The Value of a Peer, joint with Ulf Zölitz, reject & resubmit at Journal of Political Economy
This paper introduces peer value-added, a new approach to quantify the total contribution of an individual peer to student performance. Peer value-added captures social spillovers irrespective of whether they are generated by observable or unobservable peer characteristics. Using data with repeated random assignment to university sections, we find that students significantly differ in their peer value-added. Peer value-added is a good out-of-sample predictor of performance spillovers in newly-assigned student-peer pairs. Nonetheless, students’ own past performance and other observable characteristics are poor predictors of peer value-added. Peer value-added increases after exposure to better peers and valuable peers are substitutes for low-quality teachers.
On the Extent, Correlates, and Consequences of Reporting Bias in Survey Wages, joint with Marco Caliendo, Katrin Huber and Jakob Wegmann, revise & resubmit at Journal of Labor Economics
Surveys are an indispensable source of data for applied economic research; however, their reliance on self-reported information can introduce bias, especially if core variables such as personal income are misreported. To assess the extent and impact of this misreporting bias, we compare self-reported wages from the German Socio-Economic Panel (SOEP) with administrative wages from social security records (IEB) for the same individuals. Using a novel and unique data linkage (SOEP-ADIAB), we identify a modest but economically significant reporting bias, with SOEP respondents underreporting their administrative wages by about 7.3%. This misreporting varies systematically with individual, household, and especially job and firm characteristics. In replicating common empirical analyses in which wages serve as either dependent or independent variables, we find that misreporting is consequential for some, but not all estimated relationships. It turns out to be inconsequential for examining the returns to education, but relevant for analyzing the gender wage gap. In addition, we find that misreporting bias can significantly affect the results when wage is used as the independent variable. Specifically, estimates of the wage-satisfaction relationship are substantially overestimated when based on survey data, although this bias is mitigated when focusing on interpersonal changes. Our findings underscore that survey-based measures of individual wages can significantly bias commonly estimated empirical relationships. They also demonstrate the enormous research potential of linked administrative-survey data.
Feedback, Confidence and Job Search Behavior, joint with Lukas Hensel, Jonas Radbruch, Marc Witte and Tsegay Tekleselassie, submitted, pre-registered at AEA RCT registry (AEARCTR-0009698)
Job seekers face uncertainty about their abilities, and whether these match with job requirements. Such uncertainty may result in sub-optimal job search outcomes and job matches. We conduct a field experiment among job seekers in Addis Ababa, Ethiopia. Participants underwent a skill assessment and were asked about their willingness to pay (WTP) for information about their relative performance on a test of general intelligence. WTP is positive for about 80 percent of the population, and is associated with gender and personality. Feedback provision leads individuals to update their beliefs which only persists for individuals with low WTP. We provide evidence that suggests imperfect recall as potential mechanism for the lack of persistence. Feedback increases job search intensity but relatively lower for initially overconfident individuals and those with negative or zero WTP, which results in lower realized wages for these groups. The heterogeneity in belief updating, recall and job search behavior is consistent with some overconfident job seekers being unable to forget information and, thus, to maintain motivated beliefs, but being sophisticated about this inability to forget.
Adoption and Effectiveness of AI Tutoring in Tertiary Education, joint with Henning Hermes, pre-registered at AEA RCT registry (AEARCTR-0014959)
Abstract: We evaluate the adoption and effectiveness of AI-based tutoring in tertiary education using a large-scale randomized encouragement design. In collaboration with a major online education provider, we invite a randomly selected subset of students (N ≈ 7,000) to engage with an AI assistant integrated into the digital learning platform. The assistant, powered by GPT-4 and fine-tuned on course materials, provides personalized support by answering content-related questions, referencing study materials, and offering examples. Our analysis focuses on three key aspects: (i) selection into AI usage and its determinants, particularly digital skills; (ii) usage patterns across student subgroups; and (iii) causal effects of AI usage on study progress and academic performance, measured via administrative data.
Sexual Harassment in German Firms, joint with Jonas Jessen and Valentina Götz, pre-registered at AEA RCT registry (AEARCTR- 0015274)
Abstract: We study the prevalence, perceived costs and consequences of sexual harassment (SH) in German workplaces. We first use a discrete choice experiment to estimate workers' willingness to pay (WTP) for workplaces without a history of known SH cases and preventive firm measures. Women, particularly early in their careers, display the highest WTP. Preventive measures significantly increase the attractiveness of workplaces, even when there is a history of SH. Motivated by these results, we then document SH experiences using new data from the Linked Personnel Panel (LPP) and the IAB-OPAL online panel. SH is widespread: 20 percent of employees have either experienced SH at work personally or in their close work environment. Women are affected significantly more often than men. Women are also less likely to trust that leadership will respond appropriately to reported cases, and this lack of trust correlates with higher experienced incidence rates. Firms with active complaint procedures and preventive measures report greater employee awareness and more open discussion of SH. Taken together, our findings provide a strong economic rationale for preventive policies.
Schools and Teacher Mental Health, joint with Antonia Entorf and Miriam Gensowski
Abstract: We document elevated and rising rates of mental health-related health care utilization among primary school teachers in Denmark relative to other professions. Using rich administrative data, we show that these patterns are driven by the occupational demands of teaching—particularly high emotional labor and intense social interaction—rather than selection into the profession. Mental health outcomes vary substantially across schools, with large differences in teacher absenteeism, diagnoses, and prescription drug use. These differences are not well explained by observable school characteristics, but they correlate strongly with student-reported measures of classroom functioning and well-being. Leveraging a quasi-experimental mover design and event-study framework, we provide causal evidence that teachers' mental health responds to school-specific environments: teachers who move from low-stress to high-stress schools experience a deterioration in mental health outcomes.
Changing Work Climate: The Implications of Extreme Temperatures for Occupational Choice, joint with Nico Pestel, Harald Pfeiffer and Marc Witte
Abstract: Extreme weather events, intensified by climate change, impact working conditions for many jobs. As these events become more frequent, workers in outdoor-exposed occupations may reconsider their choices due to anticipated declines in conditions. Our research investigates how exposure to extreme temperatures influences occupational choices among present and future workers. We use a discrete choice experiment to estimate workers' willingness to pay (WTP) to avoid outdoor-exposed jobs, after eliciting beliefs about future likelihood of extreme weather events. We then we examine, based on administrative data from Germany and the Netherlands alongside detailed temperature records, whether workers' decisions to enter or leave outdoor-exposed occupations are influenced by extreme heat or cold episodes. Our key findings reveal that such events decrease the number of new apprenticeships in outdoor-exposed occupations and raise the likelihood of occupational changes, particularly towards less outdoor-exposed jobs. These shifts could exacerbate labor shortages in sectors crucial for transitioning to green energy, amplifying the societal costs of climate change.
Productivity Spillovers in International Teams: Insights from GitHub Activity Data, joint with Felix Holub and Beate Thies
Identifying productivity spillovers – peer effects – among high-skilled white-collar workers is notoriously difficult, due to lack of measurement of immediate effort and productivity, selective peer formation and simultaneity issues. Based on time-stamped GitHub activity from professional software developers contributing to collaborative projects, we identify endogenous peer effects in knowledge worker productivity through exogenous productivity shocks to varying shares of co-workers. Transitions into Daylight Saving Time (DST) by North-American co-workers lead to sizable reductions in productivity, measurable in quantity and quality of actions. These productivity shocks spill over to non-affected co-workers, indicating strong complementarities in software production. The spillovers are not uniform but driven by more experienced collaborators, particularly those in managerial or reviewer roles, and primarily borne by inexperienced focal developers.
1. The Returns to Language Skills in the US Labor Market, 2012, IZA Discussion Paper #7080, joint work with Mathias Sinning
2. Risky Business. The Role of Risk Attitudes in Occupational Decisions, 2010, Ruhr Economic Papers #187