Governments have long used mass media narratives as a policy tool to shape social norms and behavior, yet causal evidence on the effects of such interventions remains scarce. This paper studies the case of the U.S. government's coordinated media campaigns during World War II, which deliberately reshaped the public image of women to address wartime labor shortages. I construct a new dataset combining historical newspapers with a newly digitized archive of women's magazines, which allows to trace changes in representations of women before, during, and after the war. Exploiting local variation in war-induced labor shortages and using an event-study design, I document a pronounced wartime shift toward portraying women as workers, followed by a rapid reversal once the war ended. While this narrative was short-lived, areas more exposed to wartime pressures experienced a persistent increase in the visibility of women in public discourse. To isolate the causal effect of media narratives from local labor market conditions, I combine historical survey data with a shift-share strategy based on magazine circulation and national changes in content. Exposure to wartime narratives significantly shaped postwar attitudes toward women's roles and political participation, with effects that outlasted the economic shock itself. The results suggest that institutional messaging can rapidly reshape social narratives, even in non-authoritarian settings, with unintended consequences that may persist beyond the original shock.

ESP Workshop at the Harvard Kennedy School, Graduate Workshop in Economic History at Harvard University, Harvard CID Research Xchange, Third Political Economy CCA-Cornell-Bocconi Winter Workshop (poster)

Armenia and Azerbaijan, engaged in the Nagorno-Karabakh conflict, are known for their strong son preference and widespread use of abortion. In such a peculiar setting, we ask: does ethnic conflict affect son preference? By merging Demographic and Health Survey (DHS) data with two different sources of conflict-related casualties, we investigate the effect of ethnic conflict on fertility decisions. To do so, we leverage variations in geographic and temporal exposure to conflict-related violence in the context of the First Nagorno-Karabakh War (1988–1994), alongside variation in the gender composition of children between families. We find that families with only daughters exposed to conflict increase their fertility in the very short term. Furthermore, our analysis of births by gender reveals that this result is driven by the birth of sons. Conflict intensifies son preference, manifesting in son-targeted fertility behavior and sex-selective abortions. Lastly, we identify child replacement as a mechanism contributing to the increase in fertility, driven primarily by the replacement of male children.

Harvard Development Lunch, 2026*; SAEe 2026 Symposium of the Spanish Economic Association (Universitat Autònoma de Barcelona)*, 5th DevEconMeet, 2025 (University of Florence)*; Fragile Lives 2025 (International Security and Development Center); ArmEA 2025 (American University of Armenia)*; 2025 Mend the Gap in Economic Opportunities in Europe and Central Asia (World Bank, Axa Gender Lab at Bocconi University & EIEF)*; 2025 BAEM (Bristol Applied Economics Meetings) Workshop on Gender, Diversity and Human Capital; 2024 FROGEE Academic Conference (SSE); IELM-DD Internal Seminar (Centre d' Économie de la Sorbonne)
*indicates presentation by co-author