Any 2x2 game can be decomposed into three payoff components: strategic, behavioral, and kernel. The strategic component sufficiently determines the prediction for a large class of models focused on bounded rationality, including the most commonly used specifications for the Quantal Response Equilibrium and Cognitive Hierarchy. These models, along with others, exhibit a mathematical invariance to changes in a game's behavioral and kernel components, and this paper's primary hypothesis is that humans do not exhibit this invariance. An experiment consisting of 2x2 games shows that subjects systematically respond to changes in non-strategic information: in particular, the behavioral component.

We examine how professional traders behave in two financial market experiments. First, participants trade an asset over multiple periods after receiving private information about its value. Second, participants play the Guessing Game. Finally, subjects play our novel, individual-level version of the Guessing Game and we collect data on cognitive abilities, risk preference, and confidence. We find three differences between traders and students. Traders do not produce price bubbles observed in previous studies. Traders aggregate private information better. Traders show higher levels of strategic sophistication in the Guessing Game. These results are not driven by different individual characteristics (e.g., cognitive abilities). 

25 minute presentation 

Axios article

We study whether the COVID-19 pandemic has impacted risk preferences, comparing the results of experiments conducted before and during the outbreak. In each experiment, we elicit risk preferences from two sample groups: professional traders and students. We find that, on average, risk preferences have remained constant for both participant pools. Our results suggest that the increases in risk premia observed during the pandemic are not due to changes in risk appetite, rather, they are solely due to a change in market participants’ beliefs. Our findings support the traditional view that risk preferences are not affected by economic or social circumstances

We study the impact of COVID-19 on noncognitive skills by comparing experimental results gathered before and during the outbreak. Using a sample of professional traders, we find a sharp decrease in Agreeableness and Locus of Control and a moderate decrease in Grit, whereas Trust, Conscientiousness and Self-Monitoring are unchanged. We contrast these results with those from a sample of undergraduates whose noncognitive skills remain constant (with the exception of Conscientiousness). Our findings provide evidence against the stability of noncognitive skills, particularly in professional traders.