Bridging the Gap - A Model Comparison in Risk Behavior

This paper explores the challenges of measuring risk preferences and examines the disparity within behavioral measures. We conducted an online experiment where participants repeatedly chose between a risky and a safe stock in presence of multiple reference points and a deadline. We find evidence that convergence of behavioral risk preference measures depends on contextual similarity. Additionally, we show that risk preferences (measured by variance) near a reference point are variable in the short run. Nonetheless, Mean-Variance models outperform standard Expected Utility Theory models, and Behavioral models combining both excel. Our results indicate a shift in individual's decision making process, investing more resources (i.e. time) for the final decision before the deadline. The results emphasize the importance of context in risk preference measures and the limitations of a constant variance preference assumptions for modeling. We suggest moving beyond equating risk preference with variance preference for a better alignment with Expected Utility Theory, facilitating clearer communication of risk across fields and stakeholders. JEL Classification: D81, C91, G11