“…While some of its main comparative static results-higher audit probabilities and fines increase compliance-have by and large been confirmed (see e.g., [17,18]), several authors have argued that observed levels of compliance are too high to be explained only by aversion to the risk of being fined (e.g., [19][20][21]). 6 Partially in response to the debate about the merits and demerits of the Allingham-Sandmo model, partially as independent efforts to understand the determinants of tax compliance, experimental designs have tended to become more involved, introducing inter alia public good elements (e.g., [17,19,23,24]), heterogeneous and earned income [23,25,26], richer tax schedules [26][27][28] or endogenous audit rules [29,30]. Other studies haven taken up psycho-social factors like the role of persuasion [31], moral constraints [32,33] and emotions [33,34], and still others investigated design factors such as the role of framing (e.g., [19,25,35]) or income source [25,36].…”