Leveraging machine learning for predictive insights in robo-advisory adoption: a marketing analytics approach

The rise of machine learning has gradually reshaped the service industry. In recent decades, the emergence of many new financial technologies (Fintech), such as robo-advisors, has garnered significant attention from both academic researchers and industry practitioners. Motivated by the rapid growth of assets under management by robo-advisors and the increasing popularity of machine learning in different domains, the present study provides an example of using four machine learning techniques: Logistic regression, classification tree, random forest, and boosting to help identify potential users of robo-advisors. Data were extracted from the National Financial Capability Study 2015 (NFCS2015). Different model evaluation measurements have shown that the logistic regression model offers more prediction accuracy with a selected probability threshold compared to other models studied in this paper. Moreover, while both logistic regression and tree-based models reveal that age is a significant predictor for robo-advisory adoption, we believe that this result needs to be interpreted with care. Interestingly, we find that the influence of financial knowledge on robo-advisory adoption is only significant within people between 18 and 34, indicating a heterogeneous effect of financial knowledge based on age. Our finding underscores the importance of considering the varying effect of financial literacy when promoting robo-advisory adoption in retail markets.