Loan Approval Decisions Research Articles

Statistical Methods in Credit Risk Prediction: Analyzing Risk through Data Analytics

This study explores the factors influencing loan approval decisions by financial institutions using logistic regression analysis. By examining a dataset of loan applicants, we identify key predictors that significantly impact the likelihood of loan approval. The analysis reveals that higher income, larger family size, higher education levels, possession of a CD account, use of online banking, and having a credit card are positively associated with loan approval. Conversely, the presence of a securities account has a more complex relationship with loan approval likelihood. These findings suggest that financial institutions favor applicants with stable financial profiles and higher levels of financial engagement. Understanding these factors can help improve risk assessment models, ensuring more accurate and fair loan approval processes.

ENHANCING LOAN APPROVAL DECISION-MAKING: AN INTERPRETABLE MACHINE LEARNING APPROACH USING LIGHTGBM FOR DIGITAL ECONOMY DEVELOPMENT

This study aims to enhance loan approval decision-making in the digital economy using an interpretable machine learning approach. The primary research question investigates how integrating an interpretable machine learning approach can improve the accuracy and transparency of loan approval processes. We employed LightGBM, a gradient-boosting framework for loan approval classification, optimized via Random Search hyperparameter tuning and validated using 10-fold cross-validation. We incorporated the Shapley Additive exPlanations (SHAP) framework to address the challenge of interpretability in machine learning. The LightGBM model outperformed conventional algorithms (Decision Tree, Random Forest, AdaBoost, and Extra Trees) in accuracy (98.13%), precision (97.78%), recall (97.17%), and F1-score (97.48%). The study demonstrates that using an interpretable machine learning approach with LightGBM and SHAP can significantly improve the accuracy and transparency of loan approval decisions. This method offers a promising avenue for financial institutions to enhance their loan approval mechanisms, ensuring more reliable, efficient, and transparent decision-making in the digital economy. The study also underscores the importance of interpretability in deploying machine learning solutions in sectors with significant socio-economic impacts.

Learning Fair Policies for Multi-Stage Selection Problems from Observational Data

We consider the problem of learning fair policies for multi-stage selection problems from observational data. This problem arises in several high-stakes domains such as company hiring, loan approval, or bail decisions where outcomes (e.g., career success, loan repayment, recidivism) are only observed for those selected. We propose a multi-stage framework that can be augmented with various fairness constraints, such as demographic parity or equal opportunity. This problem is a highly intractable infinite chance-constrained program involving the unknown joint distribution of covariates and outcomes. Motivated by the potential impact of selection decisions on people’s lives and livelihoods, we propose to focus on interpretable linear selection rules. Leveraging tools from causal inference and sample average approximation, we obtain an asymptotically consistent solution to this selection problem by solving a mixed binary conic optimization problem, which can be solved using standard off-the-shelf solvers. We conduct extensive computational experiments on a variety of datasets adapted from the UCI repository on which we show that our proposed approaches can achieve an 11.6% improvement in precision and a 38% reduction in the measure of unfairness compared to the existing selection policy.

Predicting Loan Default: A Comparative Analysis of Multiple Machine Learning Models

Financial decision-making, particularly in loan approval, requires precise risk prediction. To enhance the prediction accuracy, this study utilizes various machine learning models, namely Logistic Regression, XGBoost, an Artificial Neural Network (ANN), and a hybrid XGBoost + Logistic Regression (XGB+LR). These models were selected based on their unique capacities to capture complex patterns and relationships within the data, thereby potentially improving the loan default prediction task. The training and validation of these models were performed on a meticulously prepared dataset, following crucial preprocessing steps such as one-hot encoding, feature selection, and scaling. To ensure the models' optimal performance, intensive hyperparameter tuning was conducted. The application of these techniques resulted in a robust set of models. Each model's performance was rigorously evaluated through established metrics, including the Area Under the ROC Curve (AUC) and Accuracy (ACC). Among these models, the XGBoost model demonstrated superior predictive power, achieving an AUC of 0.798 and an ACC of 0.861 on the validation set. A detailed feature importance analysis using the XGBoost model further revealed that Credit_Score and Loan_Amount were the primary factors impacting loan approval decisions. Despite slight overfitting observed in the models, the results confirm the potential of machine learning in improving financial decision-making processes. This study sets the foundation for future advancements, which may include the application of advanced regularization techniques, further hyperparameter optimization, and the inclusion of a broader feature set.

LOAN STATUS PREDICTION USING DECISION TREE CLASSIFIER

This paper investigates the effectiveness of the Decision Tree Classifier in predicting loan status, a critical task in the financial sector. The study utilizes a dataset containing various attributes of loan applicants such as income, credit score, employment status, and loan amount. The dataset is preprocessed to handle missing values and categorical variables. Feature importance is analyzed to understand the key factors influencing loan approval decisions. A Decision Tree Classifier model is trained and evaluated using performance metrics such as accuracy, precision, recall, and F1-score. The results demonstrate the feasibility of using Decision Trees for loan status prediction and provide insights into the decision-making process of loan approval.

Loan Approval Prediction using Machine Learning

In today's digital age, financial institutions are facing a significant challenge in managing the increasing volume of loan applications. Traditional loan approval methods, which involve manual evaluation by credit analysts, are time-consuming and prone to errors. To overcome these challenges, machine learning algorithms have emerged as a promising solution for automating the loan approval process. This abstract discusses the use of machine learning algorithms for loan approval prediction. The proposed approach involves collecting and preprocessing a large dataset containing historical loan applications, including various financial and personal attributes. The data is then fed into a machine learning model that predicts the likelihood of loan approval based on the input features. The model is trained using supervised learning techniques such as Random Forest, Logistic Regression, Support Vector Machine, XGboost, Decision Tree, Python. The selected model is then integrated into the lending institution's loan approval process, replacing or augmenting the manual evaluation process. The benefits of this approach include faster processing times, reduced errors, and more consistent decision-making. Additionally, machine learning algorithms can provide insights into the factors that influence loan approval decisions, enabling lending institutions to make more informed decisions and improve their overall loan portfolio management strategies. Machine learning algorithms have the potential to revolutionize the loan approval process by providing a more efficient and accurate alternative to traditional methods. As the volume of loan applications continues to grow, it is essential for lending institutions to adopt these technologies to remain competitive and provide better service to their customers

The effect of key audit matters reviewing on loan approval decisions?

The existing literature does not provide a definitive statement on whether the inclusion of Key Audit Matters (KAM) in accounting research increases the likelihood of loan approval by bank approvers. This study investigates the combined effects of approvers' trait skepticism, their professional background in accounting and auditing, as well as the familiarity and readability of KAM content, on loan approval probability. The findings indicate that an increased duration spent reading 'How' paragraph enhances the probability of loan approval by approvers. Moreover, this positive correlation is more pronounced for familiar and easily legible content, highly skeptical individuals, and those with expertise in accounting and auditing. These results offer novel insights and research perspectives for KAM while expanding upon the literature concerning factors influencing approvers' loan approvals.

Democracy favors access to credit of firms

Access to credit is one of the main obstacles for the growth of firms. We test the hypothesis that democracy exerts an impact on access to credit. We perform regressions at the firm-level on a large dataset of 46,000 firms in 108 countries. We find evidence of a negative relationship between democracy and credit constraints for firms. We further establish that democracy contributes to reduce borrower discouragement and leads to more bank loan approval decisions. Our key finding is therefore that democracy favors firms' access to credit. Our work contributes to the debate on the impact of democracy on economic development by considering one firm-level channel of transmission.

What’s in a Face? An Experiment on Facial Information and Loan-Approval Decision

Facial information is essential in daily life, but relatively little is known about whether seeing a face improves people’s decision quality. This experimental paper studies the loan-approval decisions based on the historical cash-loan data with real repayment outcomes and exogenously varies whether and how a borrower’s facial information is provided. We find that facial information does not improve subjects’ decisions, despite the fact that it can predict repayment behavior in a machine-learning algorithm. This is because subjects have various biases in evaluating facial photos, and they rely excessively on facial information in making the loan-approval decisions. This paper was accepted by Yan Chen, behavioral economics and decision analysis. Funding: J. Meng received financial support from the National Natural Science Foundation of China [Grants 71103003, 71471004, and 71822301]. Supplemental Material: Data files and the online appendix are available at https://doi.org/10.1287/mnsc.2022.4436 .

A New Prediction Approach for Preventing Default Customers from Applying Personal Loans Using Machine Learning

In the Egyptian banking industry, loan officers use pure judgment to make personal loan approval decisions. In this paper, we develop a new predictive method for default customers' loans using machine learning. The new predictive method uses the available personal data and historical credit data to evaluate the credit trust-worthiness of customers to obtain loans. We used the ABE dataset for training and testing, as we used 10 features from the application form and i- score report class that could give great help to credit officers for taking the right decision through avoiding customer selection using random techniques. The collected dataset was analysed by using various machine learning classifiers based on important selected features, to obtain high accuracy. We compared the performance of several machine learning classifiers before and after feature selection. We have found that in terms of high accuracy, the most important features are (activity – income – loan) and in terms of better performance the decision tree classifier has surpassed any other machine learning classifier with significant prediction accuracy of almost 94.85%.

Fair value accounting from the users’ perspective: an experiment on how financial analysts rely on fair value estimates in their decisions

Purpose The current paper aims to understand whether fair value accounting (FVA) affects analysts’ loan approval decisions and default risk judgments. Design/methodology/approach This study focusses on three issues: unrealized gain or loss resulting from FV measurement recognized in other comprehensive income (OCI), recognition of assets at FV or historical cost and the disclosure or non-disclosure of the FV of collateral assets. It uses an experiment carried out with a sample of 29 CFA analysts. Findings The results show that all three issues have a significant effect on analysts’ judgment and decision-making in processing FV estimates. Originality/value The paper extends knowledge on how financial analysts perceive FV estimates and disclosure and may help the accounting standard boards assess the challenges facing analysts when they apply professional judgments in interpreting FV measurements and disclosures. Moreover, it offers fresh views to the debate on the decision usefulness of FVA, particularly relevant in the post-implementation review of IFRS 13.

What's in a Face？An Experiment on Facial Information and Loan Approval Decision

Face-to-face communication is essential in daily life but relatively little is known about whether seeing a face improves people’s decision quality. This experimental paper studies the loan approval decisions based on the historical cash loan data with real repayment outcomes, and exogenously varies whether and how a borrower’s facial information is provided. We find that facial information does not improve subjects’ decisions, despite the fact that it can predict repayment behavior in a machine learning algorithm. This is because subjects have various biases in evaluating facial photos; and they rely excessively on facial information in making the loan approval decisions.

Sistem Pendukung Keputusan Penentuan Pemberian Pinjaman Dengan Metode Perbandingan Eksponensial (MPE) Pada Koperasi XYZ

The XYZ cooperation is a business entity dealing with savings and loans, building rental, convection, rice procurement and shops that are very concerned about the welfare of its members. The appraisal process for loan approval decision in the XYZ cooperative was still lack of assessment. It was based on the value of the members’ report cards which covers collateral, wealth, and behavior so that the funds were not often on the target. Moreover, the cooperative members were often careless in repayment process. The method in this research was Exponential Comparison Method. This method determined the priority order of alternative decisions with criteria. Those criteria in this research were: Report cards, BI Checking, Application Purpose, source of income, monthly income, installment period, completeness of files, guarantees, status, and the amount requested. As a result of applying this system and adding some new criteria, cooperation considers loan approval assessments in determining the eligible members. The results of functional testing by black-box method showed that the system ran as expected and the functions of the application ran properly.

Communication frictions in banking organizations: Evidence from credit score lending

By using mid-corporate loan-level data on loan approval decisions collected from a large European bank, we investigate whether spatially-based bank organizational frictions affect borrowers’ credit availability.

Do farm service agency borrowers' double minority labels lead to more unfavorable loan packaging terms?

PurposeThis study adds a new dimension in the study of racial and gender bias in farm lending. Most previous studies analyzed the separate effects of race and gender attributes on loan approval decisions. The analysis focuses on the stipulation of loan terms (loan amount, interest rate and maturity) among approved farm loan applications. The time period analyzed spans from 2004 until 2014 during which the government has undertaken reforms to improve delivery of loan services to its clientele of minority farmers. Thus, this study's findings could help validate the effectivity of such institutional reforms affecting Farm Service Agency (FSA) lending operations.Design/methodology/approachThis study utilizes a national direct loan origination data from the FSA of the U.S. Department of Agriculture (USDA) collected from 2004 to 2014. The analysis begins by identifying significant differences in cross-tabulations of loan terms among different racial and gender classes. Seemingly unrelated regression (SUR) regression techniques are then applied for a system of equations involving the three loan packaging components. The combined effects of the prescribed loan packaging terms are subsequently analyzed under a simulation-optimization framework.FindingsRegression results validate that indeed, relative to White American borrowers, certain minority borrowers are accommodated with lower loan amounts at higher interest rates and with shorter maturities. However, these decisions seem to be prompted by credit risk management considerations. The most compelling findings include the insignificance of all double minority labeling variables, except for the interest rate equation that even produced favorable results for Hispanic American females. Simulation-optimization results further reinforce that even when one or two unfavorable loan terms are included in the packaging, double minority borrowers end up with better profitability and liquidity positions.Practical implicationsThis study provides a different perspective in dealing with the controversial minority bias in lending by presenting evidence gathered from a government farm lending institution. The USDA-FSA has been sued in numerous occasions by minority borrowers. Since then, however, it has deliberately implemented institutional reforms to rectify previous errors. This study provides empirical evidence strengthening FSA's claim of its intention to improve its delivery of loan services, especially for its socially disadvantaged borrowers with double minority classification.Originality/valueThis study pioneers the analysis of the double minority labeling effect on farm lending decisions. Its contributions to literature are further enhanced by its goal to validate the effectiveness of FSA institutional reforms undertaken since the early 2000s in order to improve credit access of and delivery of credit services to minority farm borrowers, especially those that belong to more than one minority classification.

Loan approval prediction using KNN, decision Tree and Naïve Bayes models

In this modern world, financial institutions are playing a very crucial role. Nowadays, banks are developing their financial reserves by providing different kinds of loans to people who are in need. At the same time, there is also a massive increase in the count of individuals requesting loans. However, banks cannot provide loans for everyone as there are only limited reserves associated with each of them. So, banks must follow some stringent verification process to approve the loan, because if the one who got his/her loan approved failed to pay back his loan it may have a direct impact on the financial reserves of the bank and also onto the banking sector. So, banks started to provide loans only for a limited set of people who are capable of repaying their loans. But finding out who is eligible for the loan is a much typical and risky process. In this project, we will develop a model to predict who is eligible for a loan in order to reduce the risk associated with the decision process and to modify the typical loan approval process into a much easier one. Moreover, we will make use of previous data of loan decisions made by the company and with the help of various data mining techniques, we will develop a loan approval decision predicting model which can draw decisions for each individual based on the information provided by them. We will use a machine-learning-based KNN, Decision-tree, Naïve Bayes algorithms to train the model. This project primary goal is to develop a loan prediction model with a better accuracy rate.

Bank Competition and Access to Finance: Evidence from African Countries

Whether competition helps or hinders firms’ access to finance is a much-debated question in economic literature and policy issues. This study considers the consequences of bank competition on credit constraints using rich enterprise-level data set from the World Bank’s Enterprise Surveys. The study addresses 9632 firms from 27 African Countries. In addition to classical concentration measure, competition is measured by three non-structural measures (Boone indicator, H-statistics and Lerner index). The results show that bank competition worsens credit/financing constraints which supports the information hypothesis. However, further investigations show that bank competition has a significant positive effect on firms’ need for external financing, decision to apply for new line of credit and banks loan approval decision. Despite higher loan application approval rate by banks, large proportion of firms are discouraged to apply for bank loan. Small, medium and young firms are more likely to report higher financing constraints, more likely to need external financing but less likely to apply for bank loans and have low access to bank loan.

Judging Judicial Foreclosure

For the third time in the last several decades, policymakers are contemplating an overhaul of mortgage‐finance regulations. Despite the considerable attention paid to how ex ante regulations affect the availability of credit and the appropriateness of the mortgage products that lenders offer, however, our understanding of how the legal framework governing foreclosures—a form of ex post borrower protection—affects mortgage lending is incomplete. Leveraging data on loan applicants that are geographically proximate and subject to the same federal mortgage‐finance regulations and nearly identical state foreclosure regimes—but for the presence or absence of a judicial foreclosure requirement—this analysis enables the identification of the independent effects of judicial‐foreclosure requirements on loan approval decisions and the share of approved applicants who are offered subprime loans. I find that lenders adopt a more conservative posture in evaluating loan applications in jurisdictions where they must haul delinquent borrowers into court. All else equal, loan applications are less likely to be approved and approved borrowers are less likely to be offered subprime loans in judicial‐foreclosure states. Further, some models indicate that these results may be amplified for borrowers with lower socioeconomic status, suggesting that judicial supervision of foreclosures may have tempered one of the more flagrant practices of the subprime era: providing high‐rate mortgages with a greater likelihood of default to lower‐income and minority borrowers. These results suggest that, in contemplating changes to the regulation of mortgage lenders, policymakers should consider state foreclosure law to be among the tools in their regulatory toolkit.

Hierarchy of bank loan approval and loan performance

This paper analyzes data from a regional Italian bank to provide new evidence on the relationship between who, within a bank, approves a loan and the subsequent performance of the loan. The size of the bank and its pool of clients, who are primarily small- and medium-size firms, comprises characteristics of both relationship-based and transaction-based lending technologies. Our key finding is that the probability of loan default increases as the loan approval decision is made at higher levels of the lending-decision hierarchy. This evidence supports the primacy of relationship-lending technology relative to transaction-based lending technology.

Caste and Credit: A Woeful Tale?

This paper examines caste-based differences in farmers’ access to bank loans in rural India. We investigate whether banks practice taste-based discrimination on the basis of caste. In order to identify potential discrimination, we consider loan applications and approval decisions separately. We find significant inter-caste differences in application rates, and evidence of discrimination against Scheduled Tribe borrowers at the approval stage. To rule out the role of statistical discrimination, we simulate unobserved credit histories with various distributions. Evidence for taste-based discrimination persists despite accounting for unobservables. However, we find that this discrimination does not affect small farmers.