Transcript Does Money Matter? The Effect of Private Educational

Comments on “Does Money Matter? The Effect of
Private Educational Expenditures on Academic
Performance”, by Changhui Kang
Halsey Rogers, DECRG
Public-Private Partnership Conference, June 2007
Overview
1. Value of this paper
2. Econometric and data issues
3. Interpreting the findings


Does private tutoring not matter?
Why do households invest so much in tutoring?
Value of this paper

Private tutoring has received too little study

Major feature of education systems, in East Asia especially



Tutoring is quantitatively very important
Tutoring choices provide insight into the educational objectives of
households
Paper uses an appropriate empirical framework


Good data -- detailed student information linked to college
entrance exam scores
Uses IV to try to get around problem of endogeneity of
tutoring expenditures


Recognizes that parents may invest disproportionately in either highor low-ability children
Tries to control for pre-tutoring performance, by collecting
retrospective information on students’ Grade 11 ranks

Therefore more likely to be picking up tutoring value-added
Econometric and data-related issues





Appropriateness of the instrument
 Does first-child dummy really meet the exclusion restriction?
 If not, is it enough that the direct effect on test scores is likely
positive?
First-born vs. only-child effect
 Run separately (including comparison of means) without the 7%
of sample for whom only_child = 1?
Separate instruments for first-born boys and girls
 Clarify what we learn from this over-identification test?
Endogeneity of some explanatory variables
 Hours_of_self-study is likely endogenous to tutoring decision 
exclude?
Dependent variable: Test scores vs. 12th-grade grades/ranking?
Interpreting the findings 1:
Should we conclude that private tutoring doesn’t matter?

Magnitude of the tutoring effect


Korean parents spend 85% as much on private tutoring as on
public schooling
Total private tutoring expenditures/GDP = ~ 2-3%
Total expenditures
From 1998 KEDI Survey on Educational Expenditures
Educational expenditures in
Expenditures/
GDP (%)
Billion Won
Billion $US *
Public
of which:
government spending
school tuition & fees
30,484
21.7
6.8
19,096
11,188
13.6
8.0
4.3
2.5
Private
of which:
private tutoring
private-school tuitions
29,378
21.0
6.5
14,190
15,188
10.1
10.8
3.1
3.4
* Approximate; uses annual average 1998 exchange rate of 1402 won/dollar
Total expenditures
From 1998 KEDI Survey on Educational Expenditures
Educational expenditures in
Expenditures/
GDP (%)
Billion Won
Billion $US *
Public
of which:
government spending
school tuition & fees
30,484
21.7
6.8
19,096
11,188
13.6
8.0
4.3
2.5
Private
of which:
private tutoring
private-school tuitions
29,378
21.0
6.5
14,190
15,188
10.1
10.8
3.1
3.4
* Approximate; uses annual average 1998 exchange rate of 1402 won/dollar
Private expenditures per student
From 1998 KEDI Survey on Educational Expenditures
Kindergarten
Elementary
Middle school
General high school
University
Million won
$US*
1.55
3.3
2.06
2.33
3.95
$1,106
$2,354
$1,469
$1,662
$2,817
* Approximate; uses annual average 1998 exchange rate of 1402 won/dollar
Private expenditures per student
From 1998 KEDI Survey on Educational Expenditures
Kindergarten
Elementary
Middle school
General high school
University
Million won
$US*
1.55
3.3
2.06
2.33
3.95
$1,106
$2,354
$1,469
$1,662
$2,817
* Approximate; uses annual average 1998 exchange rate of 1402 won/dollar
Interpreting the findings 1:
Should we conclude that private tutoring doesn’t matter?


Magnitude of the tutoring effect
 Korean parents spend 85% as much on private tutoring as on
public schooling
 Total private tutoring expenditures/GDP = ~ 2-3%
What this means:
 If tutoring increases test scores by roughly as much as public
schooling does, and
 If test-score gains actually represent increased human capital,
then:
 Tutoring  ~1/4 of human capital acquired
 Rough calculation, but effects are likely still large  money does
matter, in an aggregate sense
Interpreting the Findings 2:
Why so much investment in tutoring? Some possibilities
Interpreting the Findings 2:
Why so much investment in tutoring? Some possibilities
Tutoring returns in terms of
Human
Test scores
capital
High (relative to
public school returnsl)
High
Reason for tutoring
investment by
households
Private hakwons deliver
education more
efficiently
Case
Private
efficiency
(Jimenez,
Lockheed, &
Paqueo 1991;
Tooley 1999)
Policy implication
Expand scope for private
schooling/choice/
innovation?
Interpreting the Findings 2:
Why so much investment in tutoring? Some possibilities
Tutoring returns in terms of
Human
Test scores
capital
Reason for tutoring
investment by
households
Case
Policy implication
High (relative to
public school returnsl)
High
Private hakwons deliver
education more
efficiently
Private
efficiency
(Jimenez,
Lockheed, &
Paqueo 1991;
Tooley 1999)
Expand scope for private
schooling/choice/
innovation?
High
Low or zero
Parents have different
objectives (e.g., rat race)
than social planner &
public schools do
(human capital)
Signalling
(Spence 1973,
Akerlof 1976)
Discourage private tutoring
(e.g., Korea in 1970s)?
Interpreting the Findings 2:
Why so much investment in tutoring? Some possibilities
Tutoring returns in terms of
Human
Test scores
capital
Reason for tutoring
investment by
households
Case
Policy implication
High (relative to
public school returnsl)
High
Private hakwons deliver
education more
efficiently
Private
efficiency
(Jimenez,
Lockheed, &
Paqueo 1991;
Tooley 1999)
Expand scope for private
schooling/choice/
innovation?
High
Low or zero
Parents have different
objectives (e.g., rat race)
than social planner &
public schools do
(human capital)
Signalling
(Spence 1973,
Akerlof 1976)
Discourage private tutoring
(e.g., Korea in 1970s)?
Moderate to low
High
Parents want children to
acquire more human
capital than public
school day allows
Lengthen public school
day/year?
Kang 2007?
Interpreting the Findings 2:
Why so much investment in tutoring? Some possibilities
Tutoring returns in terms of
Human
Test scores
capital
Reason for tutoring
investment by
households
Case
Policy implication
High (relative to
public school returnsl)
High
Private hakwons deliver
education more
efficiently
Private
efficiency
(Jimenez,
Lockheed, &
Paqueo 1991;
Tooley 1999)
Expand scope for private
schooling/choice/
innovation?
High
Low or zero
Parents have different
objectives (e.g., rat race)
than social planner &
public schools do
(human capital)
Signalling
(Spence 1973,
Akerlof 1976)
Discourage private tutoring
(e.g., Korea in 1970s)?
Moderate to low
High
Parents want children to
acquire more human
capital than public
school day allows
Moderate to low
Low or zero
Extreme version of rat
race (parents will pay for
small absolute gain in
test scores)?
Lengthen public school
day/year?
Kang 2007?
Strongly discourage private
tutoring (by changing
university admissions
process)?
Summary: Kudos, and a few suggestions



Clarify empirical and data issues
 Discuss other possible instruments?
Clarify how money matters, rather than whether it does
Explain what might be happening: why the massive investment in
tutoring?
 Deeper discussion of the context would be helpful to show why
the paper’s results are plausible
 Draw on work by sociologists, anthropologists, popular media
Comments on “Does Money Matter? The Effect of Private
Educational Expenditures on Academic Performance”, by
Changhui Kang
Halsey Rogers, DECRG
Public-Private Partnership Conference, June 2007