Events

To what extent is risk aversion a cognitive bias? We present participants with hypothetical investment decisions defined over yearly income during retirement, framed seven different ways. We then present participants with inconsistencies and intransitivities in their own decision making and observe how preferences are updated (or not), interpreting updates as moving from "untutored" to "reasoned" preferences. We further test for persistence by bringing a subset of participants back in the lab several weeks later, repeating the investment decisions and updating procedure. Finally, we use maximum likelihood to estimate how risk aversion and decision errors differ by frames and across time, focusing on heterogeneity with respect to cognition and gender. We observe a significant decline in decision errors (inconsistencies and intransitivities) within and between frames. However, we observe little convergence in risk aversion among frames, with some frames systematically inducing higher risk aversion. Higher cognition participants have fewer decision errors and are more risk tolerant. While both low and high cognition participants reduce their decision errors, their measures of risk aversion fail to converge. This is consistent with risk aversion not simply being a cognitive bias.