Research

Can political leaders use means of communication directly to consolidate their power? After the electoral victory of Hugo Chávez in 1998 and in the span of six years, Venezuela transitioned from a democracy into an autocracy. State propaganda directly used by Chávez in the shape of Aló Presidente played a crucial role in this process. I analyse the impact of exposure to Radio Nacional de Venezuela (RNV) relying on the natural ruggedness of the terrain as a source of plausibly exogenous variation. Additional variation over time in RNV's content allows me to rule out that results are due to exposure to radio in general or radio under Chávez's regime: results appear only after Aló Presidente was introduced. I identify an increase in Chávez's vote share in his first re-election, which cannot be fully accounted for by changes in electoral turnout. Underlying mechanisms include the appeal of a narrative of democracy as well as a deeply personalised style of communication.

This paper studies the economic and political effects of the legacy of African slavery in Venezuela. I find a negative effect of African slave ancestry on development outcomes and a positive effect in support of Hugo Chávez. The mechanism of historical persistence is low-human capital and occupational specialisation. To identify slave ancestry in contemporary Venezuela, I take advantage of a unique colonial norm. When slaves were brought into the continent they were registered and received the name of their ethnic groups or place of origin as surnames. This created a unique set of surnames that have persisted through time and that denote African slave ancestry. Following a two-stage least-squares strategy, I identify an effect of slave ancestry on contemporary measures of economic development. Findings include lower night-light density, literacy rates, and higher child-mortality rates. I find suggestive evidence of the historical persistence of this effect through low-human capital and poverty proxied by educational attainment, the structure of the economy, and asset-ownership. I identify a strong and consistent effect of slave ancestry in support of Hugo Chávez during his first presidential run. This effect is not fully accounted for by low economic development. Individuals with slave ancestry do not live in locations where slavery existed during the colony, measured by proximity to colonial cocoa haciendas or slave ports of arrival. The latter locations are not economically worse-off and were not more likely to support Chávez. These findings emphasise differences between mobile (e.g., culture and human capital) and immobile (e.g., laws) factors of historical persistence, provide a historical explanation to factors contributing to the rise of Chavismo, and highlight the pervasiveness of slavery even in fringe areas of a colonial empire. 

In 2013 the government of the state of Coahuila, Mexico, implemented a unique package of policies with the objective of combatting the presence of drug cartels. Two crucial dimensions included strengthening local police enforcement, and banning businesses that provided liquidity to cartels such as brothels, junk-yards, and casinos. Using synthetic controls, I test the effectiveness of this set of policies against intentional homicides---the metric targetted by Coahuila's government---as well as all other crime-types. I find a modest reduction of crimes related with the presence of cartels, and no impact on crimes unrelated to cartels. I extend the analysis to assess the economic effects of these policies. I find no statistical evidence of economic dividends of peace beyond mechanical effects parsimoniously explained by the policies adopted, such as a reduction in informality. As a practical contribution, the paper provides software for Stata that allows users to calculate confidence intervals for all the post-treatment periods for the time series of an effect calculated via synthetic controls using cross-validated Lasso. 

This paper applies current techniques in natural language processing to quantitatively analyse Hugo Chávez’s rhetoric. I scrape the transcripts of all broadcasts of Aló Presidente and all the speeches that Chávez gave. I quantify the complexity in the language used by Chávez along three dimensions: sentence complexity, word complexity, and conceptual complexity. I also quantify the distribution of topics he covers, and the political slant of his rhetoric. I can measure how his rhetoric adapts according to his audience or in response to external factors.

The Maisanta list was leaked in Venezuela in 2004 to intimidate Chávez’s opposition. The list contains all the personal information of adults registered to vote in Venezuela, and whether they signed a petition to summon a recall referendum against Chávez. The political implications of the distribution of the list remain unexplored. Combining the Maisanta list with newly digitised political data, I can observe the behaviour of Venezuelan voters aggregated at the level of the polling booth before and after the data was leaked. I interrogate incentives for political participation in the face of an authoritarian regime after the principle of a secret ballot has been violated.

The paper analyses whether the gradual rollout of 3G in Mexico had an impact on how individuals perceive violence in their local communities. By combining information on the location, type and reach of internet towers across Mexico with local household surveys and government administrative data on crime, I can map the gradual rollout of internet technologies such as 2G and 3G, and identify the impact of internet access on perceptions on violence relative to actual crime rates. To identify this effect, I follow the empirical literature in economics and rely on an instrumental variables strategy leveraging average lightning activity as a predictor for the location of internet towers.