{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,9]],"date-time":"2026-10-09T09:55:17Z","timestamp":1791539717661,"version":"4.3.4"},"posted":{"date-parts":[[2026]]},"group-title":"SSRN","reference-count":0,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>This research proposes an explainable artificial intelligence system for assessing loan creditworthiness of borrowers, forecasting the solvency of institutions, conducting macroeconomic stress testing, and automatic risk reporting. LendingClub loan data have been combined with macroeconomic variables from FRED, whereas FDIC financial data have been used for institution-level analysis. Dif erent algorithms such as logistic regression, random forest, LightGBM, XGBoost, and deep neural network were assessed using chronological training, validation, and testing samples. XGBoost demonstrated the best results in terms of performance on a borrower level with 0.7197 and 0.7139 ROC-AUC scores on the validation and testing samples, correspondingly. The deep neural network generated similar discriminatory power but had lower specificity and more false positives. Multi-task LSTM was implemented to forecast the equity-to-assets ratio four quarters ahead and performed better than the benchmark model.The techniques SHAP and LIME recognized interest rate, FICO score, level of indebtedness, loan term, homeownership, and macroeconomic environment as key determinants that af ect credit risk prediction. Stress testing of the macroeconomic environment determined which borrowers are sensitive to negative scenarios, and the agentic reporting layer created structured reports for borrowers, explanations, stress test results, and human review flags. In general, the results show that the combination of predictive modelling, explainability, stress testing, and human review is possible.<\/jats:p>","DOI":"10.2139\/ssrn.7580278","type":"posted-content","created":{"date-parts":[[2026,10,9]],"date-time":"2026-10-09T09:47:31Z","timestamp":1791539251000},"source":"Crossref","is-referenced-by-count":0,"title":["Stress Testing and Explainable AI Frameworks for Deep Learning and Agentic Models in Credit Underwriting and Solvency Forecasting"],"prefix":"10.2139","author":[{"given":"Tahani","family":"Jabir","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0006-4571-1190","authenticated-orcid":true,"given":"Redhwan","family":"Mohammed","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Amin","family":"Ilyas","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","container-title":[],"original-title":[],"deposited":{"date-parts":[[2026,10,9]],"date-time":"2026-10-09T09:47:31Z","timestamp":1791539251000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ssrn.com\/abstract=7580278"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":0,"URL":"https:\/\/doi.org\/10.2139\/ssrn.7580278","relation":{},"subject":[],"published":{"date-parts":[[2026]]},"subtype":"preprint"}}