Bank Credit Issues

Topic name

Method of use

Result

There are only three factors that have a real impact on the ability to

Economic analysis of affordability

debt repayment capacity: (1) Household size has an impact

Ojiaki &

Ogbukwa (2012)

debt of small cooperative farmers

in Yewa North Local Government Area of ​​the state

Correlation, multiple regression, Logit model

negative impact on debt repayment ability; (2) Regulation

Agricultural land use pattern has a positive impact on farmers' debt repayment ability.

Ogun, Nigeria

population; and (3) The amount of loan has a positive impact

improve household debt repayment capacity

Cardholders have low income, social status

Mathews and

Learn and explain the factors

lower tendency to use credit cards

Slocum

affect the ability to break

Logit Model

inefficient and less able to repay debt

(1969).

debt of science and technology at commercial banks.

with cardholders of high income and social status

higher society

The conclusion shows that risk factors such as cost

expenditure, debt, income, assets, economic conditions,

Agarwal et al. (2009)

Assessing the impact of individual capital information characteristics on household default probability and default outcomes

Panel data, semi-parametric

Legal environment and socio-demographic characteristics influence the likelihood of default

Topic name

Method of use

Result

Dunn & Kim (1999)

Assessing the likelihood of default on credit cards

Logit Model

The results show that socio-demographic variables such as age, marital status and number of children are strongly associated with the likelihood of default while income, education and homeownership

no impact on default

Sociological factors were found to play a role.

relatively small in debt repayment capacity. Income

is predicted to be most important in determining

debt repayment intention. Attitudinal factors are considered to be

Livingstone

& Lunt (1992)

Customer repayment forecast:

Psychological, social and economic determinants

Discriminant analysis and multiple regression analysis

important predictors of debt repayment

and non-repayment of debt. Other psychological factors, focusing on economic allocations, control, strategy,

Coping strategies and customer satisfaction

is very important and a variety of business activities

Specific realities also involve experience of

ability to pay

Ozdemir (2004)

An empirical investigation of consumer credit default risk

Logit Model

Other demographic variables have no effect on the probability of default of KHCN. Financial characteristics can explain the probability well.

default, typically variable interest rates and loan terms

Topic name

Method of use

Result

Kocenda & Vojtek (2011)

Default prediction in retail credit scoring: Evidence from Czech bank data

CART analysis and Logit Model

The results of the two models indicate that the most important financial and behavioral characteristics for the likelihood of default were found to be the amount of resources the customer possesses, the level of education, marital status, the purpose of the loan, and the number of years.

have a bank account

Arminger & Associates (1997)

Credit Risk Data Analysis: Comparing Logistic Discriminant with Classification Trees and Linkage Networks

Logit Model

The results show that the ability to repay debt is better for the following groups: Adults, people with cars, people with seniority, people with families and women.

Jacobson & Roszbach (2003)

Bank lending policies, credit scoring and risk valuation

Probit Model

Variables that have a significant impact on the likelihood of default include: Age, income, changes in annual income, and certain collateral-free credit conditions that have a significant impact on the likelihood of default.

insolvency

Peter & Peter (2006)

Risk Management Model: An Evidence-Based Assessment of Default Risk

Logit Model

The results show that the age of the household head plays an important role. Younger households tend to be adversely affected by the increasing burden of mortgage payments.

Topic name

Method of use

Result

Income and socio-demographic factors also have an influence: Low income, less education, young age, and divorce are factors that

increased likelihood of default

Dao Thi Thanh Binh (2019)

Building a credit scoring model for individual consumer loans in Vietnam

Discriminant analysis model

The classification results yield an accuracy of 89.4%. In which, the author shows that the model with the non-normalized function has a better ability than the normalized function. The two factors are the number of dependents and the account that contributes the most to the

predicting customer default

Pham Thi Thu Tra & Robert Lensink (2008)

Differences in default probability in formal, informal and semi-formal credit

Logit Model

The authors find that small households with collateral and/or guarantors rely mainly on formal and semi-formal borrowing while female contractors, large households, and borrowers without collateral or guarantors rely mainly on informal borrowing. In addition, informal lenders are at higher risk of default than formal and semi-formal lenders. Some of the terms in the loan agreement are relevant.

regarding the determination of default risk in credit

Topic name

Method of use

Result

formal lending, such as loan terms, lending interest rates and especially the role of relatives in informal lending are emphasized, whereby borrowing from relatives reduces the rate of

default rate

Le Van Triet (2010)

Perfecting the credit rating system

Asia Commercial Bank personal use

Logit Model

The descriptive results show that the average age of customers who default on their debt is 35 years old.

Dinh Thi Huyen Thanh & Kleimeier (2007)

Credit Scoring Model for Vietnam's Retail Banking Market

Logit Model

The authors concluded that the variables that have a significant impact on the probability of default are: seniority with the bank, gender, number of loans and loan term.

loan term

Kvamme & Associates (2018)

Predicting mortgage default using classification models

ANN, Random Forest

The comparison results of the two ANN and Random Forest models show that the Random Forest model has a higher ROC probability level than ANN (ROC AUC is 0.918 for neural networks and 0.926 for Random Forest classifier networks).

Topic name

Method of use

Result

Booth et al. (2014)

Automated Trading with Random Forest Performance and Seasonality

Expert, Random Forest, Logit model

The results show that the Random Forest classifiers with degenerate weights produce superior results in both profit and prediction accuracy compared to the

other techniques

Khemakhem

Credit Risk Prediction: One

The results were compared with discriminant analyses.

&

comparative study between

Artificial Neural Networks

number. The authors pointed out that, network techniques

Boujelbene

integral and neural networks

ANN, discriminant analysis

Artificial neural networks (ANNs) are more accurate about

(2015)

artificial

Prediction

Bennell & Associates (2006)

Credit Rating Models: Artificial Neural Networks and Probit Models

Artificial Neural Network (ANN)

The results for credit ratings corroborate the results of other researchers that ANNs are more efficient classifiers than Probit models.

The results indicate that the quadratic discriminant analysis

Finch & Schneider (2007)

Classification Accuracy of Artificial Neural Networks, Network Analysis and Discriminant Analysis, Logistic Regression

ANN, Logit model, discriminant analysis model

consistently performed as well as other methods while neural networks performed very similarly to linear discriminant analysis and logistic regression

Topic name

Method of use

Result

The results show the difference between the two models.

neural network model. In addition, research

Pacelli &

Azzolini (2011)

Using Artificial Neural Networks (ANN) for Credit Risk Management in Italy

Artificial neural network ANN, discriminant analysis, Logit model

The study also showed that artificial neural network models can better support traditional models. But the study also offers some advice.

Recommendations on the combined use of models

Regression analysis such as Logisitc or Probit to

support each other in forecasting

Zang (2011)

Default Risk Study of Commercial Bank Personal Loan Based on Artificial Neural Network

Artificial Neural Network ANN

The results show that the BP neural network prototypes have high prediction accuracy and good application for commercial banks in China.