research

This paper provides novel evidence on the causal effect of non-medically indicated birth interventions on patients’ health and hospitals’ follow-up care. The analysis is based on data for Germany, where the timing of births reduces uncertainty in maternity unit management, leading to induced labor being twice as common in 2022 compared to 1985. Using two years of nationwide comprehensive hospital records, the empirical design allows identifying non-random and interdependent assignment of inductions, cesarean sections, and instrumented delivery (by forceps, spatula, vacuum). Identification exploits quasi-random delivery path dependencies among healthy first-time mothers that vary the intervention likelihood of women subsequently admitted to hospital. The results reveal evidence for the key role of induction in increased prevalence of uterine ruptures, placental separation problems, and substantially impaired neonatal fitness. The detrimental impact of induction triggers an ambiguous follow-up (care) response, which highlights how leniently medical guidelines treat induction, especially compared to cesarean section.

This project aims to combine Israeli hospital data on the universe of births from 2017 onwards with records of compulsory early-childhood examinations performed by medical officers. The idea is to use my identification strategy based on relative hospital arrival times to assess the impact of physician-induced birth interventions on maternal depression, breast feeding, and spacing to subsequent births.

In this project, we combine the panel of student-level high school records (the data originally obtained for the relative age analysis) via municipality of birth and date of birth to DWD weather records for Germany. We focus on a fetus’ exposure to unusual weather events, especially heatwaves, and the implication of the in-utero treatment on secondary schooling outcomes. The analyis connects climate change driven health risks that might become ever more policy-relevant in the future to returns to education.

This paper deals with denominational schools, important providers of education in many countries around the world. Due to their focus, these schools often operate with multigrade classes, in which more than one age cohort is taught in one classroom. Multigrade classes are a cost-effective way to provide education and play a crucial role in education policy in the context of demographic change. We estimate the causal effect of attending denominational schools with multigrade classes on schooling and short-run labor market outcomes. The analysis combines administrative records of schools with comprehensive population census data, and exploits the abolition of Catholic or Protestant schools in the Saarland, a German state, in 1969, for identification of the effect. The findings document significantly detrimental effects on final grade attainment and labor market participation. Notably, the negative impact is most pronounced in the outcomes of girls.

This paper turns to intentional modern multigrade settings created for pedagogical opporturnity. While teaching more than one age cohort of pupils in one classroom has been advocated as a cost-effective way to provide education and a means of fostering academic achievements through intensified interactions, the evidence on the effects of multigrade classes on performance is mixed. We provide novel evidence on the causal effect of multigrade teaching in primary schools on literacy skills by the end of primary school. The analysis is based on student test score data of more than 68’000 fourth-graders and exploits the staggered introduction of policies targeted at making entry to primary schools more flexible across German states between 2001 and 2016 for identification. The results from a difference-in-differences design document that attending multigrade classes had negative effects on reading test scores and German grades. These negative effects are more pronounced for girls.

In this paper I use the universe of students of a high school in Germany over the period 2002-2022, to investigate the effect of relative age on various schooling outcomes. OLS regressions reveal that children that are relatively younger than their class mates perform better, even when conditioning on age (in days, which is unobserved in related work). In contrast, an instrumental variables regression using birth dates relative to a series of changes in legal thresholds for first school enrolment as instrument for relative age in class reveals the opposite: younger children perform worse. This indicates that selection in first school enrollment plays an important role for school outcomes, suggesting that knowledge on the part of parents leads to young children entering school more likely when they are sufficiently mature so that their performance is high, whereas a mechanistic enrollment would likely harm other kids that benefit from being enrolled one year later.

This project relates to political debates concerned with rigidities and path dependencies in the German education system. For identification we exploit exogenous variation across North-Rhine Westphalian municipalities and birth cohorts in primary school multigrade exposure. We explore whether students’ early life experiences in classrooms with a high age-gradient yield persistent effects in terms of no-shows and sick notes during secondary schooling conditional on current relative age rank and absolute age (in days).

This paper inspects the effects of the Colombian peace agreement of 2016 between the National Government and the Revolutionary Armed Forces of Colombia (FARC) on several violent outcomes. The methodology relies on estimating a set of event study designs considering the treatment group as the Zones Most Affected by the Armed Conflict (ZOMAC) using municipality data between 2012 and 2019. The primary findings are that the peace agreement reduced violent outcomes such as kidnappings, subversive actions, and attacks on police facilities while extortion cases increased. Our findings are stable across samples selected %through the likelihood to be part of the treatment group under a propensity score matching approach or alternative specifications employing a triple differences design.