The Impact of Government Transfer Payments on Extracurricular Tuition Expenditure for Children in Urban China

This article uses 2012–18 China Family Panel Survey (CFPS) panel data to study the impact of government transfer payment receipt on extracurricular tuition expenditure for children of urban families requiring aid. Findings have shown that government transfer payments increase extracurricular tuition spending for children and have a positive impact on total family expenditure with regard to children's education. Heterogeneity tests indicate that the impact is significant in the central and eastern regions of China, among children above seven years old and children in one-child families. The authors attribute the positive effects to parents' increased awareness of the importance of education and to increased opportunities for children from assisted families to participate in extracurricular tuition classes. This article proposes policy recommendations to improve the government transfer payment system in urban China.

INTRODUCTION

Over the past 40 years of reform and opening up, close to 800 million people in China have been lifted out of absolute poverty, contributing to close to 75 per cent of the world's poverty reduction.¹ However, China's absolute poverty line² is lower than that set by the World Bank.³ If the World Bank's corresponding poverty line were used as the standard, many people would still be estimated to be poor in China, an upper-middle-income country. Poor families in China usually have low levels of education and lack the ability to improve their skills and make a better living. Many families continue the cycle of passing poverty to their children, therefore causing intergenerational transmission. Human capital investment plays a vital role in breaking poverty's intergenerational transmission, particularly investment in children's education. An increase in educational attainment could result in rising wages and improvements in child health.⁴ Therefore, to reduce the rate of poverty returns, China should focus on improving the human capital accumulation of children from poor families.

Government transfer payments are an important measure for solving the poverty problem in China. The existing literature on the effect of government subsidies in China focuses mostly on their impact on poverty, consumption, labour supply and other variables related more to adults.⁵

Children's education is important to poor families. The lack of early childhood education has high potential to lead to failure in later social competition, thereby perpetuating poverty. However, few studies have examined the impact of government subsidies on children's educational investment and human capital accumulation from the perspective of breaking intergenerational poverty.

Simultaneously, extracurricular tuition plays an important role in early childhood education in China. Chu⁶ points out that the main burden on Chinese families at the compulsory education stage involves off-campus education. Would assisted families, restricted by economic conditions, increase their children's educational investment through extracurricular tuition after receiving government transfers? This article investigates the impact of receiving government subsidies on children's extracurricular tuition spending by focusing on educational investment in children. The authors find that receipt of government transfer payments increases children's extracurricular tuition expenditure, and that the positive impact is more obvious for children in the central and eastern regions of China, children at primary and secondary school levels, and children in one-child families. Such positive impact is, however, less obvious for children in the western region of China, children at the preschool level and children in families with more than one child. By mechanism analysis, the authors highlight that assisted families in receipt of government transfer payments have heightened parental concern about education, thus increasing the probability of children attending extracurricular tuition classes. The authors also observe that the higher the government transfer payment, the more families spend on extracurricular tuition for children.

EXTRACURRICULAR TUITION IN CHINA

Extracurricular tuition has a large market size in China. According to statistics from the Chinese Society of Education, in 2016, there were 220,000 training institutions nationwide, grossing 800 billion yuan in annual earnings, and the number of students enrolled in extracurricular tuition reached more than 137 million.⁷ Based on the past trend, the current figures should be substantially higher. A 2018 questionnaire survey conducted among parents of primary and secondary schools in Shanghai shows that 84 per cent of their children were enrolled in extracurricular tuition classes, of which 87 per cent had tuition in mathematics and 69 per cent in English.⁸ As extracurricular tuition became a trend, parents feared that children who did not enrol in tuition classes might be at a risk of being "left behind". Extracurricular tuition gives students an unfair advantage when competing with their peers.

In addition, the distribution of educational resources in China is uneven and the Matthew effect (accumulated advantage) is serious. Li and Zhao⁹ point out that interschool inequalities are large. Generally, students in top high schools are more likely to be admitted to universities and a good university degree usually guarantees a good job that would help them escape from poverty. Some studies have shown that extracurricular tuition can significantly improve children's grades in primary and secondary schools,¹⁰ and increase students' probability of entering top high schools and universities.

Chinese parents attach great importance to their children's education. Relevant data show that in 2011, the average annual education expenditure of families in first-and second-tier cities for their children receiving primary and junior high school education was almost 8,754.4 yuan, which was about one-third of annual household income.¹¹ Some studies suggest that household income level is a key factor affecting education expenditure and that an increase in income level increases the expenditure on extracurricular tuition.¹²

LITERATURE REVIEW

The authors base their research on four strands in the relevant literature. The first strand studies the impact of Chinese government transfers on household human capital investments. Based on Chinese Household Income Project (CHIP) 2012 data, Gao, Zhai and Garfinkel¹³ find that China's urban subsistence allowances increase the human capital investment of urban families, including education and medical care, instead of increasing terminal consumption expenditures. The conclusion drawn has been corroborated in subsequent research using CHIP 2007 urban survey data.¹⁴ A study by Du and Park¹⁵ using labour market survey data from 2001 and 2005 highlights that China's urban subsistence allowances increase the poor's food and education expenditures. Another research project, by Han et al.,¹⁶ using household survey data from five Chinese provinces in 2010, indicates that rural families increased their healthcare expenditures after receiving social assistance, while their educational expenditures did not change evidently. A study by Wang, Yang and Gao,¹⁷ that uses CHIP 2013 urban data, reveals that receipt of welfare benefits encourages poor recipients to increase their spending on both education and healthcare. However, the aforementioned literature focuses on the educational expenditure for families and adults and does not consider educational expenditure with children as the research subject. Therefore, there remains a lack of research on the impact of government transfer payments on investments in children's education. In addition, these studies do not distinguish between the different forms of investment in education. Filling the gap in the existing literature, this article focuses on extracurricular tuition, an important investment for children's education, to investigate whether government transfer payments increase extracurricular tuition for children from recipient families.

The second strand of literature suggests that investing in children's education is an important means of breaking poverty's intergenerational transmission. Investment in human capital accumulation is essential for eliminating poverty. In the long run, human capital accumulation is an endogenous driving force blocking poverty's intergenerational transmission.¹⁸ Among the numerous types of human capital investment, educational investment is an important form. Compared with human capital factors, such as work experience and vocational education, some scholars notice that children's education is the core human capital factor affecting poverty.¹⁹ However, the aforementioned literature does not delve into the impact of government transfer payments on investment in children's education.

The third strand of literature focuses on the impact of household income on investment in education. While family income is generally taken to be an important factor affecting household education investment, there is no unified view of the correlation. Some scholars believe that household investment in education increases with income level.²⁰ Other scholars qualify this view by pointing out that it is only when the total household income level reaches a certain critical point that household income has an impact on education expenditure.²¹ Among household education expenditures, extracurricular tuition is characterised by high fees and short cycles, and it is deemed a "high-risk" investment behaviour for families.²² Extracurricular tuition is a heavy burden, particularly for low-income families.²³ Previous studies have done some work on how income affects household spending on education, but the existing literature has paid scant attention to the impact of transfer payment income from external sources on education expenditure, especially on extracurricular tuition, a form of education investment with a large burden. This article investigates whether assisted families receiving government transfers increase their children's extracurricular tuition spending and serves to complement existing literature.

The fourth literature strand focuses on the influence of household factors and characteristics on investment in children's education. Scholars have found that in addition to family income, family investment in children's education is affected by the background and circumstances of the family, such as the number of children in a household²⁴ and their education stage.²⁵ However, literature on how households allocate transferred income among different children's educational investments is also scarce. This article explores whether the impact of government payment transfers on the education investment of assisted families varies based on their children's circumstances.

In sum, this article complements existing research in the following ways. First, it focuses on investment in children's education and studies the impact of government transfer payments on children's extracurricular tuition spending; past literature, in contrast, focuses mainly on the family or adult level. Second, this article will enrich the research on investment in children's education, specifically featuring extracurricular tuition expenditure in China. Third, this study constructs the proportion of government transfer payments receipt at the county level as an instrumental variable, using the two-stage least squares method, and conducts robustness tests by replacing the dependent variable and including fixed effects at the family level. The proposed methodology would improve the validity of empirical results. Fourth, this article examines the heterogeneity of government transfer payments on children's extracurricular tuition expenditure through subsample regression by age group and family structure, specifically whether government transfers have different effects on children's education expenditures according to the children's backgrounds in the assisted households. This focus enriches the research on families' investment preference for children's education.

EMPIRICAL STUDIES

Data Sources

This article uses panel data from the China Family Panel Surveys (CFPS) in 2012, 2014, 2016 and 2018 to examine the impact of receiving government transfer payments on children's extracurricular tuition expenditure. CFPS is a large-scale follow-up survey and research project implemented by the Institute of Social Science Survey (ISSS) of Peking University. It covers three levels of data: individual, family and community. It includes four databases: Children, Adults, Families and Communities. These four CFPS databases aid greatly in the study of China's social, economic, cultural, educational, medical and health issues. This survey has been formally implemented nationwide since 2010; since then, a follow-up survey of the original samples has been carried out biennially. The overall sample size is 16,000 families from 25 provincial administrative regions across China, which represents well the country's overall situation.

Description of Variables

The panel data used in this article are based on the CFPS children database, and the father's age and education level data in the CFPS adult database, family-level income and population size in the CFPS family database; county-level fiscal revenue and population size data in the county-level statistical yearbooks are matched with the CFPS children database (Table 1).

The dependent variable used here is the logarithm of the extracurricular tuition expenditure on an individual child, and the independent variable is associated with whether the family receives government transfer payments. Based on the relevant literature, the control variables used in this study include the child's age, gender, father's age, education level, family-level per capita annual income, number of children in a household, county-level per capita fiscal revenue and year dummy variable.

Table 1.

Description of Variables

Table 2 shows that about 53.2 per cent of urban families received government transfer payments. The average logarithm of the individual child's extracurricular tuition expenditure is 1.913.²⁶

The average age of children in the sample is 8.069 years, with boys accounting for about 53.7 per cent. The average age of fathers is 36.771 years. The average educational level of the fathers is 3.016, which is slightly higher than "middle school". Each family has an average of 1.771 children. In addition, the average family-level per capita annual income in the sample is 11,963.320 yuan. The descriptive statistics for each year are shown in Appendices I to IV.

Table 2.

Descriptive Statistics

Benchmark Regression

In the CFPS 2012–18 panel data, there are a total of 33,982 children samples, of which 12,809 children samples are from urban households, accounting for 37.69 per cent. Since the counties included in the CFPS database and those in the National Statistical Yearbook do not exactly match, when matching urban children samples with the per capita fiscal revenue variable in the National Statistical Yearbook, 6,786 urban children samples can be successfully matched with the "county-level per capita fiscal revenue" variable. When these 6,786 urban children are matched with the father's age and father's education in the CFPS adult database, 4,732 observations are successfully matched with the father's age and father's education in the CFPS adult database. The number of valid data observations for the dependent variable, extracurricular tuition expenditure, in the 4,732 urban children samples is 3,365. The number of valid data observations of the core independent variables and all control variables in the above 3,365 urban children samples is 2,995. Therefore, the final number of observations entering the regression is 2,995, accounting for 23.38 per cent of the total number of urban children samples. This process eliminates mainly observations that cannot be matched to obtain county-level per capita fiscal revenue and the father's age and education, as well as observations with missing and outlier values for the explained and explanatory variables. Therefore, the authors' process in working with the sample does not lead to biased regression results and the regression results are valid.

The authors first perform the Hausman test to determine whether to use fixed-effects or random-effects models. The p-value of the Hausman test is 0.0000, rejecting the null hypothesis that the omitted variable is unrelated to the explanatory variable. Therefore, this study first uses a fixed-effects model to study whether a family's receipt or non-receipt of government transfer payments has an impact on the extracurricular tuition expenditure on the child. The specific regression model is formulated as follows:

where γ_it, the dependent variable, represents the child's extracurricular tuition expenditure; μ_it represents the intercept; payments_it represents the independent variable; Control_it represents the control variables group; υ_i represents the unobserved heterogeneity; ε_it represents the random error changing over time; and β₁ and β₂ represent the influence of the explanatory variables on the dependent variable. Table 3 presents the regression results.

Table 3.

Impact of Receipt or Non-receipt of Government Transfer Payments on Extracurricular Tuition Expenditure

Column (1) of Table 3 presents the full-sample regression using the fixed-effects model with the initial control variables; after controlling for each control variable, family receipt government transfer payment has a positive impact on the extracurricular tuition expenditure on the child, which is significant at the 10 per cent significance level. Column (2) of Table 3 presents subsample regressions using the fixed-effects model with initial control variables, which exclude the observations of annual family-level per capita income at the highest 10 per cent of the county. Column (3) of Table 3 presents subsample regressions using the fixed-effects model with logarithm of family-level per capita annual income and logarithm of county-level per capita fiscal revenue, which also exclude the observations of annual family-level per capita income at the highest 10 per cent of the county. The results in columns (2) and (3) of Table 3 demonstrate that the positive effect of government transfer payment on the extracurricular tuition expenditure on the child still holds and the effect is significant at the 10 per cent significance level.

By using the aforementioned fixed-effects model, the authors can rule out the influence of unobservable factors that do not change temporally. Nevertheless, the estimation model may still be plagued by endogenous problems of missing variables and reverse causality. The first is to omit the time-varying unobservable variables. For example, changes in family decisions may affect whether families are recipients or non-recipients of government transfer payments and also their extracurricular tuition expenditure for their children. The second problem is reverse causality. The expenditure on family extracurricular tuition is often affected by the family income level, which is a key factor in determining whether a family should obtain a government transfer payment. If these problems cannot be overcome, errors will occur in the estimation results. The authors use the instrumental variable method to solve the endogeneity problem. The instrumental variable used is the proportion of government transfers received by other families in the same county. This variable is the sum of government transfer payments received by other households in the same county except for the household under study divided by the number of households in the entire county minus 1. The instrumental variable is borrowed from Fisman and Svensson²⁷ and Lei and Lin.²⁸ Fisman and Svensson use the mean value of bribery and taxation of other companies in the same region and industry as an instrumental variable when studying the impact of corporate bribery and taxation on corporate growth. Lei and Lin use the date of county-level New Cooperative Medical Scheme (NCMS) inception as an instrumental variable for an individual's NCMS health insurance status when studying the causal effect of the NCMS on health and utilisation of preventive care. Lei and Lin find that the county-level insurance rate of NCMS is not directly related to the model's disturbance term.

The size of the poor population varies in different regions, and the probability of households receiving government transfer payments is also different. Households in the same county have common characteristics. The probability of receiving government transfer payments for the studied households is greater in counties with a large proportion of poor people. Therefore, the probability of the studied households receiving government transfer payments is similar to other households in the same county, so this instrumental variable is correlated with the endogenous core explanatory variable. The average proportion of receiving government transfer payment (excluding the household under study) in the same county is not related to the studied household's time-varying unobservable factors (such as changes in family decisions), so it will not affect extracurricular tuition expenditure for children of the studied household in other means, thus satisfying the exogeneity of instrumental variables. Thus this is a valid instrumental variable for the following two-stage least squares regression.

Suppose Y_ijt represents the extracurricular tuition expenditure on the i-th child in county j at period t; DL_jt represents the average proportion of receiving government transfer payments (excluding the household under study) in county j at period t; X_ijt represents the control variables of the i-th child in county j at period t; and represents the outcome predicted by the instrumental variable and other control variables in county j at period t. The instrumental variable regression results are shown in Tables 4 and 5. The first-stage results of the instrumental variable regression are shown in Table 4 and the second-stage results in Table 5.

Table 4.

First-stage Results of Instrumental Variable Regression

Table 5.

Second-stage Results of Instrumental Variable Regression

Column (1) of Tables 4 and 5 shows the full-sample instrumental variable regression using the fixed-effects model; the results of column (1) of Table 5 also show that after controlling for each initial control variable, receipt of a government transfer payment has a positive impact on the child's extracurricular tuition expenditure and is significant at the 10 per cent significance level. Column (2) of Tables 4 and 5 shows the subsample instrumental variable regression using the fixed-effects model with initial control variables, which exclude the observations of annual family-level per capita income in the top 10 per cent of the county. The results in column (2) of Table 5 show that the positive effect of government transfer payments on the extracurricular tuition expenditure on the child still holds and is significant at the five per cent significance level. After using the logarithm of family-level per capita annual income and the logarithm of county-level per capita fiscal revenue as control variables, the results in columns (3) and (4) of Table 5 show that government transfer payments have a positive effect on children's extracurricular tuition expenditure at the 10 per cent and five per cent significance levels, respectively. The coefficients of the core independent variables in columns (1) to (4) of Table 4 are significant at the one per cent level. This shows that the higher the proportion of other county-level households (excluding the studied household) receiving government transfer payments, the higher the probability of the studied household receiving government transfer payments. Therefore, the correlation requirement of the instrumental variable is met. The values of the Cragg-Donald Wald F-statistic in columns (1) to (4) of Table 5 are far more than the 10 per cent critical value (16.38) of the weak instrumental variable identification test, showing that the instrumental variable used is an effective instrumental variable.

HETEROGENEITY EFFECTS

Regression by Region

Regional differences in China's development are obvious. The impact of receiving government transfer payments on the child's extracurricular tuition expenditure may vary across different regions. Therefore, the authors study the samples in three separate parts: the eastern, central and western regions.²⁹ The first-stage results of the regional subgroup regression are shown in Appendix V and the second-stage results in Table 6. For the central and eastern regions, receipt of government transfer payments has a positive impact on the child's extracurricular tuition expenditure, which are both significant at the five per cent significance level. In the western region, receipt of government transfer payments has no significant impact on the child's extracurricular tuition expenditure. The values of the Cragg-Donald Wald F-statistic in columns (1) to (3) of Table 6 are all greater than the 10 per cent critical value (16.38) of the weak instrumental variable identification test.

This phenomenon may be attributed to two causes. First, family income has a significant positive effect on extracurricular tuition expenditure.³⁰ As mentioned above, some scholars deduce that household income affects household education expenditure only when total household income reaches a certain critical point.³¹ Per capita income in the western region is lower than that in the eastern and central regions, and parents with lower income in the western region may pay less attention to their children's education and spend money on other aspects instead of extracurricular tuition. In the western region, from 2012 to 2018, average annual urban household income was 13,081 yuan and average expenditure on extracurricular tuition was 776 yuan, which accounted for about 5.93 per cent of average urban household income. By contrast, in the central and eastern regions, also over the same period, average annual urban household income was 17,763 yuan, and average expenditure on extracurricular tuition was 1,587 yuan, which accounted for about 8.93 per cent of average urban household income. The data show that urban households in central and eastern regions with higher income will indeed spend a higher proportion of total income on children's extracurricular tuition than do the western urban households. Second, extracurricular tuition institutions are concentrated mostly in economically developed areas and large cities, and most leading extracurricular tuition companies have less than 15 per cent of their teaching centres located in cities of the third tier and below.³² Therefore, since there are fewer professional extracurricular tuition providers in the western region, children in the western region are less likely to receive professional extracurricular tuition.

Table 6.

Subsample Regression by Region

Regression by Age and Number of Children

The number of children and age of children affect parents' investment in their children's education.³³ To investigate any variation in the impact of government transfer payments on children's extracurricular tuition expenditure by age and number of children, the authors first separate the children sample used in the benchmark regression into two groups, 0–6 years old and seven years old and above, to conduct grouping regression.³⁴ The second-stage results by age are shown in columns (1) and (2) of Table 7. Next, the authors separate the children sample used in the benchmark regression into an additional two groups to carry out grouping regression: only child and non-only child.³⁵ The second-stage results by number of children are shown in columns (3) and (4) of Table 7. The first-stage results of the four regressions are presented in Appendix VI.

The results in columns (1) and (2) of Table 7 show that receipt of government transfer payments has a positive impact on extracurricular tuition expenditure for urban children aged seven and above, which is statistically significant at the 10 per cent significance level. However, receiving government transfer payments has no significant impact on extracurricular tuition expenditure for urban children in 0–6 age group.

This demonstrates that the positive impact of government transfer payments on children's extracurricular tuition expenditure is manifested mainly in children in urban primary and secondary schools, while the impact on children in preschool education in urban areas is not obvious. This may be due to the different needs of Chinese children at different age stages for education investment and extracurricular tuition classes. In general, as children grow older and move up to primary and secondary schools from preschools, they face increasing pressure from academic competition and to gain admission into better schools. Therefore, children's parents are also increasing their demands for extracurricular tuition classes. A study by Zhao³⁶ corroborates this finding—when income increases by the same proportion, the increase in children's education expenditure in the compulsory education stage is greater than that in the preschool education stage. In the sample data, from 2012 to 2018, for households with children aged seven years old and above, their average annual urban household income was 16,324 yuan, and their average extracurricular tuition expenditure was 1,992 yuan, which accounted for about 12.20 per cent of average annual urban household income. For households with children in the 0–6 age group, their average annual urban household income was 17,320 yuan, and their average extracurricular tuition expenditure was 628 yuan, which accounted for about 3.62 per cent of average annual urban household income. The data highlight that urban households with children aged seven years old and above indeed spend a higher proportion of total household income on children's extracurricular tuition than those with children in the 0–6 age group. This affirms the author's hypothesis that as children grow older, the more intense is the academic competition and the greater the demand for extracurricular tuition classes.

The results in columns (3) and (4) of Table 7 show that receipt of government transfer payments has a positive impact on the extracurricular tuition expenditure in urban one-child families, which is significant at the 10 per cent significance level. However, receipt of government transfer payments has no significant impact on extracurricular tuition expenditure in urban families with more than one child. This reflects that the positive impact of government transfer payments on children's extracurricular tuition expenditure is manifested mainly in urban one-child families, while the impact on urban multiple-child families is not obvious. This can be explained by the higher attention that Chinese parents of one-child families have paid to their child's education and their willingness to invest more educational resources for their only child. Several studies support this argument. For example, Becker and Lewis³⁷ highlight that the greater the number of children, the fewer the educational resources each child is allocated. Lee³⁸ also finds that children in one-child families receive more opportunities for education than those in multiple-child families. Children's education tends to receive greater parental attention in one-child families.

Table 7.

Subsample Regression by Age and Number of Children

ROBUSTNESS CHECK

Impact on Children's Total Education Expenditure

To test the robustness of the benchmark regression results, the authors replace the dependent variable with the logarithm of children's total education expenditure in the past year. Table 8 presents the results. All regressions use two-way fixed effects models. The coefficients of core independent variables in columns (1) to (3) are all positive at the 10 per cent significance level. These results show that government transfer payments have a significant positive impact on children's total education expenditure. This further corroborates the research findings in the benchmark regression that obtaining government transfer payments can improve a family's investment in children's education.

Adding Family-level Fixed Effects

In the benchmark regression, the authors control for individual- and time-level fixed effects. CFPS panel data are micro data based on the individual and household levels. Here, the authors add family-level fixed effects to the regression. Table 9 presents the

Table 8.

Impact of Receipt and Non-receipt of Government Transfer Payments on Total Education Expenditure

results. The independent variable in columns (1) to (2) is associated with family receipt or non-receipt of government transfer payments, and the independent variable of columns (3) to (4) is associated with the amount of government transfer payments received by the household. The results of four regressions show that receiving government transfer payments increases children's extracurricular tuition expenditure and that the larger the amount of government transfer payments, the more families spend on children's extracurricular tuition expenditure. The coefficients of the core independent variables in columns (1) to (4) are all positive at the 10% significance level. This further validates the robustness of the benchmark regression results.

MECHANISM TEST AND EXPANSION ANALYSIS

Degree of Concern for Children's Education

Parental concern about children's education is also an important factor in educational investment. Chu, Furukawa and Zhu³⁹ argue that parental preferences in regard to

Table 9.

Inclusion of Family-level Fixed Effects

their children's education partly affects parents' investment in children's education. In this section, the authors further examine the impact of the mechanism of receiving government transfer payments on the child's extracurricular tuition expenditure. The authors use the degree of concern about children's education as the dependent variable. The variable is scored by the investigators based on the children's education-related items owned by the respondents' families and the value is an integer between one and five. The larger the value, the more concerned parents are about their children's education. Table 10 presents the results. Columns (1) to (3) use a panel-ordered logit model. The coefficients of the core independent variables in these three regressions are all significantly positive at the 10 per cent significance level. This shows that receipt of a government transfer payment increases the degree of parents' concern for their children's education. This explains why receiving government transfer payments increases the extracurricular tuition expenditure on the child.

Table 10.

Impact of Family Receipt or Non-receipt of Government Transfer Payments on the Degree of Concern for Children's Education

Impact on Decision to Participate in Extracurricular Tuition Classes

In this section, the dependent variable is replaced by one which represents whether the child is enrolled in an extracurricular tuition class.⁴⁰ The authors use a panel logit model to test the relationship between receiving government transfer payments and enrolment of children in extracurricular tuition classes. Table 11 presents the results. The coefficients of the core independent variables in columns (1) to (3) are all positive at the 10 per cent significance level. These results show that receiving government transfer payments has a significant positive impact on the probability of children enrolling in extracurricular tuition classes. Based on the results in columns (1) to (3), after receiving government transfer payments, the odds ratio⁴¹ becomes 1.487–1.764 times the value that represents non-recipient families of government transfer payments. The higher the probability that families allow children to attend extracurricular tuition classes, the higher the expenditure on children's extracurricular tuition. This can therefore explain the research findings that receipt of government transfer payments in the benchmark regression increases a family's expenditure on children's extracurricular tuition.

Table 11.

Impact of Receipt or Non-receipt of Government Transfer Payments on Enrolment in Extracurricular Tuition

Impact of the Amount of Government Transfer Payments on Extracurricular Tuition Spending

In the earlier sections, the authors examine the impact of receiving government transfer payments on the logarithm of extracurricular tuition expenditure for the child. Here, the authors replace the independent variable with the amount of the government transfer payment to investigate how transfer payment amount affects children's extracurricular tuition expenditure. All regressions use individual-time two-way fixed effects models. Table 12 presents the results. The coefficients of the core independent variables in all three columns are significantly positive at the five per cent significance level. This indicates that receiving greater government transfer payment increases children's extracurricular tuition expenditure. According to the results from columns (1) to (3) in Table 12, an increase of 1,000 yuan in government transfer payments will increase children's extracurricular tuition expenditure by about 8.1 to 8.8 per cent.

Table 12.

Impact of the Amount of Government Transfer Payments on Extracurricular Tuition Expenditure

CONCLUSION

Using 2012–18 CFPS panel data, this study finds that receiving government transfer payments increases children's extracurricular tuition expenditure. In addition, this effect is obvious in the central and eastern regions, but not in the western region; in terms of age and number of children, receiving government transfer payments has a significant positive impact on the extracurricular tuition spending for children in the 7–16 age group and for children in one-child households. These results are robust when the authors use instrumental variable regression, control for the family fixed and year fixed effects, and change the dependent variable to total family expenditure on children's education.

Further research has also revealed that receiving government transfer payments has increased parents' concern for their children's educational attainment. Receiving government transfer payments allows children in assisted families more opportunities to enrol in extracurricular tuition classes. Moreover, the larger the amount of government transfer payments families received, the higher is the children's extracurricular tuition expenditure.

Overall, improving the accumulation of human capital of children from poor families is an important measure to break poverty's intergenerational transmission. Findings have also demonstrated that receiving government transfer payments encourages families to increase education investment for children, thus also helping to improve children's human capital and to break poverty's intergenerational transmission in urban China. Based on the empirical results, the authors propose two recommendations: first, to further improve and expand the coverage of government transfer payments in urban China; second, to improve the targeting efficiency of government transfer payments and increase coverage to families in the central and eastern regions of China, one-child families, and families with children already in primary and secondary schools.⁴²