STUDENT SKILLS & OUTCOMES
LEAPS research uncovers how learning, skills development, family background, and external shocks shape the educational and life trajectories of children. Our studies explore how early academic experiences, socio-economic status, and motivation influence long-term progress, while also examining the differential effects by gender and other demographics. The research highlights recurring patterns of inequality, the fragile nature of learning gains, and the significant influence of both household context and unexpected disruptions on students’ futures.
Grade Repetition and Household Responses
Authors: Tahir Andrabi, Ethan Matlin, Gabrielle Vasey
Citation: Andrabi, Tahir, Ethan Matlin, and Gabrielle Vasey. 2025. “Grade Repetition and Household Responses in a Low Income Setting”, Working Paper.
Questions and Findings
What are the academic, psychological, and household effects of grade repetition for primary school students, and do these effects differ between government and private schools?
Grade Repetition Lowers Academic Performance: Students who repeated a grade scored 0.27 to 0.44 standard deviations lower in math, English, and Urdu compared to their promoted peers.
Increased Dropout Risk for Repeaters: Grade retention raised the likelihood of dropping out by 7.1 percentage points relative to similar students who advanced.
Parents Reduce Expectations and Investments: After a child is held back, parents revise downward their educational expectations, beliefs about the child’s potential, and their investments in the child’s education.
Students Become Discouraged: Repeating a grade lowers students’ beliefs in the value of their own study effort and decreases motivation.
Why This Matters
Grade repetition, while intended as a remedial support, instead imposes a sharp academic penalty, students lose nearly half a standard deviation in test performance and face a markedly increased risk of dropping out. These quantifiable setbacks are not explained by differences in health, background, or teacher judgment, pointing directly to the practice of repetition as the critical driver.
This evidence signals to education policymakers that current promotion policies may substantially worsen the trajectories of at-risk children and increase educational inequality, calling for urgent reevaluation and alternatives to automatic grade retention.
Skills, Migration, and Labor Market Outcomes
Authors: Alice Danon, Jishnu Das, Andreas de Barros, Deon Filmer
Citation: Danon, Alice, Jishnu Das, Andreas de Barros, and Deon Filmer. 2024. “Cognitive and Socio-Emotional Skills in Low-Income Countries: Construct and Predictive Validity“. Journal of Development Economics, 164, 103132.
Questions and Findings
How do cognitive skills, socio-emotional skills (SEMS), and schooling affect labor outcomes for young adults in Pakistan? How do these effects vary by gender and migration status?
Education Greatly Boosts Female Labor Participation: Women's labor force participation (LFP) jumps from 2.4% for those with primary education or less to 17.6% for those with post-secondary education.
More Schooling Raises Wages: Each year of schooling increases monthly wages by $4.10 for men (3.4% of mean) and $1.70 for women (24% of mean), mainly due to higher LFP.
Men’s Wages Respond Strongly to Cognitive and Socio-Emotional Skills: After accounting for schooling, one SD improvement in cognitive skills raises men’s earnings by $6.30–$8.30, and SEMS by $14.60–$16.50. There are too few working women to conclude similar results.
Skill Returns for Men Depend on Migration: Among non-migrant men, SEMS strongly predict wages but surprisingly, schooling has a precise 0 relationship. Among migrants, it’s the reverse. Schooling drives earnings ($4.80 more per year), but SEMS have no effect.
Why This Matters
Policymakers lack clear evidence on how the skills developed in school, beyond just attainment, translate into better job outcomes. Understanding which skills are valued is crucial for guiding investments in education and workforce development.
This study demonstrates that both cognitive and socio-emotional skills (SEMS), not just years of schooling, contribute meaningfully to young adults’ earnings in Pakistan. However, these returns vary by migration status and gender.
It appears that those who migrate need to demonstrate strong “objective” measures of skills, such as cognitive skills. However, those that stay in their home village, where their skills are often already well known, benefit more from SEMS.
By highlighting how much skill-building can increase lifetime earnings and where measurement and policy gaps remain, the findings inform how governments should prioritize education spending and the kinds of skills to emphasize.
Test Scores, Socio-Economic Status, and Schooling Attainment
Authors: Jishnu Das, Abhijeet Singh, Andres Yi Chang
Citation: Das, Jishnu, Abhijeet Singh, and Andres Yi Chang. 2022. "Test Scores and Educational Opportunities: Panel Evidence from Five Low- and Middle-Income Countries". Journal of Public Economics, 206: 1-9.
Questions and Findings
How do test scores at age 12 and socio-economic status (SES) shape later educational attainment? When do SES gaps in learning and schooling emerge?
Test Scores and SES Predict Schooling Attainment: Both higher test scores at age 12 and higher SES lead to more years of schooling and higher college attendance by age 22.
Large SES Gaps Persist at All Achievement Levels: Even among students with similar test scores, high-SES children complete 2–3 more years of schooling than low-SES peers.
Test Scores Explain Only Part of SES Gap: Test scores at 12 account for just 15–55% of SES differences in educational attainment; family background remains a major factor.
Early and Persistent SES Gaps: SES differences in test scores are evident by age 8 and remain stable through adolescence, rarely narrowing over time.
Why This Matters
Understanding when and how learning gaps emerge and persist is critical for shaping education policy in low- and middle-income countries. Early disparities in test scores signal the need for interventions well before adolescence, as these early learning gaps have persistent effects on future educational attainment.
The findings reveal the limitations of later-stage interventions and highlight the urgency of targeting young children from disadvantaged backgrounds to reduce long-term educational inequality. By identifying the period when gaps begin and understanding their trajectory, policymakers and educators can design more effective strategies for equalizing opportunities and improving social mobility.
Learning Trajectories and Fragile Learning
Authors: Natalie Bau, Jishnu Das, Andres Yi Chang
Citation: Bau, Natalie, Jishnu Das, and Andres Yi Chang. 2021. "New Evidence on Learning Trajectories in a Low-Income Setting". International Journal of Education Development, 84: 1-26.
Questions and Findings
How much do primary school children in Pakistan learn over time? What factors explain variation in their learning gains during the primary grades?
Schooling Does Lead to Learning: Children’s math and Urdu scores rose by 1.19 standard deviations between Grades 3 and 6, with 15 - 30 percentage point item-level improvements. Dropouts show no further progress after leaving school, so this is driven by schooling not age.
Closing Test Score Gaps: Lower-scoring students made the largest gains, narrowing achievement gaps and slightly reducing overall test score variance.
Learning is Fragile: Despite the general upward trend, 20% of children experience year-to-year losses, i.e. where they answer questions they had previously gotten correct incorrectly.
Limited Role for Wealth and Gender: Parental education slightly predicts learning, but household wealth and gender explain little of the variation in learning over time.
Why This Matters
In many low-income countries, education systems struggle to deliver meaningful and equitable learning outcomes for children. Understanding how, when, and for whom learning gains occur is crucial for designing policies that promote both achievement and equity. This study shows that schools can help close learning gaps when children remain enrolled, but also highlights the fragility of gains and the limits of household background in explaining achievement differences.
Policymakers must consider how to keep students in school, ensure continued progress beyond basic skills, and address the risk that learning may stall or reverse, thereby shaping strategies for more effective and inclusive education systems.
Human Capital Accumulation and Disasters
Authors: Tahir Andrabi, Benjamin Daniels, Jishnu Das
Citation: Andrabi, Tahir, Benjamin Daniels, and Jishnu Das. 2021. "Human Capital Accumulation and Disasters: Evidence from the Pakistan Earthquake of 2005." Journal of Human Resources 57 (2).
Questions and Findings
What are the long-term impacts of environmental shocks, such as the 2005 Pakistani earthquake, on children’s physical growth and learning outcomes? What factors mitigate these effects?
Early Childhood Stunting: Children in utero during the earthquake were 3cm shorter, while those aged 0-2 were 1.5cm shorter than children who lived over 30km from the fault line.
Lower Test Scores Across All Ages: Test scores dropped 0.27 SD for children within 30km of the fault line, putting them 1.5-2 years behind their peers. These differences persisted even after schooling resumed.
Potential Long-Term Impacts on Income: If the learning deficits continue into adult life, affected students could lose up to 15% of lifetime earnings, leading to 5% total earning loss in affected villages.
Maternal Education Lessened Learning Loss, Not Stunting: Children of mothers with at least primary education did not experience any learning loss. However, maternal education did not prevent stunting.
Why This Matters
This paper sheds light on the hugely negative life-long impacts of sudden, unexpected disasters on child outcomes. Natural disasters can permanently undermine children's health and learning, reinforcing social and regional inequalities. Even short schooling disruptions can have lasting negative impacts if not corrected.
Understanding both the reach of these long-term effects and the limited protective power of maternal education reveals that rapid re-building alone is not enough to secure children’s full recovery. Policymakers and humanitarian organizations must look beyond immediate relief and invest in ongoing support for child growth and learning, particularly for the youngest and most vulnerable. Addressing these deeper, persistent effects is essential to support affected families and prevent the widening of inequalities after disasters.