Using Tree-Based Algorithm to Predict Informal Workers' Willingness to Pay National Health Insurance after Tele-Collection

This study aims to determine the main factors influencing the informal worker sector's willingness to pay (WTP) the National Health Insurance System (NHIS) payments following tele-collection. The difficulty for the NHIS program's viability in terms of contributions is the low WTP for payments from informal participants. Social Security Agency for Health (SSAH/BPJS Kesehatan) has made numerous initiatives to collect payments from informal participants, but their effectiveness remains low. Therefore, the author analyzed the tele-collection data set from the SSAH using socio-demographic data, product or service factors, and billing time as independent factors to predict WTP. In addition, this study used tree-based algorithms for modelling and using accuracy values and area under the ROC curve (AUC) to evaluate predictive results. Findings/Results: (a) The determinant factors of WTP for informal participants after tele-collection in this study are: access to facility health services, billing time, number of months in arrears, and the age of the family head, which collectively yielded a predicted probability of 82.4 per cent; (b) the Light Gradient Boosting Machine (LGBM) algorithm performed more accurately on the collected data compared to other tree-based algorithms. The findings of this study can be used practically by SSAH to improve the probability of success in their tele-collection program, which was previously less effective.