Working papers and work in progress

Early Decisions, Long-Lasting Consequences: The Macroeconomics of School Tracking

with Lukas Mahler.

10/2026.

Abstract

School tracking—sorting students into different school programs—is widespread and shapes how students accumulate skills. We study its aggregate, distributional, and intergenerational effects in a general-equilibrium overlapping-generations model with endogenous school-track choices and a novel skill formation technology trading off learning gains from homogeneous groups against the risk of misallocating children whose skills are still developing. Calibrated to Germany’s early-tracking system, the model shows that delaying tracking from age 10 to 14 improves intergenerational mobility and reduces inequality but generates modest efficiency losses. These losses arise only in general equilibrium, as a falling college wage premium dampens college expansion.

Private Information in the Family

with Matthias Doepke and Michèle Tertilt.

03/2026.

WP RFBerlin
Abstract

Standard family models assume spouses share information. We challenge this with theory and evidence from a new survey module in the Dutch LISS panel, where spouses independently report knowledge of each other’s finances. Private information is pervasive: in 40% of couples, at least one partner lacks full knowledge of the other’s income. Using a mechanism design approach, we show that constrained-efficient allocations require no full revelation, and predict that income and consumption shares co-move only under imperfect information. LISS data confirm this: the income-consumption relationship appears only among poorly informed couples, even after controlling for limited commitment.

Residential and Social Mobility: A Quantitative Analysis of Parental Decisions with Social Learning

Revise and Resubmit, AEJ: Macroeconomics.

Abstract

In the U.S., beliefs about parenting and parental choices differ across socioeconomic groups. This paper develops an overlapping generations model with social learning, where young adults form beliefs about parenting by observing neighbors but may make inference mistakes due to selection neglect—imperfect correction for selection. In equilibrium, residential segregation generates information frictions that foster heterogeneous beliefs and distort parenting decisions. When calibrated to the U.S., the model suggests these beliefs reduce equity and efficiency, and housing vouchers can enhance both social mobility and aggregate income. In contrast, a model that replaces beliefs with fixed preference heterogeneity predicts income losses.

Employment Transitions: Mechanisms for Gender Earnings Gap

with Valentina Melentyeva.

Abstract

We use German administrative data to compare women’s and men’s career choices and labor income profiles over their working lives. We find a significant and robust gender gap in job-to-job transitions. When working full-time, women are less likely than men to transition from one employer to another. The first finding is that while commuting preferences and sector selection do not explain this gap, the birth event largely explains it. Second, we want to quantify the consequences of the gender gap in job-to-job transitions for the gender earnings gap. If full-time job-to-job transitions increase wage growth, this phenomenon most likely contributes to the well-documented child penalty.

Publications

Estimating Impact with Surveys versus Digital Traces: Evidence from Randomized Cash Transfers in Togo

with Emily Aiken, Joshua Blumenstock, Dean Karlan and Chris Udry.

Journal of Development Economics (2025).

Abstract

We study whether program impacts can be estimated using a combination of digital trace data and machine learning. In a randomized controlled trial of cash transfers in Togo, endline survey data indicate positive treatment effects on food security, mental health, and perceived economic status. However, estimates of impact based solely on predicted endline outcomes (generated using trace data and machine learning, which do successfully predict baseline poverty) are generally not statistically significant. When post-treatment outcome data are used in conjunction with predictions to estimate treatment effects, predicted impacts are similar to those estimated using surveys.

When Effective Teacher Training Falls Short in the Classroom: Evidence from an Experiment in Primary Schools

with Adrien Bouguen, Marc Gurgand, Valérie Munier and André Tricot.

Economics of Education Review (2024).

J-PAL WP
Abstract

Although in-service teacher training programs are designed to enhance the performance of several cohorts of students, there is little evidence on the persistence of their effects. We present the two-year results of a randomized study of an intensive in-service teacher training program conducted in France during and after the training program’s implementation. Our results highlight the short-run effectiveness of the training program: it successfully improves students’ performance but only during the implementation year. A detailed analysis of teachers’ outcomes indicates that teachers changed their pedagogical vision and practices but afterward struggled to apply skills to contents not directly covered during training.

Machine Learning and Mobile Phone Data Can Improve the Targeting of Humanitarian Assistance

with Emily Aiken, Dean Karlan, Chris Udry and Joshua Blumenstock.

Nature (2022).

CEGA
Abstract

Targeting is a central challenge in the administration of anti-poverty programs: given available data, how does one rapidly identify the individuals and families with the greatest need? Here we show that non-traditional “big” data from satellites and mobile phone networks can improve the targeting of anti-poverty programs. Our analysis compares outcomes – including exclusion errors, total social welfare, and measures of fairness – under different targeting regimes. Relative to other feasible targeting options, the machine learning approach reduces errors of exclusion by 4-21%. These results highlight the potential for new data sources to contribute to humanitarian response efforts, particularly in crisis settings when traditional data are missing or out of date.