Survival Analysis Modeling on Credit Risk With Latent Variable Indicator Models

Abstract

Survival analysis has several advantages, including the ability to handle incomplete data. In general, the explanatory variables used in survival analysis are manifest variables. In other fields such as social sciences, variables such as perception, attitudes, and psychology are often involved in statistical analysis. This research aims to model cox proportional hazard regression on data with latent variable types. Research was conducted in the banking sector using Likert scale questionnaire data. It will examine how the assessment of the 5C variable in the form of a latent variable relates to the time of delay in credit payments. Confirmatory Factor Analysis (CFA) was used to form a reflectively latent variable indicator model which was then formed into a cox proportional hazards regression model. The results show that all 5C variable assessments influence the speed of credit payments. The novelty in this research lies in the use of indicator analysis models to form latent variables before undergoing survival. The resulting interpretation and conclusions are expected to provide more in-depth information because they include the relationship between survival time and indicators.