Publications of Gábor Kertesi

The Roma/non-Roma test score gap in Hungary

This paper documents and decomposes the test score gap between Roma and non-Roma 8th graders in Hungary in 2006. Our data connect national standardized test scores to an individual panel survey with detailed data on ethnicity and family background. The test score gap is approximately one standard deviation for both reading and mathematics, which is similar to the gap between African-American and White students of the same age group in the US in the 1980s. After accounting for on health, parenting, school fixed effects and family background, the gap disappears in reading and drops to 0.15 standard deviation in mathematics.

Roma Employment in Hungary After the Post-Communist Transition

We analyse the magnitude and the causes of the low formal employment rate of the Roma in Hungary between 1993 and 2007. The employment rate of the Roma dropped dramatically around 1990. The ethnic employment gap has been around 40 percentage points for both men and women and has remained remarkably stable. Differences in education are the most important factor behind the gap, the number of children is important for female employment and geographic differences play little role once education is controlled for. Conditional on employment, the gap in log earnings is 0.3, and half of it is explained by educational differences.

Children of the Post-Communist Transition: Age at the Time of the Parents’ Job Loss and Dropping Out of Secondary School

Using data on children whose parents lost their jobs during the post-communist transition of Hungary, we address the causal effect of unexpected long-term unemployment of parents on their children's educational achievement. We estimate the effect of the children's age at the time of their parents' job loss on their probability of dropping out of secondary school (an event that follows the parents' job loss by many years). The treatment is an additional year reared in a family with regularly employed parents, which can be interpreted as additional human capital investment. We provide bounding estimates to the causal effect. The estimated bounds are tight, they show a substantial effect, and the effect is significantly stronger for preschool age children than for older ones.