Work in progress

How do independent media affect regime support in an autocracy? We carry out two field experiments in Russia by promoting the country’s only independent online TV channel to a randomized sample of cities and individuals before the 2016 parliamentary elections. In both experiments, we find that independent media foster polarization, increasing turnout and progovernment votes among regime supporters, and reducing it for nonsupporters. The effect, however, holds only for voters who rely on news from social media; among consumers of traditional media, our treatment uniformly decreases regime support. Our results highlight how social media can mediate the effect of independent media in autocracies.

We analyze media repression in Putin’s Russia (2004–2019), a smart dictatorship that mimics democratic institutions, notably relatively free elections and a relatively free press. Drawing on a unique granular dataset on journalist harassment and the pre-determined, staggered timing of local elections, we find evidence of strong political cycles of media repression. This media repression ahead of elections leads to a more favorable tonality of the news coverage of incumbents. Free press and free elections are temporally decoupled, thus disallowing them to work as effective accountability mechanisms. This secures the dictator’s power while upholding an image of competence and democratic rule.

This paper probes the persuasive power of authoritarian propaganda on social media and compares it to a similar campaign by the opposition. We carried out a field experiment during the 2018 presidential elections in Russia by collecting voting intentions and then randomly advertising a pro- and anti-regime YouTube video during a telephone survey shortly before the elections. During a follow-up interview after the elections, we gathered data on actual voting decisions to measure the effect of our treatments. We find that state propaganda on social media was effective in persuading respondents who consumed exclusively offline news sources, but failed to engage respondents consuming online news and even decreased their propensity to vote for the incumbent, conditional on their dissatisfaction with the government's performance. The anti-regime video, on the other hand, consistently decreased voting for the incumbent among both groups of respondents. Our results refute a widespread belief about the potency of authoritarian propaganda by showing its inability to engage with individuals outside its traditional audience.

This paper examines the impact of outdoor temperature on media bias. We use 12 years of daily hand-coded data on the tonality of news broadcast by the three major US news networks, ABC News, CBS News, and NBC News, all headquartered in New York City, and merge it with granular, geospatial weather data. Our identification strategy exploits detailed variations in local daily high temperatures to estimate the effect of heat on media bias in news reporting about the Republican and Democratic parties, controlling for time and network-month fixed effects. We find a positive effect of a substantial magnitude: a 1°C increment in daily maximum temperature on a hot day (>25°C) leads to a 20% increase in the media bias measured as the difference in the share of negative news about the Republicans and the Democrats. This effect exists only for maximum temperatures, as opposed to minimum or average temperatures. The results are robust to placebo tests using past or future temperatures. Our findings extend the previously established link – from hot temperatures to negative affect and a decline in cognitive ability – to the determinants of media bias.

This paper investigates the causal impact of direct cash transfers on health, well-being, and life satisfaction among adolescents in Russia during the pandemic years. We study an unanticipated introduction of a COVID-19 state support policy for families with children in the form of two one-off transfers (140$ each) for all children younger than 16 years on the date of the policy approval. Our quasi-experimental design takes advantage of the sharp age discontinuity allowing us to compare the health and well-being of adolescents marginally eligible and marginally ineligible for these cash transfers using the data from a Russian longitudinal survey. The panel structure of the survey data enables the difference-in-difference approach as we control for past values of the outcome variables in the pre-pandemic year. We find a significant positive effect of cash transfers on adolescents’ material well-being, life satisfaction, and self-assessed health in the first and second pandemic years. Placebo estimations and various robustness tests confirm the validity of our results.

This paper investigates the effect of Venezuelan transit migration on crime rates in Colombia. We exploit the reopening of the VenezuelaColombia border in 2016, which has led to a surge in transit migration, and geospatial information about the distinct routes through which the migrants crossed Columbia. Employing a difference-in-differences approach and propensity score matching, we find that transit migration increased property crime rates in crossed municipalities, with both native Colombians and Venezuelan refugees seeing higher victimization rates. Violent crimes remained unaffected. This is the first study to document a link between transit migration and crime. 

We investigate the effect of elections on underreporting COVID-19 mortality, measured as the difference between excess mortality and official statistics. Our identification strategy takes advantage of a natural experiment of the unanticipated onset of the Coronavirus pandemic in 2020 and the asymmetric electoral schedule of presidential elections around the world, in which some countries faced the pandemic with upcoming elections in the next two years, while others did not have this electoral pressure. Contrary to conventional wisdom that governments manipulate information downwards to enhance reelection probabilities, we find that democratic governments facing elections in the following years report COVID fatalities more truthfully. We explain the result by a potential aversion to the costs associated with exposed underreporting: using Gallup poll data for 2020 we show that underreporting of COVID-19 mortality potentially undermines trust in government but only in relatively democratic countries.