Un modelo de panel de datos aplicado al efecto de variables micro y macroeconómicas en la cartera vencida: el caso de los bancos colombianos

Abstract

This paper examines the determinants of Non-Performing Loans (NPLs) in the Colombian banking sector, through the implementation of a model with long-term panel data. The study is done separately for four loan categories (mortgages, business loans, consumer loans and microcredit) and its motivation is the hypothesis that both the macroeconomic behavior and the specific variables of the banks have an effect on loan quality and that these effects varies depending on the loan category. The results allow us to conclude that the NPLs in the Colombian banking system are possible to explain mainly by macroeconomic variables such as the exchange rate and the real interest rate, and by bank-specific variables such as provisions ratio and solvency. In addition, it is evident that the effects vary depending on loan categories, where mortgages are the least responsive to changes both in micro and macroeconomic variables; while the business loans are most responsive to these variables.