Studies

with Daniel Zizzo and Alexandros Karakostas - write-up stage

Motivated by observations in corporate finance where executives prefer to hire accountants with dark personality traits, we explore agents’ willingness to signal that they are antisocial and whether these signals can induce their recruitment by a principal in a lab setting. We use a two-stage game where, in stage one, the agent can dishonestly report the size of an endowment at a cost to a third party, and, in stage 2, the principal can recruit the agent to dishonestly report on their behalf. Through manipulation of the signalling environment, we examine cases where there exist separating equilibria for different types of agents based on strategic signalling of antisocial traits in stage one. In an environment where dishonesty is harmful to a third party, we find that principals are more likely to recruit dishonest agents and, to take advantage of this, agents signal that they are dishonest to increase their likelihood of being recruited. 

with Carlos Oyarzun - submitted

Norm nudges are a powerful policy tool to change behavior, with applications ranging from reducing electricity consumption to increasing charitable giving to promoting female labour force participation. On the other hand, there is a growing literature illustrating failed applications. We introduce a theory of social norms based on agents' beliefs about others' private preferences for compliance with the desired action, coupled with information design  techniques, to analyze when fully descriptive social norm nudges --revealing compliance rates--  are effective. We propose Bayesian Nudges as a method for improving the effectiveness of fully descriptive social norm nudges and derive the optimal information structure.  Full revelation is never optimal. Instead, the optimal information structure involves partitioning the set of states of the world in three events: (i) full compliance,  (ii) partial compliance, and (iii) no compliance. These events occur when their respective partition is revealed based on whether the state, which is privately determined after partition design, belongs to that partition. Our welfare analysis compares the optimal Bayesian nudges for a principal who maximizes compliance versus for a principal who maximizes social welfare.

with Philip Grossman and Alexandros Karakostas - write-up stage

This study explores the concept of ‘moral balancing’, which describes situations where subjects take into account past moral behaviour when making moral decisions. Through a lab setting, we examine the role of subjects’ prior luckiness or antisocial behaviour on subsequent prosocial and antisocial behaviour. Our experiment consists of two stages. In the first stage, subjects are provided with an endowment that, depending on the treatment, is determined by either luck or their reported outcome from two privately rolled six-sided dice. In the second stage, subjects select how to allocate an endowment between themselves and the charity Feed the Children, which is framed either prosocially or antisocially.

with Loukas Balafoutas, Jeremy Celse, and Alexandros Karakostas - submitted (JBEE)

The credibility of the social sciences is facing a significant crisis, marked by a lack of replicability, challenges in reproducing published results, and instances of scientific misconduct. This paper argues that these issues are deeply rooted in the incentive structures of academic publishing, which prioritize novel, statistically significant findings over methodological rigor and the pursuit of knowledge itself. This emphasis has fostered a "publish or perish" culture, encouraging questionable research practices (QRPs) that undermine the reliability and validity of research findings. We examine the theoretical underpinnings of scientific misconduct, exploring the individual and collective incentives that drive these practices. We review empirical evidence on the effectiveness of existing solutions like replications and preregistrations, highlighting their limitations in addressing the systemic issues. We propose that registered reports and megastudies offer a promising path towards restoring trust in science. By decoupling publication from results and emphasizing the quality of research design, these approaches realign incentives, fostering a research culture that values transparency, collaboration, and the pursuit of truth. While challenges remain, early evidence suggests that registered reports and megastudies can significantly enhance the credibility and societal impact of social science research.

with Daniel Zizzo and Alexandros Karakostas - Honours/Masters Thesis

We aimed to examine whether group identity exacerbates conflict between groups and reduces it within a group under an experimental setting. This involved the use of two experiment designs and 160 random participants. Both experiment designs consisted of two parts: (1) group identity inducement, and (2) the vendetta game. For Part 1 participants were assigned to a group based on their preferences over a set of paintings by Klee and Kandinsky and then underwent one of two group priming methods. In Experiment 1, participants had their senese of group identity primed via online communication. In Experiment 2, participants were primed via ‘common fate’, wherein participants of the same group identity shared the outcome of a lottery. Part 2 involved participants being matched with either a heterogeneous, homogeneous, or unknown counter-participant and answering whether they would Take or Don’t Take in a stylized mini-vendetta game (Bolle, et al., 2014). Overall, for both Experiment 1 and 2, we find no statistically significant behavioural differences between participants matched with a co-participant of the same group identity and those matched with a co-participants of a difference group identity. While we find no support for prosocial preferences within groups nor antisocial preferences between groups, the high instances of participants taking in the vendetta game suggest that social preferences are a motivating factor for aggressive behaviour.