CLUSTER ANALYSIS FOR LIMIT ALLOCATION POLICY TO MINIMIZE SHOPEE PAYLATER PAYMENT DELAY ON THE COVID-19 PANDEMIC SITUATION

• Main Author: Andi Michael Siahaan, Yonathan
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/79077
• Institution: Institut Teknologi Bandung

As the e-commerce sector grew, companies also expanded their business unit to synergize with their e- commerce platforms such as seller partnership, food delivery, investment gateway, and pay-later services. Paylater services are also one of the tech-enabled services in finance sector. Shopee is one of the market-leading e-commerce in Indonesia which also offer pay-later services to users, called Shopee Paylater (SPL). During the pandemic, as the late payment start to occur in SPL services, Shopee starts to reduce the maximum limit or remove it completely for all users in general, to maintain the liquidity and prevent non-performing loans. Though it is effective in its purpose, it seems like contrary efforts considering the increasing demands on such services while there is a major growth in revenues, user counts, and transactions. To address these issues, this research is designed to explore both quantitative methods to study the segregation of late payment users’ and qualitative methods to analyze cluster characteristics of different users’ profiles to design a limit allocation policy. Based on the result of this research, the three factors with the highest importance are income, monthly spending, and credit outstanding. The hierarchical cluster derived from the three factors are low-profile cluster with low income, low expense, and low credit outstanding with the mid-profile credit performance. The second cluster is the medium profile with medium income, medium expense, and medium credit outstanding with slightly higher expected credit performance. Lastly, the third cluster is the high-profile with high income, high expense, and high credit outstanding with the highest expected credit performance. The proposed policy is to increase the credit limit for users with age above 29, married, and located in Jakarta, retain the credit limit for users with age above 26, and cut-off all credit limit for users with age below 26. The new policy has a rate of accuracy at 38.82% which are above the full cut-off policy at 84.81% and below the full retain policy of 84.41%. Lastly, the new policy has an expected NPL rate of 10% which are above full cut-off policy at 0% and below the full retain policy of 15.19%.