SOCIAL MEDIA DATA TO DETERMINE LOAN DEFAULT PREDICTING METHOD IN AN ISLAMIC ONLINE P2P LENDING

Abstract

Financial technology is growing rapidly in Indonesia. One of main types of financial technologies are online peer-to-peer (P2P) lending platforms. Islamic online P2P lending is also emerging. However, credit risk is still a major concern for this platform. To address this issue, social media assessments have been developed. Therefore, in this paper, the authors have aimed to identify social media variables that could be used as default probability predictors and determine predictability level by adding social media data to the model. Six independent variables consisting of social media data and seven control variables from historical payment and demographic data were used to construct a credit scorecard and logistics. The results identified five variables that could be considered and used as default probability predictors, which are posting frequency at midnight, followers, following, employment, and tenor. Interestingly, the number of religious accounts followed on Instagram is not a significant variable. Furthermore, the model with selected variables through the combination of demographic, historical payment, and social media data could increase the predictability level by 6.6%.