Combining vignette surveys with agent-based modeling: Insights on fraud dynamics with empirically calibrated norm sensitivities

Fraud is a widespread problem in many organizations, resulting in substantial economic losses and other detrimental outcomes. Given that individual and social factors influence the prevalence and dynamics of fraud, agent-based modeling (ABM) has already been utilized to investigate the empirically difficult-to-capture phenomenon of fraud. However, providing an empirically sound model of fraud dynamics in organizations remains challenging. This paper strives to contribute to this aim. We first conduct a vignette survey informed by Bicchieri’s categorization of societal norms, assessing individual norm sensitivities across two scenarios. Then, we use the survey data to calibrate agents’ decision-making in our agent-based fraud dynamics model. We find that only slight differences in norm sensitivity distributions obtained from the two vignettes result in substantially different fraud dynamics: While fraud is contained in one scenario, it spreads and dominates the organization in the other. Overall, our study demonstrates how vignette studies and ABM can be combined to strengthen the empirical basis of models of human behavior.

Subjects

ABM

fraud

social norm theory

norm sensitivity

vignette survey

threshold model

DDC Class

330: Economics

Options

Combining vignette surveys with agent-based modeling: Insights on fraud dynamics with empirically calibrated norm sensitivities