# The Impact of Education Input on Labor Migration and Inequality in China

## LiXin(李昕)andGuanHuijuan(关会娟)

Abstract: By introducing a general equilibrium framework to China’s dual economic structure, this paper studies the microscopic mechanism of education input to narrow the urban-rural income gap and how to improve the allocation efficiency of education funds in China’s “new normal” economy. The empirical analysis results show that education input is effective in narrowing the urban-rural income gap and achieving a Pareto improvement state in both direct and indirect ways. However, the effect of education input at different stages varies. In particular, the impact of compulsory education on improving inequality is more significant than the other education levels. With adjustment for the macroeconomic slowdown, raising fiscal expenditures on education can improve potential economic growth by promoting human capital accumulation and labor productivity in the long run. In all, education input can promote labor migration and narrow the urban-rural income gap, which is conducive to alleviating the contradiction of the structural transformation lag in employment and achieving inclusive growth targets.

Keywords: education input, the urban-rural Income gap, labor migration JEL Classification Codes: J31; J41

DOI:1 0.19602/j .chinaeconomist.2018.11.07

1. Introduction

During the Twelfth Five- Year Plan period in China ( 2011- 2015), the urban- rural income gap continued to narrow. The ratio of per capita disposable income between urban and rural areas decreased from 3.13 to 1 in 2011 to 2.73 to 1 in 2015. Although the relative rate of the urban-rural income gap narrowed year by year, its value continued to expand. On one hand, the urban-rural income gap was 14833 yuan in 2011, and it expanded to 19773 yuan in 2015. The wage income gap, which contributes 75% to the urban-rural income gap, is key. On the other hand, a large number of empirical studies show that China’s income gap during that period, especially the urban-rural income gap, did not fall but continued to expand after considering non-monetary factors such as housing price, education, health care, and social security. Based on Chinese Household Income Project Survey data, Li (2003) finds that China’s urban-rural gap has been the largest in the world if income-in-kind and subsidies are included. Han and Li (2011) show that the urban-rural income gap contributes to more than 50% of the gross

national income gap and is the main factor in China’s inequality. If public health care and unemployment insurance are considered, China’s urban-rural income gap may be the largest in the world. The excessive urban-rural income gap is not conducive to the development of domestic consumption, the adjustment and optimization of economic structure, or the stability of society.

A large number of scholars have studied the influential factors of China’s urban-rural income gap, and have put forward different suggestions from multiple perspectives. Some of them explore the effect of household registration discrimination on the urban-rural income gap from China’s dual economic structure. They claim that household registration discrimination suppresses labor mobility, which expands and solidifies the urban-rural income gap. They argue that the household registration system must be reformed to narrow the gap (Sicular et al. 2007; Wan and Li, 2013). Some scholars agree that the keys to improving the inequality between China’s urban and rural areas are the modification of interregional household registration, economic openness, and the implementation of local governments’ economic policies ( Lu and Chen, 2004). Others believe that the urban- rural income gap has a negative relationship to financial efficiency and therefore, reducing the non-equilibrium of financial development can narrow the gap (Wang and Qiu, 2011). In addition to the macroscopic perspective, a large branch of research uses micro-survey data to find that education may be one of the most important factors in affecting China’s inequality (Sicular et al., 2007; Chen et al. 2010). The Country Diagnostics Report, published by The World Bank in 2016, points out that the contribution of China’s urban-rural income gap to overall inequality increased from 37% in 1988 to 54% in 2007, and the most serious inequality in China is urban-rural income inequality. Education is one of the most influential factors of that inequality.

Education input alleviates the urban-rural income gap both in direct and indirect ways. Increasing education input can not only increase the marginal production and income of labor directly, but also promote labor migration in rural areas by reducing the migration costs and alleviate the income gap indirectly. Cao and Zhang (2015) prove that if the proportion of agricultural employment decreased by 1 unit, then the urban-rural income gap would decrease by about 1.03 units. In 2015, China’s rural population accounted for 43.90% of the total population, and agricultural employment accounted for 28.3% of total employment, while agricultural output only accounted for 9.13% of GDP, and labor productivity in the agricultural sector only accounted for 22.34% of that in the industrial sector. The difference of relative proportions between the rural population, employment, output, and productivity in the agricultural sector shows that labor migration is still an effective way to improve productivity and increase incomes in the agricultural sector, and education may play an important role.

With China’s economy entering the “new normal,” national education development and education fund inputs are faced with pressures due to the decline of fiscal revenues. It is important to further research how to improve the allocation efficiency of education funds at different educational stages and alleviate the urban-rural income gap effectively, which are conducive to reducing development inequality, promoting all social members’ share in economic development achievements, and achieving inclusive growth targets. This paper complements previous studies in the following aspects: It discusses the microscopic mechanism of education input on the urban-rural income gap, referring to both the direct effect and the indirect effect, and performs empirical analysis on how to improve the allocation efficiency of education funds at different educational stages.

The rest of the paper is arranged as follows: Part II constructs a general equilibrium model with the urban-rural dual economic structure and discusses the direct effect and indirect effect of education input on the urban-rural income gap. Part III tests the hypothesis of the relationship between education input, labor migration, and the urban-rural income gap based on the dynamic spatial panel model, and distinguishes the different effects of education input at different educational stages. Part IV concludes.

2. Theoretical Framework

This paper constructs a traditional two-sector model, namely, the household sector and the firm sector. The household sector has the characteristic of intertemporal consumption, and the firm sector has the characteristic of dual structure. On this basis, this paper discusses the microscopic mechanism of education input on the urban-rural income gap.

2.1 Households

Referencing Song et al. (2011), this paper constructs a two-generation Overlapping Generation Models (OLG model).

Assume that each individual has the same preference, and the utility function is a logarithmic utility function with constant-relative-risk-aversion. Given the population growth is exogenous to zero, and the population size is standardized as 1. The individual maximizes utility (1) subject to the budget constraint (2). That is,

2.2 Firms

In the development of China’s economy, resource constraints differently impacted education levels and labor quality of individuals in urban and rural areas. The impact difference was endogenous over a long period of time and, to a large degree, resulted in the urban-rural income gap (Chao and Shen, 2014). Therefore, this paper divides the national economy into the traditional agricultural production sector a and a modern non-agricultural production sector b, with the number of laborers La and Lb respectively. Here, La+ Lb =L, and L is the total labor of the economy.

Let it be given that the traditional agricultural sector is labor-intensive and only employs workers, and the modern non-agricultural sector hires workers and rents capital in competitive factor markets.

To simplify the analysis, assume that there are no adjustment costs or depreciation of capital. The agricultural sector employs low-skilled workers and the production function has decreasing returns to scale. The non-agricultural sector hires skilled workers and the production function has constant returns. The production functions in the two sectors take the form There are a large number of identical firms in the system and each has access to the same product function and takes A as given. The firms hire workers and rent capital in competitive factor markets and sell their output in competitive output markets. The firms maximize profits and pay factors into the marginal products.

The marginal product of labor is

2.3 General Equilibrium Framework

In developing countries, there is a surplus of low-skilled labor in the traditional agricultural sector and a relative shortage of high-skilled labor in the modern sector. Assume that labor productivity and the wage level in the non-agricultural sector are both higher than that in the agricultural sector. According to the dual economic model (Lewis, 1969; 1979), the high wage will entice workers to migration from the traditional sector to the modern one and increase their wage income. The marginal product of labor and the wage income will show a reverse trend with the change in the number of laborers between the two sectors. Therefore, the labor migration will gradually narrow the urban-rural income gap in theory. However, in practice, the labor migration is based on the premise that the low-skilled workers improve their human capital and labor productivity, which involves borrowing costs, time costs, and opportunity costs that hinder labor migration (Harris and Todaro, 1970; Chau, 1997).

This paper assumes that is the cost of labor migration from the agricultural sector to the nonagricultural sector. On one hand, education input improves labor productivity and income levels. On the other hand, higher labor productivity results in more job opportunities. Therefore, education input can reduce the cost of labor migration. That is, , where is the cost of labor migration from the agricultural sector to the non-agricultural sector.

Each individual has two choices. One is working in the traditional sector as a low-skilled worker; the utility function of . The other is increasing education input and accumulating human capital to work in the modern sector; the utility function takes the form of . Because of the existing cost of labor migration, the choice made by an individual depends on the utility of that individual. When the economic system achieves its steady state, the utility is equal under the two

As mentioned above, the income gap, which contributes to 75% of the urban-rural income gap, is the key to affecting the system inequality. Therefore, this paper defines the urban-rural income gap as the difference between the non-agricultural sector and the agricultural sector.

Thus, the urban-rural income gap is positively related to the cost of the labor migration from the agricultural sector to the non-agricultural sector, and is negatively related to the education input. Education input can narrow the urban-rural income gap effectively.

2.4 Microeconomic Mechanism

This paper introduces two propositions to explore the impact of education input on labor migration and the urban-rural income gap.

Proposition 1: Education input can promote labor migration from the traditional sector to the modern sector

Proposition 1 and proposition 2 illustrate that education input can promote labor migration from the agricultural sector to the non-agricultural sector, and narrow the urban-rural income gap. On one hand, with economic development, the accumulation of material capital increases, and the marginal production of material capital declines in the non-agricultural sector (as shown in equation (7)), while the marginal production of labor increases relatively (as shown in equation (8)). Therefore, in the non-agricultural sector, the increasing demand of skilled labor results in a relative shortage of skilled labor.

On the other hand, the high wage will attract workers to migration to the modern non-agricultural sector from the agricultural sector. With the constant number the gross labor force, labor migration will decrease La and increase Lb , and Wa will increase (as shown in equation (9)) and Wb decrease (as shown in equation (10)) respectively.

As discussed above, education input can improve the accumulation of human capital and labor productivity and directly increase the wage income in both the traditional and modern sectors (as shown in equations (11) and (12)). In general, the education levels and labor productivity of workers are both higher in the non-agricultural sector than in the agricultural sector. Thus, with the assumption of the decreasing marginal production of education input, the effect of education input promoting labor productivity is more significant than that in the non-agricultural sector (as shown in equations (13)).

Therefore, increasing education input is conductive to narrowing the urban-rural income gap.

ways. In Increasing summary, the education education input input alleviates can improve the urban-rural the accumulation income of gap human both capital in direct and the and marginal indirect production of labor and directly increase the labor income in both the non-agricultural and agricultural sectors. However, the effect of education input promoting labor productivity is more significant than that in the non-agricultural sector. Thus, increasing education input can narrow the urban-rural income gap. Besides that, education input can also improve the accumulation of human capital and the labor

productivity of low-skilled workers, which allows them more job opportunities in the non-agricultural sector and reduces the cost of labor migration, which will decrease Wb and increase Wa . Because of the more significant effect of increasing than decreasing, Wb is still showing an upward trend. In addition, because the increasing rate of Wb is lower than that of Wa , education input can narrow the urban-rural income gap.

3. Empirical analysis

Based on the dynamic spatial panel model and the data of 31 provinces from 1995–2014, this paper tests the hypothesis of the relationship between education input, labor migration, and the urban-rural income gap. Moreover, we discuss the special effect of education input on narrowing the urban-rural income gap and how to improve the allocation efficiency of education funds at different educational stages.

3.1 Variables and Data

The main variables in this paper include the urban-rural income gap, education input, and labor migration. The control variables include the degree of openness, financial efficiency, the proportion of state-owned enterprises, and per capita GDP.

(1) The urban-rural income gap (GAP) is the explained variable and is measured by the ratio of per capita disposable income of urban households and per capita net income of rural households. The data are derived from the China Statistical Yearbook. The National Bureau of Statistics of the People’s Republic of China obtained the data from 1995–2012 based on the urban household survey and the rural household survey and estimated the data in 2013–2014 according to the same statistical caliber based on the Urban and Rural Household Survey data. Therefore, the statistical caliber of the data maintains consistency before and after 2012 in this paper.

(2) The education input (edu) at different educational stages is the main explanatory variable in this paper. A large number of studies have shown that educational inequality due to the differences in education input is the most important factor resulting in the widening urban-rural income gap (Bai, 2004; Terry et al. 2007; Chen et al. 2010). Education input at different educational stages has a differing effect on the narrowing of the urban-rural income gap (Becker and Tomes, 1979; Yang et al., 2015). Therefore, this paper tests the effect of education input at different educational stages, such as primary education, junior high school education, senior high school education, higher education, and compulsory education. Data are derived from the China Educational Finances Statistical Yearbook. The compulsory education input is calculated based on the education input at the primary and junior high school levels with the weight on the number of students. The average education input is calculated based on the education input and the number of students at each educational stage.

(3) Labor migration (LTR). According to Liu (2015), the estimated formula for the labor migration rate is (the number of rural employees – the number of rural primary industry employees) / the number of rural employees. The data are derived from the China Agriculture Compilation of Statistics 1949– 2008 and the China Rural Statistical Yearbook, and the Statistical Yearbook of 31 provinces.

(4) Degree of openness (FDI). The Stolper-Samuelson Theorem holds that a country’s goods with comparative advantages will become more expensive after opening to the outside world, which results in a higher return of the factor intensively used on this kind of goods. During the economic take-off period, most developing countries adapt an export-oriented strategy and develop rapidly by improving the labor-intensive industries which increase the labor demand and income level of low-skilled workers (Han et al., 2012; Han et al. 2015). This allocation of factor and income is conducive to narrowing the urban-rural income gap. In this paper, the degree of openness is measured by the ratio of foreign direct investment to gross domestic product (GDP). Data are derived from the CEIC database.

(5) Financial efficiency (Finance). With the improvement of financial efficiency, rural residents will enjoy better financial services. They have more opportunities to earn capital returns, which will help to narrow the urban-rural income gap (Clarke and Zou, 2006; Beck et al., 2010). The financial efficiency is measured by the ratio of loans to deposits in financial institutions. Data are derived from the “China Compilation of Statistics 1949–2008” and the Statistical Yearbook of 31 provinces.

(6) The proportion of state-owned enterprises (SOE). In a perfect market mechanism, the allocation and wage levels of the labor are determined by the market. At the present stage, China’s economy shows a mixed system with various types of ownership, and the allocation and wage levels of the labor are different between the state sector and the non-state sector (Démurger et al. 2006; Deng and Ye, 2012). It is difficult for the labor in the non-state sector to enter the state sector. Therefore, the state sector has monopolistic advantage and pays workers higher wages and social security than that the non-state sector. Thus, a bigger proportion of state-owned enterprises may result in higher social inequality and a greater income gap. In this paper, the proportion of state-owned enterprises is measured by the production ratio of state-owned industrial enterprises to the above-scale industrial enterprises.

(7) Gross domestic production per capita (gdp). Simon Kuznets (1955) holds that there is an inverted U-shaped relationship between economic development and income inequality. In the early stages of economic development, the non-agricultural sector with higher income inequality develops rapidly, and the income inequality in the whole society tends to deteriorate. When the economic development achieves a high level, the proportion of the non-agricultural sector increases significantly and the income gap decreases between sectors. In addition to the effect of income redistribution policies, there will be more equal income distribution. This paper chooses per capita income to measure the economic development level.

3.2 Space Econometrics Model

This paper uses the Spatial Autoregressive Model (SAR) and the Spatial Error Model (SEM) to test the effect on the urban-rural income gape of education at different educational stages. On one hand, the urban-rural income gap of a province depends on that of other provinces. The spatial autoregressive model takes into account the endogenous interaction between the explanatory variables. On the other hand, the spatial error model considers the interaction of spatial error items. The missing explanatory variables that affect the urban-rural income gap are spatially related in the spatial error model. When these two kinds of spatial panel models are used at the same time, they can overcome the influence of different potential spatial correlation factors, and the regression results will be relatively robust.

The function of the spatial autoregressive model takes the form Although the model controls a series of variables that affect the urban-rural income gap, if some important variables may be omitted, the regression results of the model are impacted. Furthermore, the urban-rural income gap may affect the education input in turn, and create endogeneity problems. Therefore, this paper introduces the first order lagged term of the urban- rural income gap as an

explanatory variable. It can not only measure the potential factors such as historical background, human environment, customs, and political influences, but also solve the endogeneity problem as an instrumental variable to a certain extent.

The function of the spatial error term model takes the form

where, is the spatial autocorrelation coefficient of the error term, and it measures the effect of stochastic disturbance of other provinces’ urban-rural income gaps. The other variables are defined as the spatial autocorrelation model.

Because both the spatial autocorrelation model and the spatial error model use the entire spatial system to calculate the spatial correlation, the regression model cannot rule out the existence of endogeneity problems. If the ordinary least squares method is used, the parameters of the spatial error model are not valid, and the parameters of the spatial autocorrelation model are neither unbiased nor consistent (Lesage, Pace, 2010). Therefore, this paper uses the Maximum Likelihood (ML) model

proposed by Elhorst (2003) to estimate the parameters. Beenstock and Felsenstein (2007) argue that if the spatial sample of the regression model is a randomly selected subsample of a population, it is appropriate to choose a random effect panel model. If the regression sample is exactly the population, it is appropriate to choose a fixed effect panel model because each space unit cannot be randomly sampled. The sample in this paper is 31 fixed-space units divided by the provincial level of mainland China, so the fixed effect model is chosen.

3.3 Empirical Analysis

Tables 1 and table 2 report the results of the SAR model and SEM model. At the same time, the model results including the interaction between education input and labor migration are given. The interaction tests whether there is interaction between them, that is to say, can increasing education input promote labor migration and can labor migration strengthen the effect of education input on the urbanrural income gap.

In general, the SAR model and the SEM model are passed through the traditional Lagrange Multiplier Test and Robust Lagrange Multiplier Test. From results of R2 and Log-L, all models have significant goodness of fit, which means the spatial dynamic panel data framework constructed in this paper describe the microscopic mechanism of education input on the urban-rural income gap very well.

The spatial correlation coefficients of the model are greater than zero and pass the 1% significance test, indicating that there is a significant positive spatial correlation relation between the urban-rural income gap of different provinces, and the estimated results of the parameters of each explanatory variable are significant.

First, education input at different educational stages and the urban- rural income gap show a significant negative correlation relationship, indicating that education input at each educational stage can narrow the urban-rural income gap. Second, the coefficient of the spatial dynamic panel model with interactive variables is significantly positive and more than that without interactive variables. Therefore, education input at different educational stages can not only improve labor productivity and directly narrow the urban-rural income gap, but also indirectly narrow the urban-rural income gap through labor transfer, which is consistent with the previous theoretical analysis. Third, education input at the compulsory educational stage, especially at junior middle educational stage, has a more significant effect on narrowing the urban-rural income gap than that at other educational stages. Therefore, it is more effective to narrow the urban-rural income gap by increasing education input at the compulsory educational stage, especially at the junior middle educational stages. Although the total amount of education input at the compulsory educational stage has increased significantly in China, there are still large gaps between regions, especially at the junior middle educational stage. Figure 2 indicates that the

public budget education expenses per student in the eastern developed provinces are five times that of the central and western provinces. In summary, in the future, the allocation of China’s education funds should be tilted to compulsory education, especially to junior high education.

The impact of each control variable on the urban-rural income gap is also consistent with the theoretical analysis in this paper. The coefficient of the openness is negative, indicating that opening to the outside world can narrow the urban-rural income gap, consistent with the Stolpa-Samuelson theorem. As a developing country, China should be actively participating in international trade and introducing foreign capital to encourage industries with comparative advantages to develop rapidly, which will increase labor demand and income levels of low-skilled workers and narrow the income gap. The coefficient of financial efficiency is significantly less than zero, indicating that financial efficiency is negatively correlated with the urban-rural income gap. Developing an inclusive financial system and improving financial efficiency can effectively narrow the urban-rural income gap in China. The impact of the proportion of state-owned enterprises is significantly positive, indicating that a bigger proportion of state-owned enterprises results in a more serious division of labor allocation which will lead to a greater income gap between the non-state sector and the state sector; The estimated coefficient of per capita GDP is also significantly greater than zero, indicating that China’s economic development is still in the rising stage of Kuznets’ inverted “U-shaped” curve. The coefficient of the lagged variable of urban-rural income gap also passes the significance test, indicating that the urban-rural income gap is impacted by the historical background, human environment, political system, and other potential factors.

Table 3 reports the regression results of education input on the per capita disposable income growth rate of urban households and the per capita net income growth rate of rural households. The spatial autoregressive model and the spatial error model both illustrate the same result. First, the regression coefficient of education input is significantly greater than zero, indicating that increasing education input can improve labor productivity and increase the income levels of rural residents and urban residents. Second, the effect of education input increasing labor income is more significant for rural residents than urban residents. Therefore, education input can narrow the urban-rural income gap.

In summary, increasing education input can not only increase the incomes of rural residents but also increase the incomes of urban residents, which reflect the Pareto improvement. Furthermore, education input at each educational stage can not only improve labor productivity and potential economic growth, but also promote labor transfer and narrow the urban-rural income gap, which is conductive to achieving economic and social inclusive growth goals.

3.4 Robustness Test

This paper uses different sample intervals to test the robustness of regression results of the dynamic spatial panel model. Considering that the reform of state-owned enterprises may impact the income distribution, I used the subsample interval 1998–2014 to test the robustness. The results show that coefficients of educational input and most control variables are significant, and there is no obvious change in the values and sign of coefficient. There is still a strong negative correlation relationship between education input and the urban-rural income gap, that is, education input can narrow the urbanrural income gap. The conclusion is robust.

4. Conclusion

From the perspective of labor migration, this paper constructs a general equilibrium model to the urban-rural dual economic structure, studies the microscopic mechanism of education input on narrowing the urban-rural income gap, and tests the hypothesis of the relationship between education input, labor

migration, and the urban-rural income gap based on the dynamic spatial panel model.

The general equilibrium theory shows that education input alleviates the urban-rural income gap both in direct and indirect ways. On one hand, increasing education input can improve the accumulation of human capital and the marginal production of the labor, and directly increase the labor income in both the non-agricultural and agricultural sectors. However, the effect of education input promoting labor productivity is more significant than that in the non-agricultural sector. Thus, increasing education input can narrow the urban-rural income gap. On the other hand, education input improves the accumulation of human capital and the labor productivity of low-skilled workers, which provides more job opportunities in the non- agricultural sector and reduces the cost of labor migration. The improvement of labor allocation efficiency is conductive to narrowing the urban-rural income gap.

The empirical results are consistent with theoretical analysis. First, in creasing education input at each educational stage can effectively narrow the urban-rural income gap in China, and the effect of education input increasing labor income is more significant for rural residents than urban residents, which is a Pareto improvement. Education input at the compulsory educational stage, especially at the junior middle educational stage, has a more significant effect on narrowing the urban-rural income gap than that at the other educational stages. Although the total amount of education input at the compulsory educational stage has increased significantly in China, there are still large gaps between regions, especially at the junior middle educational stage. During the “Twelfth FiveYear Plan” period, at the junior middle educational stage, the public budget education expenses per student in the eastern developed provinces were five times that of the central and western provinces. With the adjustment for the macroeconomic slowdown, and the decreasing growth rate of fiscal revenue, further optional allocation of education funds at different educational stages is an important guarantee for achieving the strategic objectives of “Education Plan” and building a comprehensive well-off society in 2020. Therefore, China should further increase education input, and the allocation of China’s education funds should be tilted to compulsory education, especially to junior high education.

Second, with the gradual decreasing of the “demographic dividend,” the cost of China’s factors is rising, and the industrial structure transformation from a manufacturing to a service industry has resulted in the labor surplus of low- skilled workers in the traditional agricultural sector and a relative labor shortage of skilled workers in the modern non- agricultural sector ( Chao and Shen, 2014). The lagging employment structure transformation has caused great deviation between industrial structure and employment structure. Increasing education input can promote labor productivity and income levels of low- skilled workers in the traditional sector and raise the proportion of the skilled workers in the labor market. The “migration dividend” brought by educational input both can narrow the urban-rural income gap and alleviate the contradiction of the lagging employment structure transformation.

Third, with adjustment for the slowdown of China’s macro-economy, raising fiscal expenditures on education can not only smooth economic growth by compensating for the decrease of private investment in the short term, but also improve potential economic growth by promoting human capital accumulation in the long run. In other words, increasing education input can promote labor migration and narrow the urban-rural income gap, which is an important means to achieving China’s inclusive growth targets.

Figure 1: The Direct and Indirect Effects of Education Input Narrowing Urban-rural Income Gap

Description: The Moran I is the result of the spatial correlation test. R2 and Log-L illustrate the goodness of the model. λ is the Spatial Correlation Coefficient for Spatial Autoregressive Model. ***, **, * were significant at 1%, 5%, and 10% significance levels, respectively. The value in brackets are t values of the estimated parameters.

Figure 2: Per Capita Education Expenditure for Junior Middle Education in 2015 (yuan)