Table 3: The impacts of private tutoring in selected recent studies

a. Studies that do not control for the endogeneity of private tutoring

Country Year Level/Grade/ Age Summary of Main Impacts Comments Sources
Japan 1980-1982 High-school seniors - For students in the first year out of One variable Stevenson
high school, among forms of private used in all & Baker
tutoring, practice examination and regression (1992)
correspondence course improve the models is
probability of entering college by 16% student
and 25%. academic
- For students in the second year out of standing, which
high school, attending special tutoring is likely to be
school increases the probability of endogenous
entering college by 80%. Having a because it is
private tutor significantly reduces this correlated with
probability, but this most likely reflects unobserved
the remedial character of this form of student innate
private tutoring in Japan. ability or
motivation.

Kenya 1995 Youth ages 13-18 Private tutoring reduces the chance of This study does Buchmann
repeating grades and increases student not control for (2002)
academic performance school
characteristics.
Republic of 2000-2001 Middle and high- Pre-class tutoring (private tutoring that From the Lee et al.
Korea school students teaches a school’s curriculum at least one description of (2004)
month ahead of its schedule) has no the sampling
short-term or long-term effects on procedure (p.
student academic performance 28), this study
does not
appear to
account for
student
motivation into
taking private
tutoring or not.
Singapore Not Eighth-grade students Time spent with a private tutor has a Cheo &
reported negative impact on student academic Quah
performance (2005)
Vietnam 2002 8-year-old children Although private tutoring does not This study does Ha &
significantly increase the children’s not control for Harpham
writing and multiplication test scores, it school (2005)
does double these children’s reading test characteristics.
scores

34
b. Studies that control for the endogeneity of private tutoring
Country Year Level/Grade/ Summary of Main Impacts Comments/ Instruments for Sources
Age private tutoring
Germany 1998-1999 Students in Receiving private tutoring causes a This study uses a form of Mischo &
grades five to larger improvement in academic matching, in which students in Haag
eleven performance and motivational tutoring identify a match, in an (2002)
variables. attempt to account for
unobserved differences between
tutoring participants and non-
participants.
India 2001-2004 Third- and - A remedial education program This study uses a randomized Banerjee et
fourth-grade increases average test scores for all experiment method. al. (2007)
students children in treated schools by 0.28
standard deviations. A computer-
assisted learning program increases
math scores by 0.47 standard
deviations.
- One year after the program, the
initial gains reduced to 0.1 standard
deviations.
Indonesia 2002-2003 Fourth grade Private tutoring has no impacts on A variable indicating school Suryadarma
students mathematics or dictation scores clustering in terms of private et al. (2006)
tutoring is used as the
instrument for private tutoring
Israel 1999-2001 Underperformi - A remedial education program This study uses both difference- Lavy &
ng high-school increases the mean matriculation rate in-difference and IV methods. Schlosser
students for schools and participating students Instrumental variables for the (2005)
by 3-4% and 11-12% respectively. proportion of students
- The program expenditure per participating in the program
participant is about 40% of the annual include the interaction terms of
expenditure per high-school student in school size, year dummy variable
Israel. and treatment status.
Japan 1995 Male university Years spent after high-school This study uses an IV method. Ono (2007)
and high- graduation cramming in private However, the study does not
school tutoring centers (ronin) improve the control for other household or
graduates quality of the college students go to, school or community
thus raising earning indirectly (through characteristics (such parental
this improvement in college quality). education or household
income).
US 1990-1992 High-school Coaching (commercial private tutoring - This study does not report the Briggs
students courses) increases SAT math scores by variables used as instruments for (2001)
14-15 points, SAT verbal scores by 6-8 private tutoring.
points, ACT math and English scores - This study does not control for
by 0-0.6 points, but decrease ACT school characteristics.
reading scores by 0.6-0.7 points.
1997-1999 Third- and Summer remedial programs increase This study uses a regression Jacob &
sixth-grade math and reading achievement for discontinuity method. Lefgren
students third-graders by about 12% of the (2004)
average annual learning gains. The
corresponding figures are around 6%
for sixth-graders.
Vietnam 1997-1998 Primary and Private tutoring has positive effects on This study uses the private Dang
lower student academic performance. tutoring fees charged by schools (2007b)
secondary as instruments for students
students taking private tutoring.

35
 Table 4: Government Policies Toward Private Tutoring In Selected Countries/ Territories
Type General Policy Typical Measures Country Notes
Cambodia (7), The bans were implemented at
Myanmar (1) various times in these countries,
Prohibit private
1 Total ban on private tutoring but they were ineffective because
tutoring
Mauritius (67), of government’s inability to
Republic of Korea (65) enforce them.
Croatia (58),
Most of these countries have weak
Georgia (45),
institutions and do not have the
Nigeria (6),
capacity to regulate private
Mongolia (37),
tutoring
Sri Lanka (49)
2 Ignore These countries have strong
institutions and have the capacity
Canada (94), to regulate private tutoring.
United Kingdom (94) However, they consider the private
tutoring market outside of their
sphere of responsibility.
- Generally prohibit private
tutoring in early grades; prohibit
teachers from tutoring their own
students.
- Regulations accompanied by
inspections and sanctions on Hong Kong (93),
private tutoring fees, class sizes, Lithuania (60),
Recognize and syllabi. Mauritius (67),
3
regulate - Regulations on infrastructure of Republic of Korea (65),
private tutoring centers. Ukraine (28),
- Reduce stratification in the Vietnam (29)
education system, reduce
disparities in schools, raise public
awareness about negative effects of
private tutoring.

- Encouragement policies toward

private tutoring.
Singapore (98), These countries believe that
- Subsidies for private tutoring,
Actively South Africa (71), private tutoring contributes to
4 dissemination of information to
encourage Zanzibar (Tanzania) human capital development and
link potential tutors and clients,
(43) caters to the needs of students.
training courses for tutors, taxation
incentives.
Note: The typical measures are given for illustration purposes, and the countries listed in each category may not
implement all these measures. The number in parentheses after each country is its percentile ranking in control of
corruption, with higher rankings representing higher control of corruption. This Table is constructed mainly from Bray
(2003, 2006), Silova and Bray (2006b). The cases for Croatia are added from Dedic et al. (2006), Georgia from Matiashvili
and Kutateladze (2006), Mongolia from Dong et al. (2006), Lithuania from Budiene and Zabulionis (2006), Ukraine from
Hrynevych et al. (2006), and Vietnam from Dang (2007a) respectively. The corruption rankings are taken from
Kaufmann et al. (2007).

36
 Figure 1: Demand and supply of education with private tutoring

S0 Public education
Price
Private education Public education with
private tutoring
S1
S2
Gains to parents &
C students
B
P*2 E

A D2

Gains to private
tutors

D1

Q0 Q1 Q2 Q*2 Education

1 Both authors are with the Development Research Group, World Bank; email: hdang@worldbank.org (Hai-Anh Dang)

and hrogers@worldbank.org (Halsey Rogers). The authors are grateful to Emmanuel Jimenez, Mark Bray, and three
anonymous referees for their constructive comments, and to the Hewlett Foundation for its partial support of this
research. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do
not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its
affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.
2Our focus is on private tutoring for academic subjects; Lipscomb (2007) and Barron et al. (2000), among others,
examine how non-academic extracurricular involvement affects academic achievement.
3 Strictly speaking, while it may not be correct for these public-financed tutoring programs to fit under the denomination
“private tutoring”, results from these studies are highly relevant to the discussion about equity and efficiency with private
tutoring. Furthermore, these tutoring programs have clear policy implications, since they represent what tutoring may
look like if it is to be used on a large scale by the government.
4 A similar and extreme case happens in Taiwan, where private tutoring is also popular. One Taiwanese student was even

recorded as saying “The very start of my day is at the end of school when I go to the crammer [sic]school; the
mainstream school where I spend most of my day is merely my leisure time” (Wu, 2004).
5 It is also reported that the number of private tutoring colleges listed in the Telephone Book’s Yellow Pages in Sydney—
the largest city in Australia—increased from 60 in 1989 to 222 in 2002 (Kenny and Faunce, 2004). More evidence on the

37