2005年-世界发展银行全球_School_Meals_Educational_Achievement_and_School_Competition___Evidence_from_a_Randomized_Evaluation_52页_423kb

Summary of "School Meals, Educational Achievement and School Competition: Evidence from a Randomized Evaluation"

Core Content

This paper investigates the impact of subsidized school meals on school participation, educational achievement, and school finance in a developing country context, using data from a randomized evaluation in Western Kenya. The study was conducted by a Dutch NGO, Internationale Christelijk Steunfonds (ICS), between 2000 and 2002, involving 25 treatment preschools selected at random from a pool of 50 schools.

Main Findings

• School Participation:

  • Children in treatment schools participated in school 35.9% of the time, compared to 27.4% in comparison schools.
  • The program increased participation on both the intensive margin (children already enrolled) and the extensive margin (children who would not have attended school otherwise).

• Educational Achievement:

  • The program improved curricular test scores, but only for children in schools with more experienced teachers.
  • The improvement was approximately 0.4 standard deviations per standard deviation of teacher experience.
  • There was no impact on cognitive abilities.
  • Boys' weight improved, but there was no effect on girls' height or weight or boys' height.
  • This suggests that the mechanism through which school meals improve learning is not primarily nutrition.

• Classroom Dynamics:

  • The program caused crowding in classrooms, leading to larger pupil-teacher ratios.
  • Teaching time was reduced by about 15% in treatment schools, even though meals were prepared by a cook to minimize disruption.
  • Teacher absenteeism remained high at 30% in both treatment and comparison schools.

• School Fees and Competition:

  • Treatment schools increased their fees by approximately 10% of the food cost, while comparison schools near treatment schools decreased their fees.
  • These price changes are partly due to competitive distortions in the preschool market, as some children transferred from comparison to treatment schools.
  • If school meals were provided universally, the price effects would likely be reversed, with treatment schools having lower prices and comparison schools higher.
  • The intention-to-treat estimator includes these price effects, making it a lower bound on the true effect of school meals.

• Implications for Evaluation:

  • The study highlights the importance of considering competition effects in randomized evaluations.
  • It complements literature on general equilibrium effects and contamination of comparison groups.
  • The trade-off between statistical power and contamination is a key consideration in the design of such evaluations.

Key Information

• Program Design:

  • The school meals program provided a fully subsidized breakfast in treatment schools.
  • The breakfast consisted of porridge with 422 calories per serving, made from a protein-rich flour mixture, sugar, corn oil, and water.
  • The program was not blind, and both schools and NGO staff were aware of the treatment status.

• School Characteristics:

  • Preschools in Kenya are mostly non-governmental, with low funding and minimal infrastructure.
  • Most preschools are informal non-profit centers located on the compound of government primary schools.
  • Teachers generally have low formal education and limited professional training, though they are better educated than the average woman.

• Data and Methodology:

  • The study uses intention-to-treat analysis, which accounts for school transfers and avoids upward bias.
  • The program had no effect on the quality of teaching or school participation in earlier studies.
  • Baseline surveys and follow-up assessments were conducted to measure the program's impact.

• Limitations and Considerations:

  • The results may not generalize to other contexts due to the specific characteristics of the sample schools.
  • Future data collection is needed to assess the impact of spillover effects from nearby schools.
  • The effect of school meals on attendance may be overestimated due to the price effects and market competition observed in the study.

Conclusion

The study provides empirical evidence that subsidized school meals can increase school participation, particularly in schools with experienced teachers. However, the program also led to larger class sizes, reduced teaching time, and price adjustments in the local preschool market. These findings suggest that the intention-to-treat estimator is a lower bound on the true effect of school meals when considering the broader market dynamics. The research underscores the importance of accounting for externalities and competition in evaluating the effectiveness of educational interventions.