WORKING PAPERS
My research agenda currently focuses on two main topics: the political consequences of technological change and the experiences and preferences of political elites.
Technological change
Neither Left-Behind nor Superstar: Ordinary Winners of Digitalization at the Ballot Box (R&R at the Journal of Politics)
With Thomas Kurer and Nikolas Schöll
The nascent literature on the political consequences of technological change studies either left-behind voters or successful technology entrepreneurs ("superstars"). However, it neglects the large share of skilled workers who benefit from limited but steady economic improvements in the knowledge economy. This paper studies how workplace digitalization affects political preferences among the entire active labor force by combining individual level panel data from the United Kingdom with industry-level data on ICT capital stocks between 1997-2017. We first demonstrate that digitalization was economically beneficial for workers with middle and high levels of education. We then show that growth in digitalization increased support for the Conservative Party, the incumbent party, and voter turnout among beneficiaries of economic change. Our results hold in an instrumental variable analysis and multiple robustness checks. While digitalization undoubtedly produces losers (along with some superstars), ordinary winners of digitalization are an important stabilizing force content with the political status quo.
Previous version: "Not so Disruptive after All: How Workplace Digitalization Affects Political Preferences", Working Paper 1063, Barcelona Graduate School of Economics.
Technological change, trade, and preferences for redistribution: Experimental evidence
With Alex Kuo, Pepe Fernández-Albertos and Dulce Manzano
A recurring concern in the media and among economists is that the adoption of new digital technologies in the workplace can displace workers. This paper asks if the general public is concerned about technological unemployment and which policies do citizens demand to deal with workers who become unemployed due to technological change. To study these questions, we collect novel data from a large survey in Spain with embedded survey experiments. All the experiments compare technological change with change driven by international trade. Our findings suggest that citizens believe that the impact of the introduction of digital technologies in the workplace is positive for society and for their own jobs on average. However, preoccupation varies substantially with vulnerability: workers more exposed to change are also more likely to view the impact of technology as negative. Turning to policy demands in case of technological unemployment, we find no evidence that citizens are more likely to support an unconditional cash transfer for workers affected by technological shocks than by workers affected by trade shocks. Finally, we find that economic shocks with clear outgroup beneficiaries are more likely to produce anger and lead to more demand for the government to step in and stop the change. We speculate that there are several reasons why Luddism is not widespread: the public sees technological change at the workplace as a benign force, most workers are not concerned about being personally affected, and automation does not produce clear outgroup beneficiaries to mobilize against.
Draft available on request
Political elites
Gendered language in Political Communication: A Machine Learning Approach (R&R at the European Journal of Political Research)
With Javier Beltrán, Alba Huidobro, Lluís Padró and Enrique Romero
How do male and female politicians differ when they communicate directly with the public on social media? And do citizens address them differently? A large literature demonstrates that male and female politicians are portrayed differently in traditional media. We ask if gender differences persist when politicians can communicate directly with the public online. Using over 328000 tweets written by and addressed to members of the Spanish national and regional Parliaments, we first show that male and female politicians on average use very similar language, but some words are used to different extents. We apply Lasso logistic regression models to identify the linguistic features that most differentiate language used by or addressed to females or males politicians. Words related to politics, sports, ideology, and infrastructure are more likely to be used by male politicians, while female politicians are more likely to talk about gender and social affairs. The choice of emoticons also varies starkly across genders. Finally, we find that female and male politicians receive different insults and that mentions to physical aspect and infantilizing words are more predictive of text addressed to females. To conclude, we discuss the potential of machine learning to study gender differences in political communication.
"Educated Politicians: Effects on Performance and Fiscal Policy"
With Marta Curto
Highly educated citizens are dramatically over-represented among politicians. Is this bias desirable, troubling or irrelevant? Recent studies argue that highly educated politicians perform better in office, but others find no effects. We advance a third possibility which is that education affects the preferences and beliefs of politicians, and leads them to pursue different goals and fiscal policies. Our empirical analysis is based on a novel dataset with information about the education, age and gender of elected local politicians in Spain and detailed economic and fiscal data collected between 2003 and 2011. Applying a Regression Discontinuity design, we find that when parties with more educated politicians win the election, municipalities have higher unemployment rates and do not perform better in other respects. Further analyses reveal that educated politicians are more fiscally conservative, spend less in capital investment, and prioritize different spending areas. Our results are consistent with the interpretation that more educated politicians are more fiscally conservative rather than with the claim that education is a proxy of quality. To conclude, we discuss how the elitism in the educational composition of governments can undermine political representation.
Draft available on request
Who Lies in Politics?: Evidence from a Survey of Spanish Mayors
With Katharina Janezic
Honesty is one of the most valued traits in politicians, yet because most lies remain hidden from public view it is difficult to study if some politicians lie more than others. This paper examines which individual characteristics are correlated with honesty and if higher salaries attract more or less honest politicians. In order to measure lying, we included a novel incentivized non-monetary lying game in a survey of 920 Spanish mayors conducted in Autumn 2018 (40% response rate). Mayors were first asked how interested they were in obtaining a detailed report about the survey results and at the end of the survey they had to flip a coin to decide whether they would be sent the report. We use the fact that the probability of heads is known to estimate the proportion of mayors who lied to obtain the report. The findings suggest that a large proportion of mayors lied. Women were not less likely to lie, but educated politicians lied less. To study the impact of salaries, we exploit the fact that arbitrary population thresholds determine the maximum salary that can be paid to mayors using a regression discontinuity design. Our results suggest that offering higher salaries can help attract honest citizens to office.
Draft available on request
Citizens' Feedback and Attention to Gender Issues on Twitter: A Dynamic Reinforcement Learning Model
With Nikolas Schöll and Gaël Le Mens
Previous research states that female politicians focus more on gender issues because they have different motivations and because the public treats them differently. We propose and test a feedback-based mechanism that can explain differences in attention to gender issues based on standard reinforcement models of learning in behavioral psychology. We contend that politicians talk more about issues on which they obtain more positive feedback and that citizens send more positive feedback to female representatives for talking about gender than to men. This widens initial differences in issue attention over time. We test our expectations using a large dataset of social media data covering the universe of tweets written by national and regional MPs between 2016 and 2019 in Spain (n=1.3 million). Our empirical approach combines machine learning methods with econometric models with politician and day fixed effects. We show that female politicians receive relatively more retweets and likes for writing tweets on gender issues, and that politicians respond to positive feedback by tweeting more on the respective issue. This mechanism can explain some of the observed difference in attention to gender issues between male and female politicians. We speculate that social media can exacerbate existing differences in the agendas of male and female politicians.
Draft available on request
OTHER WORK IN PROGRESS
Family structure and women political representation: From historical legacies to rapid change (with Ana Tur-Prats and Dídac Queralt)
Initial results
Unemployment Experiences of Political Elites and their Social Policy Preferences (with Alba Huidobro)
Initial results
The political legacies of violence: Evidence from the Spanish Civil War (with Laia Balcells, Elena Costas, Catherine de Vries and Hector Solaz)
Presentation available
ONGOING FUNDED RESEARCH PROJECTS
Local politicians: Selection and performance (as Principal Investigator)
Team: Marta Curto, Elena Costas, Alba Huidobro, Yeimy Ospina, Ana Tur-Prats, and Didac Queralt
Research grant (CSO2016-79569-P) started in January 2017
Digital technology, ideological polarization and intolerance
With Ruben Enikolopov (PI), Maria Petrova and Ruben Durante
Research grant (BBVA Foundation) started in January 2018
My research agenda currently focuses on two main topics: the political consequences of technological change and the experiences and preferences of political elites.
Technological change
Neither Left-Behind nor Superstar: Ordinary Winners of Digitalization at the Ballot Box (R&R at the Journal of Politics)
With Thomas Kurer and Nikolas Schöll
The nascent literature on the political consequences of technological change studies either left-behind voters or successful technology entrepreneurs ("superstars"). However, it neglects the large share of skilled workers who benefit from limited but steady economic improvements in the knowledge economy. This paper studies how workplace digitalization affects political preferences among the entire active labor force by combining individual level panel data from the United Kingdom with industry-level data on ICT capital stocks between 1997-2017. We first demonstrate that digitalization was economically beneficial for workers with middle and high levels of education. We then show that growth in digitalization increased support for the Conservative Party, the incumbent party, and voter turnout among beneficiaries of economic change. Our results hold in an instrumental variable analysis and multiple robustness checks. While digitalization undoubtedly produces losers (along with some superstars), ordinary winners of digitalization are an important stabilizing force content with the political status quo.
Previous version: "Not so Disruptive after All: How Workplace Digitalization Affects Political Preferences", Working Paper 1063, Barcelona Graduate School of Economics.
Technological change, trade, and preferences for redistribution: Experimental evidence
With Alex Kuo, Pepe Fernández-Albertos and Dulce Manzano
A recurring concern in the media and among economists is that the adoption of new digital technologies in the workplace can displace workers. This paper asks if the general public is concerned about technological unemployment and which policies do citizens demand to deal with workers who become unemployed due to technological change. To study these questions, we collect novel data from a large survey in Spain with embedded survey experiments. All the experiments compare technological change with change driven by international trade. Our findings suggest that citizens believe that the impact of the introduction of digital technologies in the workplace is positive for society and for their own jobs on average. However, preoccupation varies substantially with vulnerability: workers more exposed to change are also more likely to view the impact of technology as negative. Turning to policy demands in case of technological unemployment, we find no evidence that citizens are more likely to support an unconditional cash transfer for workers affected by technological shocks than by workers affected by trade shocks. Finally, we find that economic shocks with clear outgroup beneficiaries are more likely to produce anger and lead to more demand for the government to step in and stop the change. We speculate that there are several reasons why Luddism is not widespread: the public sees technological change at the workplace as a benign force, most workers are not concerned about being personally affected, and automation does not produce clear outgroup beneficiaries to mobilize against.
Draft available on request
Political elites
Gendered language in Political Communication: A Machine Learning Approach (R&R at the European Journal of Political Research)
With Javier Beltrán, Alba Huidobro, Lluís Padró and Enrique Romero
How do male and female politicians differ when they communicate directly with the public on social media? And do citizens address them differently? A large literature demonstrates that male and female politicians are portrayed differently in traditional media. We ask if gender differences persist when politicians can communicate directly with the public online. Using over 328000 tweets written by and addressed to members of the Spanish national and regional Parliaments, we first show that male and female politicians on average use very similar language, but some words are used to different extents. We apply Lasso logistic regression models to identify the linguistic features that most differentiate language used by or addressed to females or males politicians. Words related to politics, sports, ideology, and infrastructure are more likely to be used by male politicians, while female politicians are more likely to talk about gender and social affairs. The choice of emoticons also varies starkly across genders. Finally, we find that female and male politicians receive different insults and that mentions to physical aspect and infantilizing words are more predictive of text addressed to females. To conclude, we discuss the potential of machine learning to study gender differences in political communication.
"Educated Politicians: Effects on Performance and Fiscal Policy"
With Marta Curto
Highly educated citizens are dramatically over-represented among politicians. Is this bias desirable, troubling or irrelevant? Recent studies argue that highly educated politicians perform better in office, but others find no effects. We advance a third possibility which is that education affects the preferences and beliefs of politicians, and leads them to pursue different goals and fiscal policies. Our empirical analysis is based on a novel dataset with information about the education, age and gender of elected local politicians in Spain and detailed economic and fiscal data collected between 2003 and 2011. Applying a Regression Discontinuity design, we find that when parties with more educated politicians win the election, municipalities have higher unemployment rates and do not perform better in other respects. Further analyses reveal that educated politicians are more fiscally conservative, spend less in capital investment, and prioritize different spending areas. Our results are consistent with the interpretation that more educated politicians are more fiscally conservative rather than with the claim that education is a proxy of quality. To conclude, we discuss how the elitism in the educational composition of governments can undermine political representation.
Draft available on request
Who Lies in Politics?: Evidence from a Survey of Spanish Mayors
With Katharina Janezic
Honesty is one of the most valued traits in politicians, yet because most lies remain hidden from public view it is difficult to study if some politicians lie more than others. This paper examines which individual characteristics are correlated with honesty and if higher salaries attract more or less honest politicians. In order to measure lying, we included a novel incentivized non-monetary lying game in a survey of 920 Spanish mayors conducted in Autumn 2018 (40% response rate). Mayors were first asked how interested they were in obtaining a detailed report about the survey results and at the end of the survey they had to flip a coin to decide whether they would be sent the report. We use the fact that the probability of heads is known to estimate the proportion of mayors who lied to obtain the report. The findings suggest that a large proportion of mayors lied. Women were not less likely to lie, but educated politicians lied less. To study the impact of salaries, we exploit the fact that arbitrary population thresholds determine the maximum salary that can be paid to mayors using a regression discontinuity design. Our results suggest that offering higher salaries can help attract honest citizens to office.
Draft available on request
Citizens' Feedback and Attention to Gender Issues on Twitter: A Dynamic Reinforcement Learning Model
With Nikolas Schöll and Gaël Le Mens
Previous research states that female politicians focus more on gender issues because they have different motivations and because the public treats them differently. We propose and test a feedback-based mechanism that can explain differences in attention to gender issues based on standard reinforcement models of learning in behavioral psychology. We contend that politicians talk more about issues on which they obtain more positive feedback and that citizens send more positive feedback to female representatives for talking about gender than to men. This widens initial differences in issue attention over time. We test our expectations using a large dataset of social media data covering the universe of tweets written by national and regional MPs between 2016 and 2019 in Spain (n=1.3 million). Our empirical approach combines machine learning methods with econometric models with politician and day fixed effects. We show that female politicians receive relatively more retweets and likes for writing tweets on gender issues, and that politicians respond to positive feedback by tweeting more on the respective issue. This mechanism can explain some of the observed difference in attention to gender issues between male and female politicians. We speculate that social media can exacerbate existing differences in the agendas of male and female politicians.
Draft available on request
OTHER WORK IN PROGRESS
Family structure and women political representation: From historical legacies to rapid change (with Ana Tur-Prats and Dídac Queralt)
Initial results
Unemployment Experiences of Political Elites and their Social Policy Preferences (with Alba Huidobro)
Initial results
The political legacies of violence: Evidence from the Spanish Civil War (with Laia Balcells, Elena Costas, Catherine de Vries and Hector Solaz)
Presentation available
ONGOING FUNDED RESEARCH PROJECTS
Local politicians: Selection and performance (as Principal Investigator)
Team: Marta Curto, Elena Costas, Alba Huidobro, Yeimy Ospina, Ana Tur-Prats, and Didac Queralt
Research grant (CSO2016-79569-P) started in January 2017
Digital technology, ideological polarization and intolerance
With Ruben Enikolopov (PI), Maria Petrova and Ruben Durante
Research grant (BBVA Foundation) started in January 2018
