Determinants of the Non-Performing Loans Portfolio in Argentine Banks

The purpose of this paper is to analyze, for the first time, the characteristics of non-performing loans in the Argentine banking system in the period 1999-2005, focusing on loans to companies. These are the main findings: (1) the percentage of non-performing loans is closely related to the business cycle, even though such correlation comes to a halt during the peak of the crisis in 2002-2003 due to an easing of the regulation and its effective compliance; (2) even though the percentage of non-performing debt has tended to return to its pre-crisis levels, the percentage of non-performing debtors is still high. In 2005, approximately one out of three debtors was a delinquent debtor. This proportion was one out of five in 1999 and reached a peak of one in two in 2003. In turn, the percentage increase between 1999 and 2005 is mainly explained by a rise in the extreme tranches (the smallest and the largest debts); (3) if ordered by debt amount, the percentage of non-performing loans seems to have an inverted U shape on average, increasing up to intermediate tranches and then falling from there onwards; (4) the debtors that entered the system in 2000 and 2001 exhibit a better portfolio quality than other borrowers, thus dismissing the presumed moral hazard related to the expectation of a government rescue; (5) public banks’ non-performing loans are twice as those of national and foreign private banks in the period 1999-2005. However, it is striking that foreign banks have a higher proportion of non-performing debtors than public banks as from 2002, and both stand remarkably above the national private banks’ figure; (6) econometric results reveal that the probability of non-performing loan: (a) increases in line with the guaranteed proportion and, surprisingly, with the size of the debt; (b) increases in public banks’ debtors and falls in national private banks’ debtors; (c) increases with the assistance through Overdrafts and falls in Discount of documents and also –and this is again highly curious– in Personal loans; and (7) even though the change in probability is statistically and economically significant for all the abovementioned variables, the macroeconomic context (measured through dummy variables per year) seems to be the determinant with the highest impact on this probability.