Search Results (6)

Credit scoring is a cornerstone of financial risk management, enabling financial institutions to assess the likelihood of loan default. However, widely recognized contemporary credit risk metrics, like FICO (Fair Isaac Corporation) or Vantage scores, remain proprietary and inaccessible to the public. This study aims to devise an alternative credit scoring metric that mirrors the FICO score, using an extensive dataset from Lending Club. The challenge lies in the limited available insights into both the precise analytical formula and the comprehensive suite of credit-specific attributes integral to the FICO score’s calculation. Our proposed metric leverages basic information provided by potential borrowers, eliminating the need for extensive historical credit data. We aim to articulate this credit risk metric in a closed analytical form with variable complexity. To achieve this, we employ a symbolic regression method anchored in genetic programming (GP). Here, the Occam’s razor principle guides evolutionary bias toward simpler, more interpretable models. To ascertain our method’s efficacy, we juxtapose the approximation capabilities of GP-based symbolic regression with established machine learning regression models, such as Gaussian Support Vector Machines (GSVMs), Multilayer Perceptrons (MLPs), Regression Trees, and Radial Basis Function Networks (RBFNs). Our experiments indicate that GP-based symbolic regression offers accuracy comparable to these benchmark methodologies. Moreover, the resultant analytical model offers invaluable insights into credit risk evaluation mechanisms, enabling stakeholders to make informed credit risk assessments. This study contributes to the growing demand for transparent machine learning models by demonstrating the value of interpretable, data-driven credit scoring models. Full article

Peer-to-peer lending, a novel element of Internet finance that links lenders and borrowers via online platforms, has generated large profits for investors. However, borrowers’ missed payments have negatively impacted the industry’s sustainable growth. It is imperative to create a system that can correctly predict loan defaults to lessen the damage brought on by defaulters. The goal of this study is to fill the gap in the literature by exploring the feasibility of developing prediction models for P2P loan defaults without relying heavily on personal data while also focusing on identifying key variables influencing borrowers’ repayment capacity through systematic feature selection and exploratory data analysis. Given this, this study aims to create a computational model that aids lenders in determining the approval or rejection of a loan application, relying on the financial data provided by applicants. The selected dataset, sourced from an open database, contains 8578 transaction records and includes 14 attributes related to financial information, with no personal data included. A loan dataset is first subjected to an in-depth exploratory data analysis to find behaviors connected to loan defaults. Subsequently, diverse and noteworthy machine learning classification algorithms, including Random Forest, Support Vector Machine, Decision Tree, Logistic Regression, Naïve Bayes, and XGBoost, were employed to build models capable of discerning borrowers who repay their loans from those who do not. Our findings indicate that borrowers who fail to comply with their lenders’ credit policies, pay elevated interest rates, and possess low FICO ratings are at a higher likelihood of defaulting. Furthermore, elevated risk is observed among clients who obtain loans for small businesses. All classification models, including XGBoost and Random Forest, successfully developed and performed satisfactorily and achieved an accuracy of over 80%. When the decision threshold is set to 0.4, the best performance for predicting loan defaulters is achieved using logistic regression, which accurately identifies 83% of the defaulted loans, with a recall of 83%, precision of 21% and f1 score of 33%. Full article

At the level of the 50 U.S. states, an interconnected nexus of well-being variables exists. These variables strongly correlate with estimates of state IQ in interesting ways. However, the state IQ estimates are now more than 16 years old, and the state well-being estimates are over 12 years old. Updated state IQ and well-being estimates are therefore needed. Thus, I first created new state IQ estimates by analyzing scores from both the Program for the International Assessment of Adult Competency (for adults), and the National Assessment of Educational Progress (for fourth and eighth grade children) exams. I also created new global well-being scores by analyzing state variables from the following four well-being subdomains: crime, income, health, and education. When validating the nexus, several interesting correlations existed among the variables. For example, state IQ most strongly predicted FICO credit scores, alcohol consumption (directly), income inequality, and state temperature. Interestingly, state IQ derived here also correlated 0.58 with state IQ estimates from over 100 years ago. Global well-being likewise correlated with many old and new variables in the nexus, including a correlation of 0.80 with IQ. In sum, at the level of the U.S. state, a nexus of important, strongly correlated variables exists. These variables comprise well-being, and state IQ is a central node in this network. Full article

Machine learning and artificial intelligence have achieved a human-level performance in many application domains, including image classification, speech recognition and machine translation. However, in the financial domain expert-based credit risk models have still been dominating. Establishing meaningful benchmark and comparisons on machine-learning approaches and human expert-based models is a prerequisite in further introducing novel methods. Therefore, our main goal in this study is to establish a new benchmark using real consumer data and to provide machine-learning approaches that can serve as a baseline on this benchmark. We performed an extensive comparison between the machine-learning approaches and a human expert-based model—FICO credit scoring system—by using a Survey of Consumer Finances (SCF) data. As the SCF data is non-synthetic and consists of a large number of real variables, we applied two variable-selection methods: the first method used hypothesis tests, correlation and random forest-based feature importance measures and the second method was only a random forest-based new approach (NAP), to select the best representative features for effective modelling and to compare them. We then built regression models based on various machine-learning algorithms ranging from logistic regression and support vector machines to an ensemble of gradient boosted trees and deep neural networks. Our results demonstrated that if lending institutions in the 2001s had used their own credit scoring model constructed by machine-learning methods explored in this study, their expected credit losses would have been lower, and they would be more sustainable. In addition, the deep neural networks and XGBoost algorithms trained on the subset selected by NAP achieve the highest area under the curve (AUC) and accuracy, respectively. Full article

We study the determinants of borrowers’ default in P2P lending with a new data set consisting of 70,673 loan observations from the Lending Club. Previous research identified a number of default determining variables but did not distinguish between different loan risk levels. We define four loan risk classes and test the significance of the default determining variables within each loan risk class. Our findings suggest that the significance of most variables depends on the loan risk class. Only a few variables are consistently significant across all risk classes. The debt-to-income ratio, inquiries in the past six months and a loan intended for a small business are positively correlated with the default rate. Annual income and credit card as loan purpose are negatively correlated. Full article

A mortgage borrower has several options once a foreclosure proceedings is initiated, mainly default and prepayment. Using a sample of FHA mortgage loans, we develop a dependent competing risks framework to examine the determinants of time to default and time to prepayment once the foreclosure proceedings is initiated. More importantly, we examine the interdependence between default and prepayment, through both the correlation of the unobserved heterogeneity terms and the preventive behavior of the individual mortgage borrowers. We find that time to default and time to prepayment are affected by several factors, such as the Loan-To-Value ratio (LTV), FICO score and unemployment rate. In addition, we find strong evidence that supports the existence of interdependence between the default and prepayment hazards through both the correlation of the unobserved heterogeneity terms and the preventive behavior of individual mortgage borrowers. We show that neglecting the interdependence through the preventive behavior of the individual mortgage borrowers can lead to biased estimates and misleading inference. Full article