Econometric Analysis of Preferences under Uncertainty

This paper investigate question on “how individual investors are making decisions” and research personal factors which define individual investors’ type – financial or behavioral decision-makers. Three models were applied on an online bank customers’ trading data – CAPM, Fundamental Analysis (FA), and Reinforcement Learning (RL). Private investment decisions of clients of one of the largest Russian private banks (analysed anonymously) served as empirical basis for this study applied to four major assets: Sberbank, Yandex, VTB and Gazprom shares. Extend to which the models describe decisions made by these individuals was estimated and tested in three ways: generally, instrument-wise, and person-wise. The most important part is person-wise examination of the models. We have studied relationship between investor’s personal characteristics and how good is her decisions modeled by each approach. Results show that investors’ performance largely depend on “investment maturity” variables, such as age, income, region, and education, while. Customers with “mature investor” characteristics shows higher propensity to act according to FA than “less mature” customers. Results of general analysis are inconclusive, while instrument-wise comparison implemented on SBER, YNDX, VTBR, and GAZP securities shows that RL has the highest predictive power.