Multinomial logistic regression in the detection of socioeconomic factors associated with the decision to indebted by natural persons in Colombia.

Abstract

Personal indebtedness is influenced by multiple socioeconomic, financial, and behavioral factors whose identification is essential for decision-making in financial education and risk management. This study aimed to determine the factors associated with individuals’ decisions to incur debt through the application of a multinomial logistic regression model. A quantitative, explanatory approach was adopted, using data collected through a structured survey administered to a sample of economically active adults. The variables analyzed included sociodemographic characteristics, income level, consumption habits, credit perception, and financial behavior. The results show that income level, employment stability, financial education, and perceived necessity significantly affect the probability of choosing different levels of indebtedness, with statistically significant differences among the analyzed groups. The estimated model demonstrated adequate explanatory capacity and consistent classification of indebtedness profiles. It is concluded that multinomial logistic regression is a robust tool for identifying determinants of indebtedness, providing useful evidence for the design of financial education strategies and policies aimed at preventing over-indebtedness.