This paper presents an estimation of credit quality transition matrices for commercial banks in Colombia, using a duration hazard function model, and following the methodology proposed by Gómez-González et al (2009). Using a test developed by Weißbach et al (2005), we test for the time-homogeneity of transition matrices estimated this way, after conditioning on firm-specific and macroeconomic variables. We found that 70% of the time we could not reject the null hypothesis of time homogeneity. We also found that obtaining matrices for different subsamples was not necessary, given the similarities of the survival function.
