Université de Lausanne
Department of Economics
Quartier UNIL- Chamberonne
Office Internef 501,1
1015 Lausanne - Switzerland
Phone: +41 021 6923 3469
You can find my CV here: Link.
Abstract: What is the effect of Twitter on political participation? I address this question by studying how the spread of this social network affected voting behavior and donations to politicians during the last three US presidential elections. First, I develop a novel measure of Twitter penetration by using location data collected from users. To address endogeneity in the diffusion of Twitter across regions, I exploit variation in the popularity of sport teams that signed new players with Twitter accounts, making therefore the social network more interesting for their fans. Instrumental variables estimates do not show significant effects of Twitter on average participation, intended both as turnout and donations to politicians. On the other hand I find a differential effect across parties, with the Democratic Party being penalized in terms of votes and the Republican Party receiving more donations. I provide two pieces of evidence on mechanisms. First, I show that Twitter reduces voters' information about politics and increases political polarization. Second, by downloading and categorizing tweets written by users I show that the majority of users write about sport or entertainment and ignore politics for most part of the year. Peaks in interest happen only during presidential debates, when both the quantity of partisan tweets and the average sentiment favor the Republican Party.
(with Eleonora Patacchini and Paolo Pin)
Abstract: Using an app for smartphones we run an experiment among high school students to study the pattern of aggregation of sparsely distributed information when competing agents are arranged in small networks and can share only non-verifiable pieces of information. Our first finding is that the level of cooperation is high, especially among students that belong to the same class. Nevertheless the level of centralization of the network significantly affects the final results, with the most central node benefiting in terms of payoffs. By adding a second node with a high centrality we see that the results change significantly, with more signals traveling through the links. We then turn to a parsimonious level-k approach to characterize players according to their behavior in the game. When estimating the model we see that data are consistent with a vast majority of players acting cooperatively, with the difference across networks driven mainly by a small share of strategic players and their position in the network.
App Lab is an easy to use app to run experiments on smartphones. It is available for both Android and iOS and it allows researchers to distribute experiments directly in subjects’ pockets. It is compatible with easy to use tools such as oTree and Qualtrics, and can also be used for notifications and payments.