Cognitive abilities and risk-taking: Errors, not preferences

Abstract

There is an intense debate whether risk-taking behavior is partially driven by cognitive abilities. The critical issue is whether choices arising from subjects with lower cognitive abilities are more likely driven by errors or lack of understanding than pure preferences for risk. The latter implies that the often-argued link between risk preferences and cognitive abilities (a common finding is that abilities relate negatively to risk aversion and positively to loss aversion) might be a spurious correlation. This experiment reports evidence from a sample of 556 participants who made choices in two risk-related tasks and completed three cognitive tasks, all with real monetary incentives: number-additions (including incentive-compatible expected number of correct additions), the Cognitive Reflection Test (to measure analytical/reflective thinking) and the Remote Associates Test (for convergent thinking). Results are unambiguous: none of our cognition measures plays any systematic role on risky decision making. Using structural equation modeling and factor analysis, we show that cognitive abilities are negatively associated with noisy, inconsistent choices and this effect may make higher ability individuals appear to be less risk averse and more loss averse. Yet we show that errors are more likely to appear when the two payoffs in a given decision exhibit similar probability. Therefore, our results suggest that failing to account for noisy decision making might have led to erroneously inferring a correlation between cognitive abilities and risk preferences in previous studies.