Summary: The association between test scores in primary school and college attendance, and how this association differs by socioeconomic status (SES), remains an open question in low- and middle-income countries. Using data from long-running panel studies in Ethiopia, India, Pakistan, Peru and Vietnam, we show that children with higher test scores at age 12 report more years of schooling and higher college attendance by ∼age 22 in every country. However, variation in test scores explain only ∼15–55% of the SES gap in years of completed schooling at age 22. A striking implication is that in every country, children from low SES backgrounds who are in the 80th percentile of test scores at age 12 have similar years of completed schooling at age 22 as children from high SES backgrounds who were at the 20th percentile of test scores.

Test Scores and Educational Opportunities: Panel Evidence from Five Low- and Middle-Income Countries

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.

In low- and middle-income countries, private returns to years of schooling are now higher for those who have completed their post-secondary education. Similar to in the U.S., students from less advantaged family backgrounds are more likely to have lower test scores and to drop out of schooling. What explains test score gaps in the United States, and how policy could remediate these gaps is a key focus of current research. Yet, in low- and middle-income countries, we do not know how test scores and socioeconomic status (SES) correlate with completed schooling. This task is held back not because of conceptual complexity, but due to a lack of high-quality longitudinal data that link test scores at earlier ages to completed schooling later in life.

This is the gap we address in this paper. We examine associations between SES, test scores and college attendance in five countries— Ethiopia, India, Pakistan, Peru, and Vietnam—using data from the first longitudinal studies in LMIC that combine information on household background and test scores towards the end of primary school with a measure of schooling at age 22. These datasets are of high quality with low attrition, tests that allow us to link scores across years that were conducted by survey teams themselves. The availability of test scores at age 8 and 12 and school attainment at age 22 allows us to assess the correlation between test scores in school and school attainment at 22, as well as the evolution of test scores between 8 and 12, for children from different family backgrounds. The study countries span a large range in national income, in test scores and within-country income inequality across LMICs.

Study Design and Findings

Higher achievement is strongly correlated with more schooling, but high-SES students get more schooling at achievement levels

Children with test scores that are one standard deviation higher at age 12 report 1–2 more years of schooling by age 22 in all countries. This correlation holds after conditioning on parental education and household wealth, and holds separately for high and low-SES households. Nevertheless, at all levels of achievement at age 12, the expected years of schooling conditional on test scores is meaningfully higher for students in the top SES tercile.

Variation in test scores at age 12 explains only ∼15–55% of the SES gap in years of completed schooling

This pattern is evident in linear and non-parametric specifications. Kitagawa-Oaxaca-Blinder decompositions show that differences in test scores between low- and high-SES students at age 12 explain between 15% (in Pakistan, where the SES test-score gap is smallest) and 35–55% in the other four countries of the SES gap in schooling at age 22.

The SES gaps in test scores are mostly unchanged from ages 8 to 15

Given that test scores predict completed years of schooling in all our sample countries, we examine when these gaps emerge during childhood. In all five countries, there are substantial SES gaps in test scores by age 8, but the absolute magnitude of the SES gap does not then change appreciably between the ages of 8 and 15. The exception is Pakistan where test score gaps increase between 12 and 17, entirely due to a larger fraction of dropouts who are disproportionately concentrated in the low-SES group. This result resembles findings in the US on the evolution of test-score gaps by SES and race in different samples, where gaps that emerge early in primary schooling persist until middle school. However, we note, and show in each of our countries, that children with higher initial test scores also see the smaller subsequent increases over the four years of observation. As children from high SES backgrounds also have higher initial scores, constant absolute gaps in test scores are still consistent with a higher rate of learning for high-SES students in the primary schooling years.

The primary contribution is to present new evidence on socio-economic inequalities in education in LMIC where, in contrast to a mature literature in rich countries, such analyses have previously not been possible. That at least half of the SES gap in eventual school attainment remains even conditional on test scores suggests that policy measures to alleviate constraints for continuing high-school and college students may be important for social mobility in these settings.

Our results also provide context for a now-substantial literature on identifying effective interventions to boost test scores early on in schooling. Substantial test score differences that exist by age 8 and then persist reiterate the importance of identifying effective educational interventions targeted at disadvantaged students. The correlation of test scores with completed years of schooling at age 22 also suggests that this may translate into longer-term outcomes.

Study Resources

The following resources are for public use in presentations, papers, lectures, and more under the Creative Commons license BY-ND. Click the images below to view or download individual images, or use the button to download all.

As a condition of use, please cite as: 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.

Download All

Differences in Years of Completed Schooling by SES and Test Score Terciles

Predicted Final Schooling by Primary School Achievement and SES