Upload
educacion-innovacion
View
316
Download
0
Embed Size (px)
Citation preview
Long-Run Effects of Free Choice Among Public Schools:
College Attainment, Earnings and Social Outcomes at
Adulthood
Victor Lavy
University of Warwick, Hebrew University, and NBER
Madrid, February 29, 2016
The 1994-1996 Tel Aviv school choice experiment
The Pre 1994 Busing Program
40 percent of the students in district 9 were bused to Northern Tel Aviv
middle schools (districts 1-4)
Other 60 percent assigned to district 9 middle schools
The 1994 Choice Program
At end of 6th grade, free choice of secondary school (7th to 12th grade)
Each student could choose one of five schools
Including 3 from district 9 and 2 from districts 1-4 in Northern Tel Aviv
Choice set of 5 schools was the same for all students in a primary school
Students list friends to be with in school, at least two accommodated
Schools were not allowed to select students
Admission was centrally done by the City school authority
Schools could not dropout students based on academic achievement
School enrollment flexible, expanding capacity if over-subscribed
Lavy V. “Effect of Free Choice Among Public Schools”
Review of Economic Studies, July 2010
Focused on short and medium term effects:
Reduced dropout rate during 7th to 12th grade by 6.5 percentage points
(a 32 percent decline)
Increased matriculation rate by 8 percentage points (a 25 percent
increase)
Increased Bagrut average score by 7 points (a 12 percent increase)
Improved quality of education:
Number of Bagrut credit units increased by 13 percent
Number of credit units in science subjects increased by 33 percent
Number of Bagrut subjects at honor level increased by 20 percent
The program achieved these gains by allowing closure of schools
and permitting enrollment flexibility in over subscribed schools.
This Study
Main question: whether effects of free school choice persist beyond attainment
and test scores in high school and lead to long term enhancements to human
capital and well-being.
Now, 20 years after choice was made, is an unusual opportunity to evaluate
whether free school choice among public schools in Tel Aviv have a lasting
and long term impact on social and economic outcomes.
This study provides the first evidence of links between school choice and
students’ employment and earnings, and social outcomes at adulthood.
Background
Research in economics of education about the effectiveness of
educational programs and interventions have centered on the
evaluation of impact on short-term outcomes, primarily
standardized test scores, as a measure of success.
However, the ultimate goal of education is to improve lifetime
well-being.
Therefore, attention shifted recently to long term consequences
at adulthood with an initial focus on post-secondary attainment
in light of the increasing economic returns to higher education
(Heckman and LaFontaine 2010; Acemoglu and Autor 2010).
Literature on long term effect of school choice
There is little causal evidence on the long term effect of school choice even though
it is still a very controversial policy.
Deming et al (2014) study the impact of a public school choice lottery in Charlotte-
Mecklenburg schools on college enrollment and degree completion. Findings
suggest significant gains from school choice among girls with no effect on boys.
Girls who attend their first choice school are 14 percentage points more likely to
complete a four-year college degree.
Chingos and Peterson 2013, study an experiment offering a private-school voucher
to low-income families. Overall, no significant effects on college enrollment of the
offer, however, large significant impacts for African-American students and smaller
but statistically insignificant impacts for Hispanic students.
Wondratschek et al 2014. study short and long term effect of Sweden 1992 school
choice reform, find it had very small positive effects on marks at the end of
compulsory schooling, but zero effects on university education, employment,
criminal activity and health at age 25.
Recent ‘visible’ Studies on Long Term Effects of Other
Educational Programs
Garces et al (2002) and Deming (2009) on Head Start
Schweinhart et al (2005) on the Perry Preschool program
Chetty et al (2011) on kindergarten classroom size and peers
on early adulthood earnings
Chetty et al (2014a and 2014b) on earnings consequences of
primary and middle school teachers’ quality
Lavy (2015) on effect of pay for performance for high school
teachers on students post-secondary schooling and earnings at
adulthood.
Summary of findings
School choice increased academic college schooling:
Treated students are 5 percentage points more likely to enroll in academic
colleges
Complete almost an additional quarter year of college schooling
These gains reflect a 15 percent increase relative to pre-program means
and they are similar in magnitude to the gains in high school in
matriculation outcomes
No gain in enrollment in research universities.
Not surprising result since those affected are marginal students
from low SES families who would not enroll in any academic
schooling if not for the school choice program.
Average annual earnings at age 30 increased by 5 percent.
No systematic effect on social outcomes such as marriage and
parenthood.
Data
Administrative data from Israel’s National Insurance Institute (NII), for
students who graduated high school in 1999-2001, in 2011 adults aged 28-30.
NII is responsible for social security and mandatory health insurance.
NII allows restricted access to this data in their protected research lab.
The underlying data sources include:
Population registry data, which contain information on personal status, number of
children and their birth dates (Ministry of the Interior)
NII records of postsecondary enrollment from 2000 through 2011 based on annual
reports submitted to NII every fall term by all of Israel’s post-secondary education
institutions
NII records of all welfare allowance, including unemployment and disability
payments.(NII data)
Israel Tax Authority information on income and earnings of employee and self-
employed individuals (Ministry of Finance)
For this project, NII constructed an extract containing these data and linked it
to data from Minisitry of Education (used in Lavy 2010).
The post high school academic schooling system in Israel
Includes seven research universities
over 50 academic colleges that confer academic undergraduate
degrees (some of these also give master’s degrees),
All universities require a Bagrut certificate for admission with at least
one subject at advance level and English at intermediate or high level.
Most academic colleges also require a Bagrut, though some look at
specific Bagrut components without requiring full certification.
For a given field of study, it is typically more difficult to be admitted
to a university than to an academic college.
Definitions of outcomes at adulthood
Post-Secondary Education
University enrollment indicator = 1 for enrollment at least
one year in any university institution.
Completed years of university schooling defined as the
number of years of enrollment in any university during the
period 2000-2013.
Respective variables for college schooling
Definitions of outcomes at adulthood
Labor Market Outcomes:
Earnings: annual gross earnings from salaried and non-salaried
employment in 2000-2012.
Employment: An indicator with value 1 for individuals with non-zero
number months of work in a given year, 0 otherwise.
Eligible for Unemployment Benefit: An indicator = 1 if received in a
given year unemployment benefits.
Unemployment Benefit Compensation: Annual benefits of
unemployment compensation.
Identification and Estimation
In previous work (Lavy 2010) I used difference in differences
(DID) and regression discontinuity (RD) methodologies to
estimate the effect of the school choice.
Three alternative comparison groups were used in the DID
estimation: first included adjacent Tel Aviv school districts
that were not treated initially, the second included two Easterly
adjacent cities with an independent school authority, the third
included a Southerly adjacent city with an independent school
authority.
All three comparison groups yielded exactly the same
estimates for treatment effects even though one comparison
group had worse characteristics and outcomes and the other
two was better at benchmark pre intervention cohorts.
Identification
I therefore combine in this paper all three groups into one
comparison group to gain estimation efficiency due to a larger
sample size.
In previous work I also used and RD methodology in a sub-
sample that yielded a balanced treatment and control group.
This analysis yielded similar evidence regarding the impact of
the choice program and I therefor use the RD sample to look at
the effect of school choice on LT outcomes as well.
17
Estimation
I first present a controlled comparison of treated and untreated
students using samples of pre and post treatment cohorts based
on the following regression:
(1) Yijt = Xijt + Zj d + Uijt
where Yijt is the ith student's outcome in school j and year t;
Xijt is a vector of the same student’s characteristics;
Zj is the treatment indicator
d is the treatment effect.
Estimation
In addition, I use the before-and-after cross section data as
stacked panel data for primary-school fixed effects, using three
years of cross-section data combined.
The treatment indicator Zjt is now defined as the interaction
between a dummy for the year 1994 and the district 9
indicator, as follows:
(2) Yijt = µj + πt + Xijt + Zjt d+ ijt
where µj is the primary school fixed effect and πt is a year (i.e.,
1992, 1993 and 1994) effect.
Descriptive Statistics
Table 1: Descriptive Statistics and Pre and Post Treatment-Control Comparison of Outcomes Means
(11 Years since high school graduation)
Pre: 92 and 93 cohorts Post: 94 cohort
Treated
schools
mean
Non treated
Schools
mean
Mean
Difference
(Standart
error)
Treated
schools
mean
Non treated
Schools
mean
Mean
Difference
(Standart
error)
A. Enrollment in Post-
Secondary Schooling
University 0.178 0.235 0.057- 0.158 0.231 0.073-
(0.012) (0.016)
Academic College 0.210 0.272 0.062- 0.254 0.282 0.028-
(0.012) (0.017)
B. Post-Secondary Years of
Schooling
University 0.720 1.019 0.299- 0.646 0.975 0.330-
(0.058) (0.077)
Academic College 0.600 0.866 0.265- 0.759 0.865 0.106-
(0.045) (0.062)
Table 1: Descriptive Statistics and Pre and Post Treatment-Control Comparison of Outcomes Means
(11 Years since high school graduation)
Pre: 92 and 93 cohorts Post: 94 cohort
Treated
schools
mean
Non treated
Schools
mean
Mean
Difference
(Standart
error)
Treated
schools
mean
Non treated
Schools
mean
Mean
Difference
(Standart
error)
C. Employment Outcomes
Employed (1 = Yes, 0 = No)0.861 0.837 0.024 0.841 0.844 0.003-
(0.010) (0.014)
Average Annual Earnings (NIS)82,407 86,245 3,838- 77,914 80,501 2,587-
(2,098) (2,780)
Received Unemployment
Insurance Benefits (1 = Yes, 0 =
No) 0.068 0.070 0.001- 0.062 0.070 0.008-
(0.007) (0.010)
Total Unemployment Insurance
Benefits Received (NIS) 854 912 58- 750 818 68-
(110) (139)
Table 1: Descriptive Statistics and Pre and Post Treatment-Control Comparison of Outcomes Means
(11 Years since high school graduation)
Pre: 92 and 93 cohorts Post: 94 cohort
Treated
schools
mean
Non treated
Schools
mean
Mean
Difference
(Standart
error)
Treated
schools
mean
Non treated
Schools
mean
Mean
Difference
(Standart
error)
D. Personal Status
Married (1 = Yes, 0 = No) 0.525 0.505 0.020 0.457 0.393 0.064
(0.014) (0.019)
Children (1 = Yes, 0 = No) 0.447 0.408 0.039 0.372 0.281 0.091
(0.014) (0.018)
Number of children 0.796 0.692 0.104 0.612 0.443 0.169
(0.029) (0.034)
Effect Of School Choice on Post-
Secondary Academic Schooling
-0.05
0
0.05
0.1
0.15
0.2
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Un
ive
rsit
y En
rollm
en
t R
ate
Year
Mean and Treatment Effect: Enrollment Rate in Universities
Treatment Effect Estimate
Mean
*Shaded area indicates two sided confidence intervals, 10% significance level.
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Year
s o
f U
niv
ers
ity
Sch
oo
ling
Year
Mean and Treatment Effect: Years of University Schooling
Treatment Effect Estimate
Mean
*Shaded area indicates two sided confidence intervals, 10% significance level.
-0.05
0
0.05
0.1
0.15
0.2
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011Enro
llme
nt
Rat
e in
Aca
de
mic
Co
llege
s
Year
Mean and Treatment Effect: Enrollment Rate in Academic Colleges)
Treatment Effect Estimate
Mean
*Shaded area indicates two sided confidence intervals, 10% significance level.
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Year
s o
f U
niv
ers
ity
Sch
oo
ling
Year
Mean and Treatment Effect: Years of Academic Colleges Schooling
Treatment Effect Estimate
Mean
*Shaded area indicates two sided confidence intervals, 10% significance level.
Table 4: Differences-in-Differences Estimates of The Effect of School Choice on Academic Post-
Secondary Schooling (After 11 years since high school graduation)
Enrollment Years of Schooling
Mean, 1992-
1993 Cohorts
in Treated
Schools
Treatment
Mean, 1992-
1993 Cohorts
in Treated
Schools
Treatment
A. University Schooling 0.219 0.000 0.924 0.038
(0.414) (0.013) (1.975) (0.064)
B. College Schooling 0.278 0.042 0.848 0.179
(0.448) (0.019) (1.589) (0.051)
Table 4: Differences-in-Differences Estimates of The Effect of School Choice on Academic Post-
Secondary Schooling, RD SAMPLE (After 11 years since high school graduation)
Enrollment Years of Schooling
Mean, 1992-
1993 Cohorts
in Treated
Schools
Treatment
Mean, 1992-
1993 Cohorts
in Treated
Schools
Treatment
A. University Schooling 0.232 0.123 0.980 0.486
(0.423) (0.054) (2.052) (0.278)
B. College Schooling 0.246 0.038- 1.437 0.187
(0.431) (0.050) (0.713) (0.150)
Effect Of School Choice on Earnings at
Adulthood
-10000
0
10000
20000
30000
40000
50000
60000
70000
80000
-4000
-2000
0
2000
4000
6000
8000
10000
12000
14000
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
20
09
New
Isra
eli
She
kels
(N
IS)
20
09
New
Isra
eli
She
kels
(N
IS)
Years Since High-School Graduation
Mean and Treatment Effect: Annual Earnings - 2009 Prices NIS
*Shaded area indicates two sided confidence intervals, 10% significance level.
Table 4: Differences-in-Differences Estimates of The Effect of School Choice on Employment and Income
11 years after high school
graduation
Stacked Regression 9-11 years
after high school graduation
mean, 1992-1993
Cohorts in
Treated Schools
Estimate
1992-1993
Cohorts in
Treated Schools
Estimate
Employment Indicator (1 = Yes, 0 = No) 0.846 0.014-
(0.361) (0.014)
Total Annual Earnings (2009 NIS) 80270.6 4120.9 76641.1 3517.9
(2430.782) (2173.967)
Can The Improved High School Outcomes Account for
the Gains in Post-Secondary Schooling and Earnings?
Table 7: Estimated Treatment Effect of School Choice when Adding High School Educational Outcomes to the DID
Regression
Added Control Variables
Original
Estimate/
No Added
Variables
Average
Matriculati
on Score
Received
High
School
Matriculati
on
Number of
Credit Units
in
Matriculatio
n Exams
Number of
Science
Credit
Units
All High
School
Outcome
Variables
(1) (2) (3) (4) (5) (6)
A. Post-Secondary Educational Outcomes
Full Sample (N = 14,818)
Enrollment in College Post Secondary Schooling
(1 = Yes, 0 = No) 0.042 -0.003 0.006 -0.003 0.026 -0.010
(0.021) (0.017) (0.019) (0.018) (0.020) (0.017)
Completed Years of College Post Secondary
Schooling 0.179 0.008 0.044 0.012 0.106 -0.027
(0.079) (0.064) (0.078) (0.069) (0.071) (0.065)
Table 7: Estimated Treatment Effect of School Choice when Adding High School Educational Outcomes to the DID Regression
Added Control Variables
Original
Estimate/
No Added
Variables
Average
Matriculatio
n Score
Received
High School
Matriculatio
n
Number of
Credit Units
in
Matriculation
Exams
Number of
Science
Credit Units
All High
School
Outcome
Variables
(1) (2) (3) (4) (5) (6)
B. Earnings at Adulthood
Full Sample (N = 14,688)
Total Annual Earnings in 2011 (NIS) 4129.9 1,894 2,950 1,801 3,782 1,314
(2,430.8) (2,398) (2,281) (2,392) (2,252) (2,309)
Regression Discontinuity Sample
(N = 2,314)
Total Annual Earnings in 2011 (NIS) 8,287 5,730 7,392 6,316 7,678 3,220
(3,407) (4,240) (4,189) (4,062) (4,269) (4,108)
Conclusions An important question of interest to society is the impact of educational
interventions on long-run life outcomes. Recent research has begun to look
at this issue, but much work remains to be done, particularly with regard to
the long-term effects of interventions explicitly targeting improvement in
the general quality and students’ educational attainment.
The high school system in Israel and its high-stakes exit exams are very
similar to those in other countries, and the school choice program studied in
this paper have many similar features with related program implemented in
recent years in the US and in European and other OECD countries.
As a result, the lessons learned from this research are transferable and
applicable to other educational settings throughout the developed countries
of the world.
Conclusions
The school choice program studied here had positive longer term outcomes
at adulthood.
The evidence clearly suggest that not only it improved sharply students’
exit high school outcomes six years later, but it also impacted positively
their path to post-secondary schooling, increased meaningfully their
earnings over a decade and a half later.
These results are important because the school choice experiment targeted a
disadvantaged population in some of the more deprived parts of Tel Aviv.