Do the individual characteristics of politicians affect the performance of local governments? Do electoral rules and local regulations affect who runs for office and who gets elected? Why and how do politicians matter? Until recently, theoretical models of public policy disregard the role of individual politicians and explain political performance based on structural factors, the preferences of voters, and the incentives provided by institutions. An emerging field of inquiry is starting to examine if leaders matter for policies in a systematic way using state-ofthe- art methods which allow going beyond case studies, correlational studies, and anectodal evidence. The LEADERS project builds on this body of literature and contributes to it by identifying three main lacunas in previous research and addressing them.

The first shortcoming we identify is that previous research has concentrated on top leaders but has rarely examined the aggregate composition of governments. Politics is never performed by leaders alone, but by teams of politicians in different positions. Arguably, outcomes also depend on all members of governments, rather than just those at the top, and the joint dynamics of teams. Besides, previous research about the effects of leaders on performance has studied a small number of outcomes, especially GDP growth, to the neglect of other important outcomes. The project places a particular focus on the impact of individual characteristics on corruption.

The second shortcoming is the scarcity of well-identified research about political selection. Here we will exploit the discontinuities in salaries, competencies and electoral systems depending on the size of municipalities. We distinguish between self-selection factors such as the motivations of politicians to run for office, including salaries and the power of local governments, and institutional factors, in particular the use of open or closed party list.

The third shortcoming, and perhaps the most important, is that very little previous research has investigated the mechanisms through which the individual characteristics of political leaders affect their performance. Most research in political economy uses administrative data rather than survey data, which would allow examining the specific reasons, psychological and otherwise, why different types of leaders perform differently. We study how beliefs, honesty, preferences, leadership styles and motivations affect government performance and how they change due to holding office.

The project will study local politicians in Spain. First, we will use a novel dataset on the individual characteristics (age, gender, occupation, and education) of all local politicians elected in Spain between 1979 and 2007 made available by the Spanish Ministerio de Hacienda y Administraciones Públicas and apply Regression Discontinuity Design, a powerful tool that exploits the randomness of outcomes in very close elections. Second, we will collect original survey data in a two-wave panel study of politicians during their time in office (in 2017) and after it has finished (in 2019). We complement this data through text analysis of official documents. By addressing substantively important questions with cutting edge methods, the project will advance knowledge about the causes and consequences of political leadership and will help inform public debate about this topic.