Essays in Applied Economics: new empirical approaches to study individual behavior and improve policy targeting

This PhD thesis is composed of three, seemingly almost unrelated, chapters. The first chapter, titled “Twice Losers: How the shadow of cheating affects tax behavior and norms” relies on a lab experiment to study whether the way income is generated in a society can impact individuals’ willingness to pay taxes and judgements on the acceptability of tax evasion. I focus on whether the suspicion that some individuals in the society could have got their income by cheating at the expenses of others alters individuals’ behavior and acceptability ratings on tax evasion. The data collection for this project was made possible by a generous grant received by IFREE in 2018. The second chapter, titled “Machine learning in the service of policy targeting: The case of Public Credit Guarantees” originates from a joint work I developed with some colleagues at Bank of Italy, where I spent a few months as a research intern. This project relies on a combination of tools from Machine Learning and causal inference to propose an alternative targeting rule for Italy’s main public guarantee program, aimed to ease SMEs access to credit through publicly funded collaterals. The third and last chapter, titled “Social preferences and strategic incentives for cooperation in infinitely repeated Prisoner Dilemmas”, which I first started working on while visiting UCSB in 2019, bridges my interests for applied econometrics and experimental economics. This paper investigates the role of structural game parameters and of social preferences in shaping cooperation in infinitely repeated Prisoner Dilemmas: in the first part, I collect data from previous experiments to run a meta-analysis aimed to test, using simple supervised learning algorithms, the predictive power of structural game parameters. In the second part, I develop a novel experimental design to study the role of social preferences on cooperation in infinitely repeated Prisoner Dilemmas.