This article examines how military losses shape public engagement with propaganda. We combine data on Russia's military losses in its war against Ukraine with unique data from social media groups of Russian public schools to estimate the effect of war fatality reports on engagement with patriotic and pro-regime propaganda. Exploiting idiosyncrasies in the timing of these reports for identification, we document a sustained decline of up to 36% in likes, shares, and views for content promoting the authorities after the first such report in a municipality. Engagement with patriotic propaganda increases, but this increase is limited to the first year of the invasion and at some point the effect reverses. The effects are most pronounced in educated, populous, and young municipalities and where the obituaries appear in school groups, highlighting how information access shapes the public response to propaganda. Further analysis of the obituaries with a large semantic model shows that users engage most with personal stories of the KIA soldiers and disengage when their deaths are framed within nationalistic and pro-regime narratives. Together, these patterns suggest that war fatalities can hinder the spread of propaganda and erode public support for and reach of the state-controlled narratives during the conflict.
The effect of the first KIA report in municipality on engagement with patriotic (left) and pro-regime propaganda (right)
Journal of Historical Political Economy: Vol. 2: No. 1, pp 65-87.
In this paper, I address collective memory as a potential transmission vehicle and study how the commemoration of political violence might promote the associated effects. I exploit variation in the location of 1930s political arrest sites in Moscow and in the locations of memorial plaques that commemorate these arrests. I find that individuals currently residing nearby the arrest sites are less likely to engage in pro-social behavior, namely online donations. Most importantly, the effect appears insignificant in the absence of commemoration. These findings suggest that commemoration and collective memory revitalization might play a crucial role in the persistence of historical legacies even in transient communities.
The 1937 arrests (left) have no effect on the intensity of online donations (right) unless there is a commemorative plaque installed nearby.
with Aleksei Kiselev
This study investigates whether news about corruption can affect the likelihood of sovereign default by examining the impact of Alexei Navalny's blog posts on Russian sovereign debt markets from 2008 to 2011. It focuses on how Navalny's blog posts about corruption in state-affiliated companies influenced sovereign spreads after the Global Financial Crisis. The analysis uses data on sovereign spreads and the timing of Navalny's posts, showing that unexpected disclosures of corruption in Russian state-owned enterprises (SOEs) significantly increased sovereign borrowing costs. The findings indicate a statistically significant relationship, with corruption revelations explaining up to two-thirds of daily variation in sovereign spreads in 2008-2009, especially when oil prices were low. This research underscores the importance of sovereign credibility, the role of independent media in exposing corruption, and the broader economic mechanisms beyond sovereign debt. It highlights the critical role of information transparency and anti-corruption efforts in maintaining sovereign credibility and debt market stability.
All posts
In-depth corruption investigations
The figure above reports the rolling window estimates for the days around Navalny's blog posts. While the impact of ordinary posts fades out by the second half of 2009, the in-depth investigations pertain their effect on the sovereign spreads.
We explore the impact of fear on demand for government intervention. Our empirical strategy takes advantage of a unique quasi-experiment: During our survey in Russia, a very popular television program misreported the riskiness of the Covid-19 virus, thereby providing short-term exogenous variation in the fear of infection. Using a shift-share instrument based on this media intervention, we estimate a causally positive effect of fear on popular approval of a variety of government interventions. The magnitudes of this effect are remarkable, with a one-standard-deviation change in fear explaining as much as 77-103 percent of a standard deviation in our aggregate measure of demand for regulation, and 99-167 percent and 85-96 percent of a standard deviation in obligatory mask wearing and stay-home orders, respectively. However, fear had little impact on demand for interventions unrelated to the virus, such as censorship and housing policy. We explore potential mechanisms and establish that fear heightens perceptions of noncompliance with safe behaviour by others — a mechanism in line with a neoclassical view that free-riding concerns increase calls for government intervention. Our study informs debates on the demand for regulation, the role of emotions in shaping policy preferences, and the impact of media on political attitudes.
with Ekaterina Borisova and Koen Schoors
The Neolithic transition, marked by the shift from hunting and gathering to sedentary farming and herding, has been proposed as a potential contributor to persistent changes in labor-related gender norms and ongoing gender inequality in the labor market. In this study, we test this hypothesis using a novel dataset of sustainable genetic markers, specifically Y-DNA haplogroups associated with the Neolithic transition, to trace Neolithic ancestry in contemporaneous populations. Our findings indicate that countries with higher levels of Neolithic ancestry have lower female labor force participation rates and residents with more gender-discriminatory attitudes towards female labor market participation. Conversely, locations with higher levels of hunting-gathering ancestry exhibit the opposite trend. These results are consistent at the country level, individual level, and for subsamples of second-generation migrants with respect to the agricultural ancestry of their country of origin. They also remain robust to the use of alternative datasets of genetic markers and the inclusion of a wide range of control variables. Moreover, our Neolithic ancestry measure is a better predictor than related measures such as years since the Neolithic transition, grammatical gender intensity or plough use. We conclude that differences in gender roles and norms can be traced back to a persistent cultural response to the change in the division of labor between genders that resulted from the Neolithic Revolution.
OLS estimation results performed on a sample of second-generation migrants in the US show that female respondents whose parents come from countries with a longer history of sedentary agriculture are less likely to be employed.