Final Report Summary - EDUCATION-LONG-RUN (Long-Run Effects of Education Interventions: Evidence from Randomized Trials)
The main objective of this proposal was to analyze the long-term effects of a wide scope of educational interventions and to determine which interventions are more effective at improving long-term outcomes. In each study I exploited creative and convincing natural experiments, using unique administrative data. I provide below summary of each of the research outputs according to two parts of the proposal:
I. Interventions that affected educational inputs and school climate/environment
1. Increasing school resources and instructional time
Six main findings. (1) more school resources and a longer school week have positive and significant effects on students' academic achievements. (2) any additional hours in school that are not devoted to instruction have no effect on students’ cognitive outcomes. (3) no systematic effect of lengthening the school week on pupils’ misbehavior and violence. This is the first study to provide a comprehensive analysis of the causal effect of the school budget and its components on pupil cognitive and non-cognitive outcomes. (4) an additional hour of classroom instruction in Math, science and English have similar positive effects on students’ exam scores. (5) cross subject effects are very small and not significantly different from zero. (6) homework time in each of the core subjects significantly increases with the subject’s increased instructional time in school.
2. Remedial education through individualized instruction
Main finding: remedial education to high school students increased significantly college education and earnings at adulthood, mainly of students from below-median-income families.
3. Effect of School Quality on Students from Deprived Background
In a previous paper, we found that good primary schools improved Ethiopian immigrant children education attainment by end of high school. Following these children until age 33, we find that improved primary school quality did not cause any improvement in post-secondary schooling nor in earnings. I document three explanations: (1) Ethiopians in Israel are discriminated against in the labor market, reducing their incentive to invest in schooling. (2) Tradition, culture and norms in the Ethiopian community in Israel demand young adults to contribute to family income through employment. (3) Lack of information about the financial benefits of investing in higher education.
4. The Long-Run Economic Consequences of High-Stake Examinations: Evidence from Transitory Variation in Pollution
In this study, we evaluate whether cognitive performance during high-stakes exams can indeed be affected by transitory random disturbances, and assess their long-term consequences. Exploiting variation across the same student taking multiple exams, we find that transitory PM2.5 exposure is associated with a significant decline in student performance and also have long-term consequences, lowering post-secondary attainment and earnings at adulthood.
5. Effect of class size on students' academic achievements
We use Maimonides Rule as an instrument for class size in large Israeli samples from 2002-2011. In contrast with Angrist and Lavy (1999), newer estimates show no evidence of class size effects. The new data also reveal enrollment manipulation near Maimonides cutoffs. A modified rule that uses birthdays to impute enrollment circumvents manipulation while still generating precisely estimated zeros. The disappearance of Israeli class size effects may reflect changes in the Israeli education production function.
6. Changing the social network of students in school
We investigate the influence of social networks on educational attainment of children in school. We show that the effect of social network depends both on the 'quality' of friendship ties, and the ‘quality’ of friends. We find that reciprocal friends have positive effect on a student while non-reciprocity in friendship has negative effects on a student. However, students benefit from being popular even if these friendships are non-reciprocal.
7. Gender-based discrimination by teachers and its short and life-long effects on students
In two papers that are part of this ERC proposal I examine how teachers’ gender role attitudes and stereotypes influence the gender gap by affecting the school environment. In the first study (joined with Edith Sand), we estimate the effect of primary school teachers’ gender biases on boys’ and girls’ academic achievements during middle and high school and on the choice of advanced level courses in math and sciences during high school in Tel-Aviv, Israel. Our results suggest that assignment to a teacher with a greater bias in favor of girls (boys) has positive effects on girls’ (boys’) achievements. Such gender biases have also positive impact on girls’ (boys’) enrollment in advanced-level math courses in high school. In a related second paper, I explore the extent to which teachers’ gender bias in high school influences students’ school attendance and academic performance in high-stakes university admission exams and students’ choice of university field of study in Greece. We (jointly with Rigissa Megalokonomou) find a very high persistency in teachers’ gender favoritism behavior over several years. We estimate large effects of these teacher biases on students’ school attendance and performance in university admission exams, quality of enrolled degree and the given field of study at the university.
8. Religion-Based Discrimination in Education
We document the existence of significant in-group bias in grading based on the religious status of students. We find that the bias is almost entirely driven by male examiners. Third, using bunching in the grading distribution, we find evidence that bias, at least at the top of the distribution, is largely driven by religious examiners.
II. Interventions that introduced 'market' type incentives in schools
1. Incentivizing teachers
I examine the effect of teachers’ pay for performance on long-term outcomes and find that it led to an increase in university education and in earnings at adulthood of the treated high-school students. These gains are largely mediated by the positive effect of the program on several high-school outcomes, including quantitative and qualitative gains in the high-stakes matriculation exams.
2. Long-term effects of free school choice in public education
I study free school choice offered to primary school students and find that it led to significant gains in post-secondary education and earnings at adulthood. The effects are heterogeneous across gender. I find that choice enabled students to attend better high school and achieve a better match between students and schools.
3. Incentivizing students
I followed individuals from age 18 to age 33 and found that financial incentives offered in high school had no gains on post-secondary schooling, employment and earnings even though they led to improvements in matriculation exams test scores at age 18.
4. Importance of 'CEOs' in public education as a mechanism to enhance education quality
In this project, I focus on measuring the causal effect of CEOs in public education on the education outcomes of students and schools and identifying the mechanisms behind these effects. I find that higher quality CEO’s have positive impact on students’ learning and behavioral outcomes. This effect is highly non-linear, driven mainly by the most effective CEOs. As mechanisms, our focus is on school management practices.
5. Increasing the return to schooling: effect on children’s education
Using an unusual natural experiment that resulted in a substantial increase in the financial returns to schooling, we find that it led to significant improvements in students’ high school outcomes and later also in their post-secondary schooling. The spillover over effect to peers in schools led also to improvement in their high-school outcomes. I a related study we also found that adults schooling investments increased significantly, mainly in STEM subjects.
I. Interventions that affected educational inputs and school climate/environment
1. Increasing school resources and instructional time
Six main findings. (1) more school resources and a longer school week have positive and significant effects on students' academic achievements. (2) any additional hours in school that are not devoted to instruction have no effect on students’ cognitive outcomes. (3) no systematic effect of lengthening the school week on pupils’ misbehavior and violence. This is the first study to provide a comprehensive analysis of the causal effect of the school budget and its components on pupil cognitive and non-cognitive outcomes. (4) an additional hour of classroom instruction in Math, science and English have similar positive effects on students’ exam scores. (5) cross subject effects are very small and not significantly different from zero. (6) homework time in each of the core subjects significantly increases with the subject’s increased instructional time in school.
2. Remedial education through individualized instruction
Main finding: remedial education to high school students increased significantly college education and earnings at adulthood, mainly of students from below-median-income families.
3. Effect of School Quality on Students from Deprived Background
In a previous paper, we found that good primary schools improved Ethiopian immigrant children education attainment by end of high school. Following these children until age 33, we find that improved primary school quality did not cause any improvement in post-secondary schooling nor in earnings. I document three explanations: (1) Ethiopians in Israel are discriminated against in the labor market, reducing their incentive to invest in schooling. (2) Tradition, culture and norms in the Ethiopian community in Israel demand young adults to contribute to family income through employment. (3) Lack of information about the financial benefits of investing in higher education.
4. The Long-Run Economic Consequences of High-Stake Examinations: Evidence from Transitory Variation in Pollution
In this study, we evaluate whether cognitive performance during high-stakes exams can indeed be affected by transitory random disturbances, and assess their long-term consequences. Exploiting variation across the same student taking multiple exams, we find that transitory PM2.5 exposure is associated with a significant decline in student performance and also have long-term consequences, lowering post-secondary attainment and earnings at adulthood.
5. Effect of class size on students' academic achievements
We use Maimonides Rule as an instrument for class size in large Israeli samples from 2002-2011. In contrast with Angrist and Lavy (1999), newer estimates show no evidence of class size effects. The new data also reveal enrollment manipulation near Maimonides cutoffs. A modified rule that uses birthdays to impute enrollment circumvents manipulation while still generating precisely estimated zeros. The disappearance of Israeli class size effects may reflect changes in the Israeli education production function.
6. Changing the social network of students in school
We investigate the influence of social networks on educational attainment of children in school. We show that the effect of social network depends both on the 'quality' of friendship ties, and the ‘quality’ of friends. We find that reciprocal friends have positive effect on a student while non-reciprocity in friendship has negative effects on a student. However, students benefit from being popular even if these friendships are non-reciprocal.
7. Gender-based discrimination by teachers and its short and life-long effects on students
In two papers that are part of this ERC proposal I examine how teachers’ gender role attitudes and stereotypes influence the gender gap by affecting the school environment. In the first study (joined with Edith Sand), we estimate the effect of primary school teachers’ gender biases on boys’ and girls’ academic achievements during middle and high school and on the choice of advanced level courses in math and sciences during high school in Tel-Aviv, Israel. Our results suggest that assignment to a teacher with a greater bias in favor of girls (boys) has positive effects on girls’ (boys’) achievements. Such gender biases have also positive impact on girls’ (boys’) enrollment in advanced-level math courses in high school. In a related second paper, I explore the extent to which teachers’ gender bias in high school influences students’ school attendance and academic performance in high-stakes university admission exams and students’ choice of university field of study in Greece. We (jointly with Rigissa Megalokonomou) find a very high persistency in teachers’ gender favoritism behavior over several years. We estimate large effects of these teacher biases on students’ school attendance and performance in university admission exams, quality of enrolled degree and the given field of study at the university.
8. Religion-Based Discrimination in Education
We document the existence of significant in-group bias in grading based on the religious status of students. We find that the bias is almost entirely driven by male examiners. Third, using bunching in the grading distribution, we find evidence that bias, at least at the top of the distribution, is largely driven by religious examiners.
II. Interventions that introduced 'market' type incentives in schools
1. Incentivizing teachers
I examine the effect of teachers’ pay for performance on long-term outcomes and find that it led to an increase in university education and in earnings at adulthood of the treated high-school students. These gains are largely mediated by the positive effect of the program on several high-school outcomes, including quantitative and qualitative gains in the high-stakes matriculation exams.
2. Long-term effects of free school choice in public education
I study free school choice offered to primary school students and find that it led to significant gains in post-secondary education and earnings at adulthood. The effects are heterogeneous across gender. I find that choice enabled students to attend better high school and achieve a better match between students and schools.
3. Incentivizing students
I followed individuals from age 18 to age 33 and found that financial incentives offered in high school had no gains on post-secondary schooling, employment and earnings even though they led to improvements in matriculation exams test scores at age 18.
4. Importance of 'CEOs' in public education as a mechanism to enhance education quality
In this project, I focus on measuring the causal effect of CEOs in public education on the education outcomes of students and schools and identifying the mechanisms behind these effects. I find that higher quality CEO’s have positive impact on students’ learning and behavioral outcomes. This effect is highly non-linear, driven mainly by the most effective CEOs. As mechanisms, our focus is on school management practices.
5. Increasing the return to schooling: effect on children’s education
Using an unusual natural experiment that resulted in a substantial increase in the financial returns to schooling, we find that it led to significant improvements in students’ high school outcomes and later also in their post-secondary schooling. The spillover over effect to peers in schools led also to improvement in their high-school outcomes. I a related study we also found that adults schooling investments increased significantly, mainly in STEM subjects.