University of Patras
Αξιολόγηση επίδοσης του τραπεζικού μάνατζμεντ μέσω της μέτρησης της πιστοληπτικής ικανότητας των δανειοδοτούμενων μικρομεσαίων επιχειρήσεων (SMEs) με τη χρήση μοντέλων credit scoring, κατά τη διάρκεια της κρίσης
Abstract
dc:descriptionThe current crisis of the Greek economy led to a rapid increase of non-performing loans. The control and management of bad debts is a matter of vital importance for the sustainability of the Greek banking system. The asymmetric information at the time of granting of loans, led to adverse selection, resulting the exposure of financial institutions to a higher credit risk. In an effort to reduce exposure to additional risks during the crisis, the bank managements applied policies of credit rationing, which led to a credit crunch of the national economy.In parallel, the interventionist policy of Greek Government, as a result of the inability of conventional treatment of depression (inability to exercise fiscal - monetary policy), led to an increase in the percentage of borrowers who consciously choose not to service their debt obligations (moral hazard).In the present study we evaluate the effectiveness of bank management by measuring the creditworthiness of SMEs during the crisis.Using data from applications of business loans issued by one of the fourth domestic systemic banks in 2005, we study their performance in the period August 2010 - July 2012. Initially, we observe a significant increase of non-performing loans as crisis deepens. Also, we notice that the loans developed more than 90 days delay in the first year, indicated a further deterioration in the 2nd year. As a result, it seems to be an urgent focus of banks to avoid transition of loans in more than 90 days late.Subsequently, using the binomial logistic regression, we show that the factors related to the financial soundness of borrowers (real estate, private facilities and private residence), the strong collateral, the previous deposit cooperation with the borrower and the multi-annual operation of the business, note temporal negative correlation with the probability of a loan to be classified as non-performing. Conversely, the existence of adverse factors, the high "loan to turnover" ratio, the existence of preexisting debt cooperation and the age of the borrower, positively affect the possibility of a loan to been characterized as non-performing. Subsequently, we use the model multinomial logistic regression in order to observe the evolution of the influence of the independent variables in the formulation of additional non-performing loans, during the two years. We observe that the agents “own facilities”, “loan to turnover” ratio, "pre-existing deposit cooperation" and "rented house" showed reversal of the effect during the month of July 2012.Finally, we compare the credit scoring model applied by the bank with models: binomial logistic regression, decision trees, neural network "Multilayer perceptron" and neural network "Radial Basis Function". The most effective model is the neural network "Multilayer perceptron", followed by binomial logistic regression and decision trees. The model used by the bank showed the worst performance, over time. In general, the bank management options could reduce non-performing loans. Additionally, bank management should focus on the control of non-performing loans in order to avoid further increase, as well as to focus on financing viable enterprises, using modern credit scoring models.
Degree
thesis:*- Grantor dc:publisher
- University of Patras
- Year dc:date
- 2015
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Giannopoulos, Vasilios
- Γιαννόπουλος, Βασίλειος
Subjects
dc:subject × 24- Αξιολόγηση πιστοληπτικής ικανότητας
- Τραπεζικό μάνατζμεντ
- Ηθικός κίνδυνος
- Εσφαλμένη επιλογή
- Καταμερισμός πιστώσεων
- Λογαριθμική παλινδρόμηση
- Νευρωνικά δίκτυα
- Δένδρα αποφάσεων
- Μη εξυπηρετούμενα δάνεια
- Χρηματοπιστωτική και οικονομική κρίση
- Credit scoring
- Bank management
- Moral hazard
- Adverse selection
- Credit rationing
- Logistic regression
- Artificial neural networks
- Decision trees
- Non-performing loans
- Financial crisis
- Κοινωνικές Επιστήμες
- Οικονομικά και Επιχειρήσεις
- Social Sciences
- Economics and Business
Rights
dc:rights- Statement dc:rights
-
- BY
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/36143