith migrant workers; Xi denotes whether the ith migrant workers participates in GPVST; Zi denotes the mediating variable of the ith migrant workers; and Ci denotes the other control variables affecting the participation or non-participation of the ith migrant workers in the GPVST, the mediating variable, and the income status. δ2,γ2, and λ3 denote the coefficients of the rest of the control variables, and εi denotes the random error term. (3) δ1 denotes the total effect of participation in GPVST affecting the income status of the ith migrant workers; γ1 in Eq. (4) denotes the effect of GPVST on the mediator variable; and λ1 and λ2 in Eq. (5) denote the direct effect of GPVST and that of the mediator variable on the income status of the ith migrant workers. Substituting Eq. (4) into Eq. (5), we can obtain the mediating effect of GPVST (i.e., the indirect effect of GPVST on the income status of migrant workers) using the mediating variable.

Samples and data

The data utilized herein was obtained from the CLDS, which targets the working-age population between the ages of 15 and 64 years old. The study utilizes a probability sampling method that is proportional to the size of the labor force in a multi-stage, multi-level manner, and it applies a rotating-sample tracking method; moreover, it covers 29 provinces in China (with the exception of Hong Kong, Macao, Taiwan, Tibet, and Hainan). Because the CLDS questionnaires in 2016 and 2018 added new elements such as GPVST, the research needs of this study are fulfilled. Therefore, this study takes 2016 and 2018 as the two phases of implementing GPVST, respectively, and constructs balanced panel data for the two periods.

According to the content of the 2016 and 2018 CLDS questionnaires and related instructions, the samples of the experimental group were mainly from migrant workers who had completed the GPVST before the 2016 CLDS questionnaire research. Respondents can participate in one and more training programs. However, due to the limited amount of data available in the questionnaire, only the most basic distinction can be made, i.e. whether or not they have participated in GPVST.

To obtain balanced panel data, this study retains only the individual migrant workers who were surveyed in both 2016 and 2018 (i.e., 16–55-year-old females and 16–60-year-old males with an agricultural household registration, who have been away from their families for more than 180 days in a year, and who are currently engaged in non-agricultural jobs; Zhang & Lu, 2009); furthermore, the study deletes the sample of students who are enrolled in school and those who have worked for a number of days equal to zero, and who are missing some crucial information (e.g., education level, gender, wage level, and participation in training). After removing core variables with missing values, two-period balanced panel data with a 1823 sample size are obtained; the control group is 1686, and the treatment group is 137, accounting for 92.48% and 7.52%, respectively. This observation is not only consistent with the proportion of migrant workers participating in vocational training in 2016 as announced by the State (the total number of migrant workers in China was 281.71 millionFootnote 3, of which 21 million received vocational trainingFootnote 4, a proportion of 7.45%), but is also basically consistent with the statistical results documented in the relevant literature (only approximately 9% of migrant workers have received training; Shi et al. 2021), and therefore exhibits satisfactory representation.

Variable selection and descriptive statistics

The selection, definition, and assignment of variables, which are depicted in Tables 1–3, include four categories, namely explained variables, explanatory variables, control variables, and mediating variables.

Table 1 Variable Selection, Definition, and Assignment.
Full size table
Table 2 Indicators for measuring the livelihood capital of migrant workers.
Full size table
Table 3 Descriptive statistics.
Full size table

Explained variables

The explained variable, herein, is the income status of migrants. Based on existing studies (Ning et al., 2022), this study utilizes IL and IS to measure the income status of migrant workers. CLDS asks each respondent for the total of all types of income for the year (e.g., agricultural income, wage income, and business income) and evaluates the current level of income through an integer between 1 (very dissatisfied) and 5 (very satisfied).

Explanatory variables

The explanatory variable is whether individual migrant workers have participated in GPVST. CLDS asks each migrant worker whether he/she has received GPVSTFootnote 5 (e.g., cook, elderly caregiver, and electrician), and indicates that the migrant worker has participated in GPVST if the individual answers yes.

Control variables

In regard to estimating the propensity score, this study controls a series of initial characteristic variables (Xi) that affect whether individuals participate in GPVST or not, which are obtained from the individual perspective of migrant workers and refer to the related literature (Shi & Sun, 2019; Li & Li, 2022), and the variables mainly include gender, age, marital status, etc. In addition, assuming that the income status of migrant workers and the probability of obtaining GPVST are influenced by the city level, this study also controls for city-based regional dummy variables in the empirical process.

Mediating variables

Herein, under the framework of the DFID sustainable livelihood theory and based on the existing research results on the quantification of livelihood capital, the indicators through which the livelihood capital of migrant workers can be measured are selected from five dimensions, namely human (Wu, 2023), social (Yang & Yang, 2022; Zhang & Zhao, 2023), physical (Wu & Wu, 2021; Zhang et al. 2023), natural (Li & Chen, 2023; Gao & Sun, 2023), and financial (Li, 2016) capital, and the measurements methodology is mainly based on the research results of studies conducted by researchers including Sharp (2003) and Dong & Yan (2023), and mainly adopts the entropy method to weight the 15 indicators and obtain the weights of each indicator. The specific index system is illustrated in Table 2.

Results

Descriptive statistics

Table 2 depicts the descriptive statistics of the dependent and control variables in the reference and treatment groups. Regardless of the year, the IL of migrant workers who participated in GPVST (treatment group) were higher than those who did not participate in GPVST (reference group), with a 0.28 and 0.15 increase in the mean of the logarithmic value pertaining to the IL of the treatment group and that of the reference group, respectively (differential result: 0.13). For IS, the mean value pertaining to the IS of the treatment group increased from 2.98 to 3.04 (a 0.06 increase), whereas the mean value pertaining to the score of the reference group decreased from 2.93 to 2.96 (a 0.03 increase); migrant workers who participated in GPVST exhibited a more considerable IS increase than those who did not. The result apparently verifies the following theory: participation in GPVST facilitates the enhancement of migrant workers’ income status.

Sample matching quality test

The common support hypothesis and balance test are prerequisites for applying the PSM method. Herein, the propensity score common support domain bar chart and the kernel density estimate are provided as a test of the common support hypothesis (Figs. 3; 4), and the results indicate that most of the observations are within the common range of values, with only a small number of sample losses. In the balance test, the standardized deviations of all the control variables after matching are significantly reduced, and their absolute values have been lower than 10%. T-test results indicate that there is no systematic difference between the control variables of the treatment group and those of the reference group after matching, and it is no longer possible to judge whether the migrant workers have participated in the GPVST through the control variables. The Pseudo-R2 value in the overall goodness-of-fit statistic of the model is significantly lower, and the LR statistic is no longer significant, which indicates that the matching results can balance the distribution of the control variables in the two groups of samples more optimally, and the balance test is verified (Chen & Zhai, 2015). Due to the space limitation of the study, the test results will not be repeated here.

Fig. 3
figure 3

Common range of values for the propensity scores for IL (left) and IS (right).

Full size image
Fig. 4
figure 4

Kernel density plot.

Full size image

PSM-DID estimation results

PSM-DID estimates are obtained after standardizing some of the ordered variables. Table 4 indicates that GPVST exerts a significantly positive effect on the IL and IS of migrant workers, regardless of whether control variables are added or not, which indicates that GPVST significantly increases migrant workers’ IL and IS. GPVST significantly increased IL by approximately 14.6% to 34.7% and also significantly increased IS by approximately 6.9% to 9.3%. Thus, hypotheses 1a and 1b are empirically tested.

Table 4 Results of PSM–DID estimation.
Full size table

To ensure the accuracy of the conclusions, this study changes matching methods (1:1 and 1:3 nearest neighbor matching), and it replaces the independent variables for robustness tests (replacing with the number of times of GPVST); the results are basically consistent with the preceding conclusions. In addition, the omitted variables were tested with reference to the methodology proposed by Oster (2019), and the results indicated that the impact of GPVST on the income status of migrant workers is quite stable, and that due to the omitted variables, the impact does not change significantly. The aforementioned empirical analysis confirmed the satisfactory robustness of the conclusions and reconfirmed the research hypotheses 1a and 1b.

Heterogeneity analysis

To further clarify the impact of GPVST on the income status of migrant workers, this study chooses age and hukou as the dividing criteria. The old generation of migrant workers (born in the 1960s or 1970s) and the new generation of migrant workers (born after the 1980s) exhibit large differences in personality traits, life differences, and social identities (Lee and Qi, 2021; Xu et al. 2023), whereas China’s dual household registration system excludes foreign hukou from the enjoyment of urban area right of welfare benefits (Ma & Hu, 2018); thus, a large gap between their willingness and probability (Zhang & Li, 2020) to participate in GPVST is observed (Smith, 2014). Therefore, this study extends Model (3) as follows.

$${Y}_{i}\,=\,{\delta }_{0}+{\delta }_{1}{D}_{i}+{\delta }_{2}T+{\delta }_{3}{W}_{k}+{\delta }_{4}{D}_{i}\times T\,+\,{\delta }_{5}{D}_{i}\times T\times {{W}_{k}}+{\delta }_{6}{{Control}}_{{it}}+{Z}_{i}+{\varepsilon }_{{it}}$$
(6)

where the Wk variable indicates whether it is a new generation of migrant workers or not, and whether it is a local migrant worker or not; the coefficients of the three times interaction term \({D}_{i}\times T\times {W}_{k}\) are of interest herein as a measure of the heterogeneous impact of GPVST on the income status of different types of migrant workers, and the rest of the variable definitions are consistent with those in the previous study.

Based on①-④ in Table 5, compared with the older generation, GPVST exerts no significant effect on the new generation of migrant workers’ IL; however, the coefficient of the interaction term on IS is significantly negative, which indicates that the GPVST mainly enhances the IS of the older generation. This observation may be rationalized as follows: the older generation of migrant workers will gradually reduce their expected goals for career development and income growth as their body function response sensitivity and physical strength decline with age (Chen, 2021).

Table 5 Heterogeneity analysis.
Full size table

⑤-⑧ indicate that the effect of GPVST on IL is more apparent among foreign migrant workers, and that the effect on IS is more apparent among local migrant workers. This relationship is mainly rationalized as follows: the human capital accumulated by foreign migrant workers before mobility can only be partially transformed in a foreign place (He et al. 2015), which will reduce the IL of this group. Meanwhile, foreign migrant workers may be discriminated against by social systems such as household registration and exhibit difficulty in obtaining remuneration that meets their expectations of returns to education, which affects the individuals’ subjective perception of income.

Mediating effect of livelihood capital

Based on the aforementioned empirical results, GPVST can significantly increase the IL and IS of migrant workers; however, what is its mechanism of influence? In conformance with the theoretical analysis in Section II, GPVST exerts an impact on IL and IS by influencing migrant workers’ livelihood capital. To verify this mechanism, the KHB method proposed by Karlson et al. (2011) is utilized herein; thus, the mediating effect paths between the independent and dependent variables are revealed, and the specific results are depicted in Tables 6 and 7. Models ①-⑤ test the mediating effect of different livelihood capitals, and Model ⑥ tests the mediating effect of livelihood capital as a whole.

Table 6 Impact mechanisms (IL).
Full size table
Table 7 Impact mechanisms (IS).
Full size table

Table 6 illustrates the impact mechanism between GPVST and the IL of migrant workers. The total effects of models ①-⑥ indicate that with regard to controlling for other variables, the IL of migrant workers who have received GPVST is 18.2% to 18.3% higher than that of other migrant workers. After the introduction of the mediating variable related to livelihood capital, model ⑥ indicates that livelihood capital as a whole exerts a significant mediating effect, explaining 63.93% of the GPVST effect. Specifically, when introduced into the model alone, the mediating effects of human and financial capital are large, explaining 42.08% and 29.51% of the impact of GPVST on the income increase of migrant workers, respectively; furthermore, the mediating effect of social capital is small (accounting for 2.73% of the total), whereas physical and natural capitals do not produce significant mediating effects.

Table 7 depicts the influence mechanism between GPVST and the IS of migrant workers; the results indicates that the total and direct effects of GPVST on migrant workers’ IS are significantly positive, and that the indirect effects of human capital, social capital, and financial capital are also significant; however, physical and natural capital have failed to exert an intermediation role, which proves that the GPVST will indirectly enhance the IS of migrant workers by increasing their human capital, social capital, and financial capital. Specifically, the direct effect of GPVST on IS dominates (59.59%), whereas the indirect effect of the other three livelihood capitals occupies a secondary position with a contribution rate of only 40.41%.

It can be observed that the direct effect of GPVST on the migrant workers’ IS is greater than the effect on the IL, whereas the indirect effects of human capital, social capital, and financial capital should also be noted. So far, Hypothesis 2a, Hypothesis 2c, and Hypothesis 2e have been confirmed.

Discussion

The research utilized the CLDS panel data pertaining to 2016 and 2018, analyzing the impact of GPVST on the objective IL of migrant workers, while also considering the crucial role of IS in measuring the effectiveness of GPVST (Fig. 5). Studies have observed that GPVST exerts a significant income-enhancing effect, mainly related to the fact that GPVST enhances the returns to education for migrant workers, adapts to industrial structure upgrading, and provides more information on employment, a finding that is confirmed by markets in both developed and develo** countries (Winter-Ebmer & Zweimüller, 2003; Chakravorty et al., 2021). On the other hand, GPVST can also enhance the IS of migrant workers. When migrant workers receive GPVST as a means of obtaining certain professional skills, it can not only yield higher IL for individuals and thus enhance their subjective evaluation of income (Kapteyn et al., 2013), but also enable migrant workers to gain a deeper understanding of the industry and enhance the degree of matching between people and jobs; thus, it changes the objective function and expected benefits of migrant workers’ employment, which is then transferred to their work efficiency and the evaluation of overall satisfaction (Hu et al., 2015).

Fig. 5
figure 5

Structural model of GPVST–Livelihood Capital–Income Change.

Full size image

With respect to conforming to the benchmark regression hypothesis, this study, which is based on the livelihood capital theory, dedicating more attention to the effect occasioned by the change of other livelihood capitals on the basis of human capital, adopts the mediation effect model; thus, it analyzes the impact mechanism of GPVST on the income of migrant workers. The results indicate that GPVST enhances the three types of livelihood capital (i.e., human capital, social capital, and financial capital), and, thus, increases the IL and IS of migrant workers; in general, for livelihood capital, the indirect effect pertaining to the effect of GPVST on the migrant workers’ IL is stronger; likewise, the direct effect pertaining to the effect of GPVST on the IS of migrant workers is stronger.

First, based on the human capital perspective, the impact of GPVST on the income of migrant workers belongs to the skill-enhancing effect. It is apparent that the human capital of workers exerts a significant positive effect on their IL. Compared to basic education, which is characterized by high costs and slow results, GPVST can enhance labor skills in the short term; thus, a specialized human capital that exerts a signaling value-adding effect is formed (Attanasio et al. 2017; Abebe et al., 2021). Consequently, return on experience and productivity is increased (Almlund et al., 2011), which effectively contributes to the growth of migrant workers’ income (Zhang & Zhang, 2020; Wei & Lu, 2021) and increased IS (Sahi, 2013; Liu, 2014).

Second, the impact of GPVST on the income of migrant workers is also reflected in the information-interaction effect occasioned by social capital. Migrant workers mobilize social resources to influence income returns through the information-interaction effect achieved by social capital (Coleman, 1990; Hao & Wen, 2013). Migrant workers enter the city; the rupture of native social relations leads to structural tension and imbalance in their social relations; and through GPVST with a common goal, it is easier to form new social relations. Thus, the expansion of social capital is stimulated, which provides a natural supply of contacts (Li, 1996) and a wealth of human resources and information resources, and leverages the interactive effect of this information as a means of achieving increased income (Agulera & Massey, 2003). This notwithstanding, by overcoming income disparity (Grootaert, 1999), reducing the gap between actual and expected income, and increasing IS by enhancing the bargaining power of migrant workers, social capital also exerts a positive effect (Lu and Zheng, 2019).

Third, financial capital exerts a trickle-down effect between GPVST and the income of migrant workers. GPVST enhances the financial literacy of migrant workers while strengthening professional skills training, and also enables them to obtain vocational skills level certificates and, thus, exclusive loans, such as the skills-loan (i.e., an exclusive financial product tailored to the skilled personnel group in Zhangjiagang City, Jiangsu Province), which has increased the benefit of financial resources to low- and middle-income groups. As the availability of migrant workers through credit increases, it not only prompts them to utilize savings or credit services to expand their income acquisition channels (Xu et al., 2017), but also changes the initial endowment of the household, optimizes the inputs of production factors (Meng, 2010), and increases the disposable and fundable cash of the migrant workers (Adetiloye, 2012; Zhang et al., 2018), which exerts a positive impact on narrowing the income disparity between the groups (Greenwood & Jovanovic, 1990), and on mitigating the subjective feelings occasioned by the inequality of incomes.

However, the mediating effect of natural and physical capital is not confirmed by the empirical results of this paper, which may mainly stem from the following three reasons: First, the government utilizes GPVST to address the specific needs of certain types of industries or sectors. For example, Nan**g Municipal Government in Jiangsu Province is vigorously pursuing employment skills training in the areas of new energy vehicle testing and maintenance and drone manipulation; the training is more targeted, and it is unable to satisfy the multifaceted training needs of all migrant workers. Second, since the current exit mechanism and compensated transfer mechanism of natural resources such as land have not been perfected (Yu et al., 2023), direct abandonment for migrant workers may lead to property loss in regard to natural resources. Meanwhile, Chinese farmers are deeply influenced by the Confucian culture of returning to their roots (Chen et al., 2021), and land is the basic guarantee for migrant workers to retire; therefore, the probability of migrant workers retaining natural resources such as land in their old homes is higher. Third, the accumulation of physical capital consumes a long time, and a certain lag effect exists; however, because this study utilizes panel data for a two-year period, it is difficult to reflect the role of the GPVST in influencing physical capital in the short term.

Conclusion and policy implications

This study utilizes CLDS data pertaining to 2016 and 2018 to constructe two periods of panel data, measuring the impact of GPVST on migrant workers’ IL and IS, and analyzes heterogeneity according to age and hukou location. Therefore, with respect to influencing the IL and IS of migrant workers, the mediating role effected by multidimensional livelihood capital in GPVST is explored based on the livelihood capital theory. The results indicate that participation in GPVST exerts a positive effect on the IL and IS of migrant workers; GPVST is more capable of raising the IL of migrant workers, and the training exerts a greater positive effect on the IS of the older generation of migrant workers and local migrant workers. The mechanism analysis indicated that for GPVST, human capital, social capital, and financial capital exert the skill-enhancing effect, the information-interaction effect, and the trickle-down effect; thus, the income of migrant workers is affected.

This conclusion not only explains the impact of GPVST on the objective IL of migrant workers but also theoretically enriches the subjective perspective of impact - IS - and verifies the subjective income effect of GPVST. Simultaneously, based on the livelihood capital theory, this new mechanism, which no previous attempt has been made to analyze, expands the study of the intrinsic mechanism between GPVST and the income of migrant workers and provides a new theoretical explanation of the causes of the changes in the income of migrant workers.

Based on the preceding conclusions, it can be observed that GPVST exerts a considerably positive effect on both the IL and IS of migrant workers, which also indicates that GPVST for migrant workers, which have been actively advocated and supported by the Chinese government, have achieved more optimal results. The vast majority of migrant workers have expressed a desire to receive training as a means of upgrading their skills; thus, they can increase their employment opportunities and work income (Cui & Wu, 2017). However, according to the 2020 Migrant Worker Monitoring Survey Report, the proportion of migrant workers who have received skills training in non-agricultural occupations is low (around 30%) and on a downward trend, which indicates that the training needs of most migrant workers cannot be met. Currently, migrant working is still an important strategic choice for Chinese farmers, and it is only by enabling migrant farmers (migrant workers) to maintain and even strengthen their capabilities and assets in the present, and in the long run, through methods such as vocational training, that it is sustainable, and consistent with the overall Chinese sustainable development goal.

In optimizing GPVST, the government should expand its breadth and depth; by including migrant workers from multiple dimensions and perspectives, livelihood capital can be enhanced.

First, with regard to the construction of the training system, it is not only in regard to breadth that we should encourage the participation of enterprises, social organizations, and educational institutions to make more migrant workers can receive vocational training. In regard to depth, the curriculum should also comprehensively consider economic and social development and changes in the employment scenario; establish a full-cycle training system that is hierarchical, categorized, and phased; and ensure that the content of the curriculum covers the basic skills and knowledge required by rural migrant workers, such as computer operation, safety, and legal knowledge; thus, their needs can be satisfied at different stages of their careers.

Second, in order to ensure the quality and effectiveness of training, on one hand, it is necessary to realize the combination of training and employment, thereby ensuring that migrant workers can not only apply what they have learned, but also provide enterprises with qualified labour; thus, a win-win scenario can be achieved. On the other hand, China should promptly establish an authoritative system for the certification of vocational qualifications and ensure that the standards and procedures for issuing vocational qualifications are open and transparent. The qualifications acquired by migrant workers through their technical skills can not only enhance their competitiveness in employment and the possibility of stable career development, but also offer them priority in borrowing and housing.

Finally, to enable migrant workers obtain more livelihood capital to enhance their IL and IS, the government should also create employment information exchange platforms to provide employment information, vocational training information, services pertaining to the interpretation of policies and regulations, and other services centered on social capital, financial capital, and other forms of livelihood capital, thereby expanding migrant workers’ interpersonal resources, access to information, and other tangible or intangible capital elements that are conducive to their survival and development.

Limitations

This study should be pursued with regard to at least two dimensions. (1) Data. This paper utilizes panel data obtained from the CLDS pertaining to 2016 and 2018. Due to the changes in the external market environment and the upgrading of the industrial structure, it is necessary to utilize the latest data under the CLDS data update to constitute the 2- or 3-period panel data; thus, the main conclusions obtained herein can be further validated. (2) Classification of GPVST. GPVST is generally categorized into skills training and entrepreneurship training; due to data limitations, the analysis process herein mainly focuses on the impact of vocational skills training on the income of migrant workers, and subsequent research can also examine the impact of entrepreneurship training. In further research, it will be worthwhile to develop methods of selecting appropriate theoretical models through which the mechanism between GPVST and migrant workers’ income can be selected, and to develop methods of selecting indicators through which the various livelihood capitals and the impacts they generate can be reflected more optimally.