{"paper_id":"2609.27654","title":"FedIncome: Federated Learning for Income Estimation in Digital Lending Under Data Sovereignty Constraints","authors":["Sultan Amed","Tanmay Sen","Sayantan Banerjee"],"paper_date":"2026-09-23","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2609.27654","page_url":"https://thequant.space/flowcharts/fedincome-federated-learning-for-income-estimation-in-digital-lending-under/","scores":{"empirical_rigor":8,"math_complexity":6,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes FedIncome, a federated learning framework for income estimation that enables institutions to train a shared model without pooling raw borrower records.","quote":"We introduce FedIncome, a federated learning framework for income estimation that enables institutions to train a shared model without pooling raw borrower records.","summary":"Verified income is often unavailable in digital loan applications, forcing lenders to rely on reported income and potentially leading to over-lending, overly conservative offers, or rejection of creditworthy applicants. Cross-institutional data-sharing constraints make this problem especially…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Observed income labels are treated as the true target without measurement error.","type":"data","stated":"explicit","quote":"Assumption 3.1 (Verified training labels). For the training population, the observed income label yi is treated as the target income without additional measurement error."},{"assumption":"All participating clients share the same set of predictors with common definitions.","type":"data","stated":"explicit","quote":"Assumption 3.2 (Common feature space). All participating clients observe the same d predictors with common definitions and measurement conventions."},{"assumption":"Clients follow the training protocol honestly and do not manipulate updates.","type":"execution","stated":"explicit","quote":"Assumption 3.3 (Non-strategic participation). Participating clients transmit the updates prescribed by the training protocol and do not strategically manipulate them."}]},"datasets":{"status":"available","items":[{"name":"LendingClub loans","provider":"LendingClub","kind":"other","access":"unknown","period":null,"universe":"50 state-level client partitions","quote":"Using more than one million LendingClub loans partitioned into 50 state-level clients, we simulate a heterogeneous lending consortium."}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"unclear"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R3","candidates":{"front_matter":"n/a","evidence_card":"n/a","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"forecast-accuracy","findings":[{"finding":"FedNova achieves out-of-time R2 = 0.608, compared with 0.619 for the pooled centralized benchmark (≈1.8% relative gap).","kind":"performance","quote":"FedNova attains OOT R2 = 0.608, compared with 0.619 for the pooled centralized benchmark, corresponding to a relative R2 gap of approximately 1.8%."},{"finding":"Low‑sample clients (mean n ≈ 936) gain an average out-of‑time R2 improvement of 3.8 percentage points under FedNova.","kind":"predictive-accuracy","quote":"Low‑sample clients, with mean training size n = 936, obtain average R2 improvements of"},{"finding":"Replacing reported income with FedNova predictions raises simulated approval rates by 4.6 pp (verified) and 7.8 pp (not‑verified) while default rates increase only 0.29 pp and 0.31 pp respectively.","kind":"performance","quote":"Replacing reported income with FedNova‑predicted income increased simulated approval rates by 4.6 percentage points in the source‑verified group and 7.8 percentage points in the not‑verified group, while observed default rates among approved loans increased by 0.29 and 0.31 percentage points, respectively."},{"finding":"The empirical crossover where centralized training overtakes federated gains occurs at roughly 4,800 training observations.","kind":"statistical","quote":"the fitted client‑level relationship places the empirical crossover at approximately 4,800 training observations in the present experiment."},{"finding":"FedNova reduces out‑of‑time MAPE by 7.5 pp for low‑income and 4.7 pp for middle‑income borrowers relative to the centralized benchmark.","kind":"predictive-accuracy","quote":"FedNova’s MAPE advantage is concentrated in the low- and middle-income terciles, where MAPE is lower by 7.5 and 4.7 percentage points, respectively."}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":39,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":1,"extracted_at":"2026-11-14T17:04:35Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}