Misclassifications in Financial Risk Tolerance

Abstract

This paper analyzes the empirical risk tolerance of individuals. Rare empirical evidence shows the role of personal behavior in both propensity toward financial risk and risk aversion. By using a test which mimics the financial decision process in a laboratory setting for 445 individuals, we obtained an ex-post experimental measure for individual risk tolerance. Predictive classification models allow us to evaluate the forecasting accuracy of two alternative risk tolerance assessments: a psychometrically derived questionnaire and a psycho- physiological experiment. Our findings show that misclassifications resulting from the questionnaire are massive: individuals asked to self-assess their risk tolerance reveal a high probability of failing their judgment, i.e., they behave as if they were risk takers, while defining themselves as risk-averse (and vice versa). Conversely, when considering somatic activation, misclassifications are considerably lower. Emotions are confirmed to drive the financial risk- taking process, enhancing the accuracy of the individual risk tolerance forecasting activity.

Keywords: decision-making, individual behavior, classification methods, risk tolerance questionnaire, Iowa gambling task, skin conductance response

JEL Classification: D81, C91, C38