Lesson 8 : Education II

Question 1

What is the Mincerian method of estimating returns to schooling?

Answer 1

• Simple OLS regression model by Mincer (1974)
• ln (w) = a + ß1 S + ß2 X + ß3 X^2 + u

where
x: labor market experience
ß1: average rate of return to schooloing to wage employment

Results: approximately 10% (robust to cross-country analysis by Montenegro & Patrinos)

Question 2

What are some general statistics about return to schooling?

Answer 2

• estimates return to schooling are higher for women than for men
• returns to tertiary education are the highest, to secondary the lowest
• Returns are higher in low/ middle income than in high income economies

Question 3

How do you solve the endogeneity in returns to education estimation?

Answer 3

Problem: access to schooling is non-random

Solution: staggered school construction:
* differences in cohort exposure to new schools
* differences in construction across regions

DiD:
ln (w) = a + ß SA + γ + λ + W θ + ε

where
w: adult hourly wage
SA: dummy for having access to school
W: controls
γ: locality fixed effect
λ: cohort FE

-> you can also split ß into many coefficients, one for each time (age) -> gives treatment effect dependent on age

Question 4

What is the empirical evidence on returns to education in LMICs?

Answer 4

• Primary school investment is productive: school investment raises adult wages
• secondary education: published evidence shows limited labor market impacts
• secondary education -> suggestive evidence in kenya of rise in fromal employment probability (through RDD)
• secondary education -> free scholarshops, no impacts on men, increased civil service jobs for women, RCT in Ghana

Question 5

Name the results of Duflo’s paper on school construction in Indonesia (Duflo, 2001).

Answer 5

• each primary school constructed per 1000 children lead to an average increase of 0.12 to 0.19 years of education
• estimated returns to education ragned from 6.8& to 10.6%

Question 6

Explain the intuition and findings in the paper on tech-aided schools by Navarro-Sola (2024)

Answer 6

Motivation: ICTs as a potential solution for public service access in LMICs
-> post primary education is challenging to scale up for countries with limited state and fiscal capacity
-> use of ICTs (telesecundarias) as a potential solution (lectures through TV, daily classes with remote instructions & in-class work)

**Identificaton: **
* staggered roll-out of schools
* Treatment: density of TV-schools available when 15 yo
* Variation: staggered construction of TS across Mexico over 20 years, Municiplaities switching to high tv-school intensity

Findings:
* increase in educational attainment
* lower secondary graduation (strong increase), years of education (strong increase)

Counterfactuals for TV-schools (what would have happened if TV schools did not get constructed?):
* Full Sample: more people in Tv school, number of students in standard schools drops -> suggestive of switchers
* low-service sector: increase in TV school students, barely impact on regular schools -> either you go to TV school or you work
* high services sector: larger fraction of TV students that would have attended standard schools instead (people have more access to both schools)

• Increased adult income (only partially due to changes in labor market composition)
• 3pp (7.5%) increase in probability of working in services
• 8% increase in hourly wage

Question 7

What are the threats to identification in the paper on TV schools? How can you test them?

Answer 7

Concern: In-migaration and out-migration differentially changing sample composition

Evidence:
1. Impacts on cohort sizes (size of each cohort of non-migrants), effects on gender ratio
-> no tv-schools impacts on cohort population and share of females
2. use census to assess short-term impacts and migration
3. investigate inter-municipal migration using alternative datasets (surveys)

Question 8

What are the labor market impacts of TV-schools?

Answer 8

• implied returns to education: an additional year of education resulting from an extra TV-school per 50 children increase of 10% adult hourly income
• an additional year of education: through tvs: extra 1,141 pesos, through normal schools extra 864 pesos
• estimates benefit-cost ratio is at least 3.4
• returns of tv-schools vary by region (sector), more diversified areas offer higher wages

Question 9

What is the policy relevance to the findings on TV-schools?

Answer 9

• ICTS offer scalable, cost-effective ways to reach remote populations
• teacher shortages persist in rural post-primary schools, existing incentives are insufficient
• long-term evidence on low-tech standardized instructions, showing that hiybrid approaches can be effective
• expanding digital infrastructure could enable even graeter future gains thorugh ICT-enhanced models