When schools are the ones that choose: the effect of screening in Chile Dante Contreras Sebastián Bustos Paulina Se

1. When schools are the ones that choose: the effect of screening in Chile Dante Contreras Sebastián Bustos Paulina Sepúlveda June 2007

2. Contribution of the paper • This study provides evidence on the use of student selection mechanisms applied by private subsidized schools in a competitive context  This study examines the effects of this practice on the results’ gap between private and public schools and its impact on academic performance. • Two main results: • We present evidence indicating that student selection is a widespread practice among private subsidized schools. • After controlling for a series of selection criteria  there are no differences in results between public and voucher schools.

3. Motivation • Chile  voucher scheme • nationwide system • Since early 80s • Assumption: consumers (families) choose schools for their children. • Parental choice  quality [increasing quality] • Increase opportunity set for families  free choice context.

4. The Chilean Education Market Design… • Public schools are obligated to accept all students. • Private subsidized schools can select students according with their educational objectives. • Most of the PS schools are for profits. • Same voucher. • So what is going to happen under such conditions • PS schools have incentives to select students  the less expensive to educate.

5. Given the market design…

6. Epple y Romano, AER, 1998 • Show that schools will choose the highest ability students from the highest income families. • This is because less capable students imply higher educational costs • They require greater resources than higher ability students to achieve the same results. • Teachers must also spend more time with students with learning difficulties, thus negatively affecting the other students. • If the objective of voucher for profit schools is to maximize profits, then student selection is an easy and economical method for attaining those goals and improving academic results. • Direct Effects • Indirect effects  peers [ Zimmer and Toma, 2000, McEwan 2001, 2003, Sacerdote 2001]

7. Evidence in Chile • This is not the first paper to compare private and public school achievement: • Aggregate school level data (e.g., Bravo et al. 1999, Mizala and Romaguera 2000) • Student level data that attempts to control for selection bias (e.g. Anand et al., 2006: Contreras 2000, Sapelli and Vial, 2002; McEwan, 2001). • After controlling for selection bias and peer effects, the evidence [for Chile] indicates that the gap between public and private subsidized schools is positive and small.

8. Background information required by schools

9. Selection categoriesown elaboration based on parental reports Note: A dummy variable is defined for each of these selection criteria, which takes the value of 1 if the school applies selection criteria and 0 if otherwise. we consider that a school used a given type of selection if more than 50% of the responses were affirmative for each selection category

10. Schools’ student screening

12. In summary… The selection strategy used by Chilean voucher schools is consistent with the theory. [Epple and Romano] Private subsidized schools commonly use selection mechanisms  the most recurrent: Student ability Family income Why?  reduce educational costs + achieve better results. • The evidence indicates  Active selection made by voucher schools.

13. Selection in the other way around… • Parents “want” to be selected? • Parent prefers schools which performs this kind of selection? • If so  school may be reacting to parental preferences • What we do know about parental choice? • Parental choice not correlated with educational achievement. [Elacqua et., al.,JPAM] • Main reasons behind school choice by families is the proximity to the home or workplace. • Parents do not have the necessary information to compare the quality of schools. • Be careful with the “causality” interpretation, but at least the parental choice according to quality of education is not operating.

14. Effects of selection on educational achievement • Direct Effect: screening [Hsie + Urquiola, JPUBE, 2006] • Indirect Effect  better peers through the selection process. • The indirect effect correspond to the benefit of attending a school where the socio-economic profile of students is higher than of schools without selection.

15. Data and methodology • Individual data: SIMCE 2005 [fourth grade] • Parents answer how their children were selected into the school (S = 1 if more than 50% respond affirmatively). • Production function approach. • OLS and IV estimates [similar models, data and empirical strategies  new control variables] • IV strategy • School choice = F (parental education, family income, other hh characteristics, school availability density - previously used in literature)

16. The model …

17. Peers differences Quantifying the Effects: Direct and Indirect Effects of Student Selection

18. Impact – Math test points Statistical significant coef./ inconditional stand. deviation

19. Direct effect Indirect effect

20. The results doesn’t change when we use IV to control the parents school selection bias

22. Conclusions • The evidence indicates that the different selection methods are widely used by voucher schools, and especially in schools with high socio-economic profiles. • Student ability selection is the most frequently used, and produces the greatest effects on subsequent academic results. • Selection of students is an important issue  student discrimination using public resources. • The results by OLS and IV indicate that, after controlling for family and school characteristics and student selection criteria, the public-private gap shown in earlier studies drops to zero after controlling for the selection criteria used. • This evidence also helps to understand the segregation in the Chilean educational system.