A Study on Building the Prediction Model and a Behavior Scorecard for Investors’ Risk Tolerance:A Case Study in Bank A

• Main Authors: Yun-Tung Hsieh, 謝昀東
• Other Authors: Te-Hsin Liang
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/95845474439128432747

碩士 === 輔仁大學 === 統計資訊學系應用統計碩士班 === 101 === After the financial crisis, the banks not only face the rapid-changing in global financial environment but also the competition from other financial institutions. Know Your Customer is the most important factor for banks. If the banks can evaluate the Investors’ Risk Tolerance seriously, then financial consultants can design the combination of financial products according to Investors’ risk tolerance. It can not only promote the professional image of banks and customer loyalty, but also improve the marketing promotion on target groups in future. Data were collected from Bank A by using the survey of “The cluster of wealth management clients’ investment behaviors” from July 2012 to September 2012. Important variables were selected by two steps. The first step selected the significant variables by using Chi-Square Independent Test and reduced the multicollinearity in the data set by combining independent variables. The second step used random sampling to select 80% of the entire training dataset. Resampling was done 50 times to construct 50 multinomial logistic models. The final important variables would be select if the number of voting is more than or equal to 25 times from the 50 models and Cramer’s value is more than or equal to 0.1. Finally, the behavior scorecard will be build base on multinomial logistic models. The feature of Behavior scorecard in this study can provide the financial consultants the suggestions of making an appropriate diversity and assets allocation decision according to investors’ risk total score in different score interval.