Between Attention and Portfolio Adjustment: Insights from Machine Learning-based Risk Preference Assessment

Financial firms recommend products to customers, intending to gain their attention and change their portfolios. Based on behavioral decision-making theory, we argue attention’s effect on portfolio adjustment is through the risk deviation between portfolio risk and their risk preference. Thus, to fully understand the adjustment process, it is necessary to assess customers’ risk preferences. In this study, we use machine learning methods to measure customers’ risk preferences. Then, we build a dynamic adjustment model and find that attention’s impact on portfolio adjustment speed is stronger when customers’ risk preference is higher than portfolio risk (which needs an upward adjustment) and when customers’ risk preference is within historical portfolio risk experience. We conducted a field experiment and found that directing customers’ attention to products addressing the risk deviation would lead to more portfolio adjustment activities. Our study illustrates the role of machine learning in enhancing our understanding of financial decision-making.