2011年-世界发展银行全球_Average_and_Marginal_Returns_to_Upper_Secondary_Schooling_in_Indonesia_40页_1mb

Summary of "Average and Marginal Returns to Upper Secondary Schooling in Indonesia"

Core Content

This working paper investigates the average and marginal returns to upper secondary schooling in Indonesia using a non-parametric selection model estimated by local instrumental variables (LIV). The data comes from the Indonesia Family Life Survey (IFLS), and the study focuses on male workers aged 25-60 with complete wage and schooling information.

The paper introduces the concept of marginal treatment effect (MTE) as a central parameter to understand heterogeneity in returns to schooling. It emphasizes that the return to education is not uniform across individuals and that policy analysis should focus on the marginal student rather than the average student.

Main Points

• Returns to schooling vary widely across individuals:

  • The return for the marginal student (who is indifferent between attending or not) is 14.2% per year.
  • The return for the average student who attends upper secondary schooling is 26.9% per year.
  • The return can be as high as 50% per year for those very likely to enroll or as low as -10% per year for those very unlikely to do so.

• Distance to the nearest secondary school is used as an instrumental variable (IV) to estimate the returns to schooling. It is a strong determinant of enrollment and is self-reported by the community head.

• The study controls for several individual and community characteristics, including:

  • Parental education (father and mother)
  • Whether the individual lived in a village or town at age 12
  • Religion
  • Rural vs. urban residence
  • Province dummies
  • Distance to the nearest health post

• The assumption of exogeneity of the distance variable is supported by evidence showing no correlation with:

  • Whether the individual repeated a grade in elementary school
  • Test scores in elementary school (math, science, Bahasa, social studies)
  • Whether the individual worked while in primary school

• The paper proposes a methodological innovation to estimate the marginal treatment effect (MTE) and related parameters using simulation rather than non-parametric estimation of high-dimensional conditional densities.

• The MTE is defined as the expected difference in wages between individuals with and without upper secondary schooling, given observable and unobservable characteristics. It is used to construct other important parameters such as:

  • Average Treatment Effect (ATE)
  • Average Treatment on the Treated (ATT)
  • Average Treatment on the Untreated (ATU)
  • Policy Relevant Treatment Effect (PRTE)

• The paper also introduces three different definitions of marginal individuals, based on the proximity of the decision threshold:

  1. $|P - V| < \varepsilon$
  2. $|Z\gamma - U_s| < \varepsilon$
  3. $|\frac{P}{U} - 1| < \varepsilon$

• It estimates the average marginal return to schooling using the second definition of marginal individuals, which corresponds to the marginal treatment effect (MTE).

• The study compares its findings with previous estimates in the literature, such as those from Duflo (2000) and Petterson (2010), and finds that the returns to schooling in Indonesia are higher than those reported in earlier studies, although the dataset, instrument, and time period differ.

Key Information

• The sample size is 2608 working-age males.
• The dependent variable is the log of hourly wages.
• The independent variable is schooling, collapsed into two categories:
  • Completed lower secondary or below
  • Attendance of upper secondary or higher
• The methodology includes:
  • Local Instrumental Variables (LIV)
  • Semi-parametric selection model
  • Simulation of conditional densities to avoid high-dimensional non-parametric estimation
• The results suggest that the returns to education are heterogeneous and that policy interventions should be tailored to different groups of individuals.

Conclusion

The paper concludes that the returns to schooling are not uniform and that policy analysis should focus on the marginal student. It also highlights the importance of using appropriate instrumental variables and methodological innovations to better understand the heterogeneity in returns to education. The findings contribute to the understanding of schooling choices and their impact on labor market outcomes in Indonesia.