This thesis includes three empirical chapters, which are selfcontained
but all related to education inequality in China.
Chapter 2 aims to examine the causal effect (nurture effect) of
parental education on children’s education. Parents and their
children share many common characteristics which are often
unobserved – this causes the omitted-variable bias. To eliminate
this bias, this chapter uses school interruption during the
Chinese Cultural Revolution (CR) as an instrument. The Chinese
Cultural Revolution (1966-1976) significantly interrupted
one generation’s education, but has had no direct effect on the
next generation. It naturally provides a valid instrument. The
empirical results suggest that in urban China, an one year decrease
in parent’s schooling because of school interruption during
the CR leads to a 0.27-0.38 year decrease in the child’s schooling;
if a parent did not obtain a university degree because of
school interruption during the CR, the child is 35-53 percent less
likely to obtain a university degree. The results also suggest that
maternal education has a greater influence on children’s education
than paternal education. Overall, for the particular group
whose parental education was changed by the CR, this chapter
ix
x
confirms a significant and sizable nurture effect.
Chapter 3 estimates the causal effect of private tutoring on
Chinese and mathematics test scores of primary school students
in urban China. Because the unobserved determinants of schooling
achievement often influence private tutoring expenditure,
OLS cannot provide a consistent estimate. This chapter adopts
a heteroskedasticity based identification strategy proposed by
Lewbel (2012) to handle this problem. The estimation results
show that, on average, private tutoring expenditure has a small
but statistically significant effect on the mathematics test scores
of primary school students, but has no statistically significant
effect on the Chinese test scores. A 1000 yuan (about 55% of
a standard deviation) increase in private tutoring expenditure
raises the primary school students’ mathematics test scores by
0.80-0.87 percentage point (about 11% of a standard deviation).
The instrumental variable quantile regression combined with the
Lewbel method suggests that private tutoring is more likely to
improve student achievement at the bottom end of test score distribution.
When moving upward to the top end, the effect becomes
smaller and even negative, despite not being significant.
Chapter 4 measures and analyzes the rural-urban schooling
achievement gap in primary education in China. Using the RUMiC
2010 data, this chapter finds that the test scores of urban children
xi
are on average about 5.7 percentage points higher than that of
rural children. This is approximately equivalent to 54-56% and
65-72% of the standard deviations of urban and rural children’s
test scores, respectively. The regression analysis and OaxacaBlinder
decomposition analysis suggest that urban children outperform
their rural counterparts mainly because of their better
socio-economic background (e.g. parental education and family
income per capita), better school quality and higher government
budgetary spending on education per student in cities.