| The issue of "gender discrimination" in credit granting has been extensively studied.In order to effectively manage credit risk,loan approvals are beginning to use machine learning algorithms with higher prediction accuracy.However,few scholars have studied whether the application of machine learning algorithms will intensify or reduce "gender discrimination" in credit.By studying the problem of "credit discrimination",this paper finds that there is "gender discrimination" against female borrowers in both traditional bank credit market and P2 P lending market,as female borrowers are associated with higher loan rejection rates or loan interest rates and lower loan sizes.On this basis,this paper uses the credit data sets of PPDAI and Home Credit to explore whether machine learning algorithms can help reduce the problem of "gender discrimination" in credit.First,we use the Logit method to analyze whether there is a difference in actual default performance between female borrowers and male borrowers.Secondly,the four machine learning algorithms of SVM,RF,XGBoost and Light GBM are applied to loan approval for personal credit risk prediction,then the five-fold cross-validation method and grid search method are used to select hyperparameters to train the machine learning model.The loan pass rate of each borrower under the four models is predicted respectively.Finally,the effect of borrower gender on the predicted pass rate was analyzed using the OLS method.The study finds that female borrowers have a lower default risk,and machine learning algorithms can accurately identify the creditworthiness differences of borrowers of different genders,giving female borrowers with a lower repayment default rate a higher pass rate,and at the same time not compromising Loan rates and amounts for women create additional limits.This paper argues that the introduction of machine learning algorithms into credit approval,the application of machine learning algorithms can help reduce the problem of "gender discrimination" in credit.Finally,this paper puts forward the safeguard measures to reduce credit gender discrimination from three aspects: the supervision department,the financial department and the science and technology department. |
