The effect of refugees on native adolescents’ test scores: quasi-experimental evidence from PISA

Semih Tumen

Journal of Development Economics, Volume 150 (2021)

Working paper available here. 


This paper investigates the impact of Syrian refugees on the school performance of adolescent children in Turkey. Earlier research suggests that immigration may affect native children’s school performance through two main mechanisms that operate in opposite directions: (1) the labor market mechanism that improves the educational outcomes of natives and provides additional incentives to continue education due to increased competition for available jobs in the low-skilled segment of the labor market; and (2) the educational experience mechanism that negatively affects the educational outcomes of natives because the interaction between native and immigrant students in school environments has a negative effect on the quality of education (for example due to lower-quality peer interactions, language barriers, and looser teaching standards).

The analysis is based on the OECD’s Programme for International Student Assessment (PISA) microdata, which captures standardized test scores for 15-year-old students. The author uses 2009 and 2012 data (prior to the refugee influx) together with 2015 and 2018 data (after the refugee influx). The author isolates the impact of the labor market mechanism from the educational experience mechanism by separately estimating effects using 2015 data (when Syrian refugees were excluded from secondary education, and consequently the education experience mechanism did not operate) and 2018 data (when both the labor market and education experience mechanisms jointly operated and together affected the test scores of native adolescents).

The paper accounts for the common “endogeneity issue”, i.e., the likelihood that the location choices of refugees are not random, by using standard econometric techniques, such as a difference-in-differences approach (based on province-year variation in refugee intensity) and an instrumental variable strategy.

Key findings:

  • Math, Science, and Reading scores of Turkish native adolescents increased following the Syrian refugee influx, conditional on parental education (a proxy for unobserved ability).
  • The increase in test scores mostly comes from the lower half of the test score distribution, and from students with lower maternal education (mothers with less than a high school education). This suggests that the refugee influx has reduced the test score inequality among natives.
  • There is suggestive evidence that the PISA scores of male adolescents increased more than those of females.
  • As the presence of refugee adolescents increases in the Turkish education system (2018 data), the education experience mechanism starts eroding the test scores gains that are initially obtained through the labor market mechanism.

The author argues that the labor market forces that emerged in the aftermath of the Syrian refugee crisis have led native adolescents, who would normally perform worse in school, to invest in their human capital more intensively, i.e. the increase in refugee concentration generates pressures in the low-skill labor market and those pressures provide incentives for increased school achievement.