Caroline Le Pennec-Çaldichoury


Department of Economics

University of California, Berkeley


I am a PhD candidate in Economics at UC Berkeley.

I conduct research in political economy, with an emphasis on electoral competition. My dissertation investigates the extent to which campaigns and elections matter for representation in western democracies. I am also interested in innovative empirical methods, in particular using text as data.

I will be on the job market this coming fall (2019-2020).

Working Papers:

  • Job market paper: Converging on words, not policies: Evidence from Campaign Manifestos (with Paul Vertier)- draft coming soon!

Abstract: In many contexts, politicians compete in local races under a national party ticket. How do individual candidates respond to local competition when they face strong national constraints and when their policy platform is determined at the party level? We exploit the natural variation in competition in two-round elections and estimate the within-candidate change in campaign strategy as a result of the change in electoral supply between rounds. We assemble a unique dataset of more than 30,000 manifestos circulated by candidates to the French legislative elections before each election round, between 1958 and 1993. Using computational text analysis, we scale manifestos on a left-to-right axis and show that candidates who make it to the runoff tend to moderate their discourse before the second round. We further show that discourse moderation reflects rhetorical convergence between politicians who move away from their party platform to send a more personalized and consensual message. Consistent with a model of competition with exogenous policy positions, we find evidence of strategic selection into discourse moderation: politicians who campaign on non-policy issues instead of their policy platform are electorally vulnerable - but possibly competent - and ready to exert more effort on non-policy representation if elected. Our paper implies that electoral competition at the individual level matters for representation, even when policy platforms are determined at the national level.


  • Vote choice formation and the minimal effects of TV debates: Evidence from 61 elections in 9 OECD countries (with Vincent Pons) - draft coming soon!

Abstract: We use repeated survey data from 250,000 voters and 61 elections in 9 OECD countries since 1952 to study the formation of vote choices and policy preferences in the electoral season and assess the contribution of TV debates to this process. We find that the share of voters who state a pre-election vote intention corresponding to their final vote choice increases by 12 percentage points in the two months preceding the election. Changes in individual vote choices likely result from changes in beliefs on competing candidates, and they generate aggregate shifts in predicted vote shares. Instead, policy preferences remain remarkably stable over time. We use an event study to estimate the impact of TV debates, campaigns' most salient events, and find that they affect significantly neither individual vote choice and preference formation nor aggregate vote shares. This suggests that information continuously received by voters exerts more influence on their behavior.

Work in Progress:

  • Money and Ideology: Evidence from French Legislative elections (with Julia CagĂ© and Elisa Mougin)