Purpose: The purpose of this paper is to examine the link between banking crises and the subjective well-being of individuals. In addition, the authors examine the transmission of crises from the banking sector to well-being and show that negative financial shocks have significant adverse effects. Design/methodology/approach: The authors employ agent-based modeling to test for the direct and indirect welfare effects of banking crises. The model includes a support vector machine (SVM) optimized subjective well-being function. The existing literature suggests that this is influenced by both the negative psychological effects of recessions and the adverse economic effects of income loss and increased unemployment. Findings: The authors show that the different choices of policy response to a banking crisis carry different opportunity costs in terms of welfare and that societal preferences should be taken into account. The authors demonstrate that these effects influence different population classes in an asymmetric manner. Finally, the results demonstrate that the welfare loss of a bank failure is much higher than the cost of a bailout. Practical implications: The authors are able to propose to the authorities the best policy mix in order to handle banking crises in the most adequate manner, according to society's preferences between financial stability and public goods. Social implications: The findings extend the existing literature on subjective well-being, by quantifying the welfare cost of banking crises and showing that authorities should reconsider bank bailouts as a policy solution to bank distress. Originality/value: The originality of this article lies in the use of an agent-based model to model the relationship between societal well-being and financial stability. Also, the authors extend existing agent-based methodologies to include machine learning optimization techniques.