International Evidence on Educational Vouchers

advertisement
Choice in International
Contexts
ERIC BETTINGER
STANFORD UNIVERSITY
SCHOOL OF EDUCATION
10 MAY 2010
School Choice and Vouchers Abroad
 School Choice Takes Many Forms Abroad




Educational Vouchers
Contract Schools
Open Enrollment
Privately Financed Initiatives
 Vouchers are More Integral to Many Educational
Systems Abroad Than to the US


National Voucher Systems in Chile, Sweden
Targeted Voucher Programs
Means-Tested: Puerto Rico, Colombia
 Women: Bangladesh

 International Literature is still in its infancy in many
ways
International Evidence and the US
 Design of School Systems
 Limits to Generalizeability
 Political Circumstances
 Institutions
 Funding Schemes
 Role of Private Sector
 Motivation for Voucher Systems
 Market Forces
 Equality of Opportunity
 Overcrowding
 Decentralization
Why Look Internationally?
 Most Voucher Literature is From the US
 International programs are much larger than in the
US

Chile, Bangladesh, Colombia
 Long-run Equilibrium?
 Chile’s voucher program is now almost 30 years old
 Existing funding systems and institutions may make
vouchers more feasible
 Many previous reviews: Neal (2002), West (1996),
Zimmer and Bettinger (2008), Wößmann (2007)
Plan of Talk
Discuss Major Research Questions in Int’l Voucher
Literature
2. Major Voucher Programs
1.


Chile
Colombia
3. Effects of Vouchers on Voucher Users
4. Effects of Vouchers on Overall System
Key Research Questions
How do Vouchers Affect Students Who Use Them?
1.





Academic Outcomes
Non-Academic Outcomes
Heterogeneous Outcomes
Cost Effectiveness
Mechanisms?
2. How do Vouchers Affect the System as a Whole?

School Entry and Exit

Overall Attainment and Outcomes

School Staffing

Competition and Resource Loss
Why Study Voucher Users?
 Policy relevance
 Motivation for many policies
 Part of knowing overall impact of system
 Depends on size of system
 Sheds light on other educational questions
 Effect of private schools
 Peer groups
 School Organization
 Cost Effectiveness
Overall Impact of the Voucher
 Efficiency of Overall System
 Friedman (1955)
 Tiebout Competition (1956)
 Alternative Explanations
 Resource Loss
 Limits on Competition
 Sorting
International Voucher Programs














Bangladesh
Belize
Canada
Chile
Colombia
Guatemala
Japan
Lesotho
The Netherlands
New Zealand
Poland
Puerto Rico
Sweden
United Kingdom
Chilean Voucher Program
 Part of Larger Decentralization Effort by Pinochet
Government in 1980
 All students were eligible

Starting in 2008, “bonus” subsidy for admitting low-income
student
 Increase in Private Schooling
 15% in 1981 to 42 percent in 2005
 Public, Voucher, and Private Non-Voucher
 “Topping Off” Allowed in 1993
 Selective Admission Policies in Voucher Schools
Attendance Across Sectors in Chile
Source: Bravo, Mukhopadhyay, and Todd (2009)
Colombia Voucher Program
 From 1992 to 1997, Decentralization Effort by Colombian
Government

Goal to Improve Secondary School Enrollment
 Means-Tested



Focus on Students Entering Secondary School (Grade 6)
Students Coming From Public Sector
Covered About 144,000 Students
 Admissions


Students had to be admitted at a voucher school prior to voucher
application
Vouchers assigned by lottery if oversubscribed
 Voucher Value

Declined over time from 100 to 50 percent
Enrollment Rates in Colombia, 1993
13
Overall
Rate
Rate for
Poorest
Quintile
Rate for
Richest
Quintile
Grade 1-5 enrollment/
Population aged 6-11
89
78
97
Grade 6-11 enrollment/
Population aged 12-18
75
55
89
University enrollment/
Population aged 19-24
30
12
51
Source: Sanchez and Mendes, 1995
Effect on Users
 Not clear what to expect
 Imagine that the voucher at least does no harm
Parents can always reverse decision
 However, “original” school may be altered by voucher program

 Raw correlations
 Colombia: voucher schools have similar outcomes to public
schools (King et al 1997)
 Chile: comparisons depend on types of covariates included
(McEwan and Carnoy 2000)
 Cleanest evidence to date is from Colombia
 Uses randomization to identify the voucher
Evidence from Colombia
15
 Angrist, Bettinger, Bloom, King, Kremer (AER 2002)
 Surveyed Voucher Applicants from Bogotá 1995 Lottery
 Compared Voucher Lottery Winners and Losers
 Effects after Three Years
 Angrist, Bettinger, and Kremer (AER 2006)
 Administrative Data on All Applicants
 Compared College Entrance Exam Scores
Effects on Users: Colombia
16
 Effects after 3 Years






Increased Usage of Private Schools
Higher Educational Attainment
No Difference in Drop-out Rates
Less Grade Repetition
Higher Test Scores
Less Incidence of Teen-age Employment
 Long-Run Effects




30 percent of Voucher Applicants Take College Entrance Exam
Increase of 7 percentage points for Voucher Winners
25 Percent Relative Effect
Impact on Test Scores
Users: Evidence from Chile
 Hard to Interpret
 Counterfactual hard to identify
 New private schools and rapid entry into the market
 Were “control” students affected
 Conclusion depends on Covariates
Some Evidence of Positive Effects on Users (e.g. Sappelli
and Vial 2002)
 Evidence is Largely Mixed as to Whether Vouchers
Improve Outcomes (McEwan and Carnoy 2000)

Evidence on Users: Mechanisms
 Randomization facilitates identification of “intent to
treat” parameter in Colombia

Does not help identify specific mechanisms
 Possible mechanisms

Private schools are better


Peers are better


But school quality was the same (King et al 1997)
Bettinger, Kremer, and Saavedra (2008) examine vocational school
applicants
 Winners “stuck” in vocational schools – peers are worse
 Winners outcomes are better despite worse peers
Voucher included incentives
Students lost voucher if they failed.
 Alternative was labor force participation

Effect on Overall System
 Competing Hypotheses
 Competition
 Resource Loss
Loss of money from enrollment depends on average versus
marginal cost
 Marginal cost of leavers is likely lower than average cost


Inability to close public schools in Chile
 Raw Correlation
 Positive relationship between degree of private competition
and outcomes (Wößmann 2009)
 Lack of pre-program data for Chile
Evidence on System: Chile
 Hsieh and Urquiola (2006)

Difference-in-Differences approach


Average test scores do not rise


Compare areas with significant increase in private school to those
without
Vouchers resort peers. Best students move to private voucher schools.
Some outcomes (e.g. grade repetition) are worse
 Gallego (2005, 2007)

Instrumental Variable approach


Uses stock of Catholic priests in 1980 as instrument
Immediate expansion of parochial schools led overall outcomes of all
schools to improve
Evidence on System: Chile
 Bravo, Mukhopadyay, and Todd (2009)
 Life-cycle model of earnings and schooling decisions
 Use labor force data for individuals educated before and after
the voucher reform
 Estimate impact of choosing private school
 Simulate what educational attainment would have been in
absence of the program.

Positive impacts on educational attainment, high school
graduation, college attendance and graduation, and wages
Evidence on Systemic Effects: Other Countries
 Sweden
 Comparisons based on penetration of private sector


Sandstrom and Bergstrom (2005): Positive effects on overall
system
Value-added comparisons

Ahlin (2003): Positive effects on overall system
 Israel
 Open enrollment => Positive Effects Lavy (2006)
 UK
 Relationship between degree of choice and outcomes
(Gibbons, Machin, and Silva 2008)

Small, positive effects.
Evidence on Systemic Effects: India Experiment
 Kremer and Muralidharan experiment
 Sample of communities who “might be eligible for voucher
study”


Matched randomization on which communities received
voucher


Comparisons between non-applicants across voucher and nonvoucher cities
Within communities with vouchers, lotteries determine
voucher assignment


Gathered applications
Differences across users
Preliminary results positive on both accounts
Future Directions
 Other Outcomes
 Parents choose schools for non-academic reasons


How do these outcomes vary?
Wage impacts

Bravo et al (2009) is first study to explore
 Cost Effectiveness
 Few studies so far
 Public expenditure on vouchers generally smaller
 Mechanisms
 Staffing issues
Summary
 International Voucher Literature in Infancy

Significant contributions in last few years
 Effects on Users



Positive Effects from Colombia
Not clear on the mechanisms and the importance of the voucher for
the effects
Chilean evidence is hard to interpret

Growing consensus of positive effects
 Effects on System



Positive Effects in Chile
Comparisons of areas with and without voucher penetration
More evidence arriving soon
Download