Regime Loyalty during Wartime: Evidence from Nazi Germany 

with Felix Haass, Sascha Riaz and Julian Voss

Manuscript

Measuring regime support in closed autocracies is notoriously challenging due to preference falsification, state censorship, and pervasive propaganda. We introduce a novel behavioral measure of regime loyalty based on subtle expressions of allegiance in soldier obituaries published in Nazi Germany between 1939 and 1945. Our empirical analysis draws on a large-scale dataset of over one million scanned pages from roughly 160,000 newspaper issues across 260 unique local news outlets. Using Large Language Models for OCR and data labeling, we detect expressions of regime support, such as praise for Hitler, National Socialism, or the Fatherland, in approximately 600,000 obituaries. Our approach yields the first spatially and temporally granular measure of Nazi regime support during World War II. Our descriptive findings nuance the prevailing historical consensus: we find that regime loyalty began to erode immediately following the 1941 invasion of the Soviet Union, not after the Battle of Stalingrad. By contrast, militaristic rhetoric emphasizing soldiers' heroism persisted at high levels throughout the war.

Parliamentary Representation and Right-Wing Violence: Evidence from Nazi Stormtrooper Riots in the Weimar Republic

with Daniel Bischof, Felix Haass, and Henry Thomson

Manuscript

Radical right-wing movements often resort to violence, particularly in declining democracies. While elections are expected to channel such conflict peacefully, they can often also fuel right-wing unrest. In this paper, we investigate the conditions under which elections can drive right-wing violence at the local level. We study Nazi paramilitary riots after the July 1932 Reichstag elections in Weimar Germany. We match data on street violence from digitized Prussian police records to the home towns of Nazi party candidates. Using a regression discontinuity design, we compare towns of candidates who narrowly won mandates to those of candidates who narrowly lost. We find that post-election violence was significantly higher in the home towns of elected Nazi candidates. Supplementary analyses demonstrate that these differences were driven by increased frustration among local stormtroopers blocked political pathways to power and enhanced recruitment capacity in mandate winner's home town. Our findings have important implications for our understanding of electoral violence after elections, the consequences of radical right-wing representation, and how elections can erode democratic stability.

Child Penalties in Autocratic Bureaucracies - Motherhood, Loyalty Signals, and Career Paths in the German Democratic Republic

with Felix Haass and Jan Pierskalla

Manuscript 

Many autocracies publicly emphasize gender equality and adopt policies to foster women’s career advancement, yet the higher echelons of their bureaucracies remain overwhelmingly male. What explains this discrepancy? While gender and child penalties are well-documented in Western labor markets, much less is known about comparable penalties inside autocratic bureaucracies. We study the German Democratic Republic (GDR), a most likely case for low penalties given strong equality laws and extensive childcare. Analyzing biographical data on more than 370,000 cadres, we find a large and persistent motherhood penalty: women’s promotion prospects fall sharply after the first birth, whereas men’s careers continue on trend. Additional analyses indicate that motherhood reduced time-intensive regime engagement, thereby reducing women’s ability to send expected signals of regime loyalty. These findings have important implications for understanding bureaucratic politics in autocracies, the logic of child penalties in high-profile, effort-demanding professions in Western democracies, as well as gendered legacies of autocratic rule.

Long-Term Sequences of State-Building and Violent Conflict

Manuscript

A large body of research demonstrates that state capacity is an important correlate of intra-state conflict. However, similar levels of present-day state capacity can result from different historical trajectories of state-building. This paper investigates how the spatio-temporal patterns of long-term expansion of state authority and control can influence current levels of violent unrest. In particular, I argue that specific sequences of state-building can increase the present-day risk of violent conflict: geographical obstacles can isolate territories from state-building processes for extended periods, strengthening non-state institutions. The risk of violence increases once the state has accumulated sufficient capabilities to close this state-building gap. The extension of authority into peripheral territories triggers resistance from among local elites that can mobilize their constituencies against the central state. I investigate this argument in a mixed-methods design that exploits exogenous variation in early state-penetration in Nepal. Areas located below the Himalayan foothills were heavily Malaria-infested preventing the state from establishing meaningful presence until the implementation of a malaria eradication program in the 1950s. I make use of this specificity in Regression Discontinuity (RD) estimations and a qualitative process tracing approach. Findings lend support to the primary arguments of the paper.